Inﬂuence of consumer reviews on online purchasing decisions in older andyounger adults - Tài liệu tham khảo | Đại học Hoa Sen

Inﬂuence of consumer reviews on online purchasing decisions in older andyounger adults - Tài liệu tham khảo | Đại học Hoa Sen và thông tin bổ ích giúp sinh viên tham khảo, ôn luyện và phục vụ nhu cầu học tập của mình cụ thể là có định hướng, ôn tập, nắm vững kiến thức môn học và làm bài tốt trong những bài kiểm tra, bài tiểu luận, bài tập kết thúc học phần, từ đó học tập tốt và có kết quả

20 10 lượt tải Tải xuống

Chia sẻ Báo cáo tài liệu

Contents lists available at ScienceDirect

Decision Support Systems

journal homepage: www.elsevier.com/locate/dss

Inﬂuence of consumer reviews on online purchasing decisions in older and

younger adults

Bettina von Helversen

a *,

, Katarzyna Abramczuk

, Wiesław Kopeć

, Radoslaw Nielek

Department of Psychology, University of Zurich, Switzerland

Institute of Sociology, University of Warsaw, Poland

Polish Japanese Academy of Information Technology, Poland

A R T I C L E I N F O

Keywords:

Consumer decision making

Older adults

Consumer ratings

Consumer reviews

Anecdotal evidence

A B S T R A C T

We investigated how product attributes, average consumer ratings, and single aﬀect-rich positive or negative

consumer reviews inﬂuenced hypothetical online purchasing decisions of younger and older adults. In line with

previous research, we found that younger adults used all three types of information: they clearly preferred

products with better attributes and with higher average consumer ratings. If making a choice was di cultﬃ

because it involved trade-oﬀs between product attributes, most younger adults chose the higher-rated product.

The preference for the higher-rated product, however, could be overridden by a single aﬀect-rich negative or

positive review. Older adults were strongly inﬂuenced by a single aﬀect-rich negative review and also took into

consideration product attributes; however, they did not take into account average consumer ratings or single

aﬀect-rich positive reviews. These results suggest that older adults do not consider aggregated consumer in-

formation and positive reviews focusing on positive experiences with the product, but are easily swayed by

reviews reporting negative experiences.

1. Introduction

Understanding how people make online purchasing decisions is of

growing importance. With an increase of 19.9% in 2016 and a fore-

casted growth of 17.5% for 2017, global business to consumer (B2C) e-

commerce is now accounting for 8.7% of retail sales worldwide.

Overall, e-commerce is still dominated by younger and middle-aged

consumers, but older consumers (55-year-old and older) are increas-

ingly buying goods or services online [1]. So far most research has

focused on younger adults, leaving it unclear how older adults deal with

the challenges involved in online consumer decisions (for notable ex-

ceptions see [ ]).1 3–

The goal of the present research is to contribute to understanding

how older adults make on-line purchasing decisions. Do they diﬀer in

their decision process from younger adults? What information do they

consider? And last but not least: how can we use this knowledge to

ensure better decision making on their part? We focus on how older

adults use three main types of information: product attributes, average

consumer ratings, and single positive and negative reviews that contain

an aﬀect-rich and vivid description of the reviewers' experiences. We

also take into account how the products are presented i.e. whether they

are presented simultaneously or sequentially and which product is

presented as the ﬁrst/on the left.

In the following, we ﬁrst review the literature on the inﬂuence of

consumer ratings and reviews on online purchasing decisions and on

how decision making processes change with age. Then, we report two

experimental studies investigating how younger and older adults use

consumer reviews in hypothetical online purchasing decisions. Finally,

we discuss the results of the studies and consequences of our ndingsﬁ

for designing e-commerce systems.

2. Related work

2.1. Inﬂuence of consumer reviews on attitudes and purchasing intentions

The eﬀect of consumer reviews on online decisions is widely re-

cognized. Numerous studies have shown that consumer ratings and

reviews impact people's purchasing behavior and intentions, as well as

attitudes towards products and retailers (e.g., [ ]).4–6

According to recent meta-analyses, the most important features in-

ﬂuencing sales and attitudes are the valence and the volume of reviews

[5 7, ]. In general, more positive reviews increase sales and attitudes,

https://doi.org/10.1016/j.dss.2018.05.006

Received 13 November 2017; Received in revised form 15 March 2018; Accepted 28 May 2018

Corresponding author at: University of Zurich, Binzmühlestr. 14, Box 19, 8050 Zürich, Switzerland.

E-mail address: b.vonhelversen@psychologie.uzh.ch (B. von Helversen).

https://www.ecommercewiki.org/Prot:Global_B2C_Ecommerce_Report_2016.

Decision Support Systems 113 (2018) 1–10

Available online 18 June 2018

(http://creativecommons.org/licenses/BY-NC-ND/4.0/).

whereas negative reviews reduce them (e.g., [5,8]). Their eﬀect, how-

ever, also depends on review exposure [9], the characteristics of the

reviewer [10], and the source of the review [ ].5

Although positive and negative reviews can sway consumers' be-

havior, some research has indicated that they diﬀer in their impact.

Purnawirawan et al. [7] reported that negative reviews had the stron-

gest eﬀect on attitudes and usefulness, suggesting that negative reviews

may carry more weight than positive reviews [11,12] — a ﬁnding that

resonates with research in further areas of communication [ ].13,14

However, other research has reported that with consumer reviews the

negativity bias is limited to hedonic goods [12]. Furthermore, Wu [ ]15

suggested that consumers may not weigh negative reviews more

strongly per se, but perceive them as more informative because they

often are rarer and of higher quality.

Besides the valence of the review, the format of the information

matters. Online platforms often provide consumer reviews in two for-

mats: average ratings giving an overview over the overall perceived

quality of the product (i.e., statistical information) and single reviews

that contain personal narratives of experiences made with a speci cﬁ

product. The relative importance of these types of information is still

under debate. A recent consumer survey indicated that customers rate

average ratings as most important [16]. Hong and Park [17] found that

both statistical information and narrative information are equally

convincing, whereas Ziegele and Weber [18] reported that although

average ratings were considered important, single vivid narratives

overrode average ratings. This picture is consistent with research in the

medical domain showing that anecdotal or narrative evidence can be

more convincing than statistical evidence of treatment quality [ ].19-21

The question of how strongly single reviews inﬂuence behavior is

particularly important because people often only read a small number

of reviews before making a decision, focusing on the most recent re-

views [ ].16

In sum, research suggests that younger adults' purchasing decisions

are strongly inﬂuenced by average consumer ratings. Average ratings of

a product, however, may loose their inﬂuence on decisions if they are

inconsistent with a well-written, single review [18]. Furthermore, some

research indicates that negative reviews exert stronger inﬂuence than

positive ones [7] suggesting that negative single reviews may carry

more weight than positive single reviews. In contrast, little is known

about how older adults make online consumer decisions and react to

consumer ratings and reviews.

2.2. Aging, decision making, and online purchasing

Aging is characterized by a number of changes in cognitive abilities,

aﬀect and motivation [23-25] that impact how older adults make de-

cisions (e.g., [ ]).25,26

In terms of cognitive abilities, growing old is related to a decrease in

ﬂuid cognitive abilities such as working memory capacity, processing

speed and visual processing, resulting in older adults having di cultiesﬃ

in a number of cognitive tasks (e.g., [27 29– ]). This age-related decline

also aﬀects the decision making process. Older adults tend to perform

worse than younger adults, in particular, if tasks are complex, demand

the processing of large amounts of information (e.g., [30,31]), or re-

quire learning [ ].32,33

Despite the decline of ﬂuid abilities, older adults show an increase

in crystallized abilities; that is, higher levels of declarative knowledge

and experience [23]. Using this knowledge and experience, older adults

can devise strategies to compensate for their limited ﬂuid cognitive

abilities (e.g., [26]). Speciﬁcally, they are more selective in their in-

formation search and frequently rely on less information-intensive

strategies [26 34, ]. Moreover, older adults may simplify decision pro-

blems by focusing more on aﬀective cues [25]. Although these simpler

strategies often perform somewhat worse than more information-in-

tensive strategies, they perform very well if they are suited to the task

(e.g., [35]). Accordingly, the loss in decision quality can be quite small

[36 37, ].

In line with this, research in consumer contexts indicates that older

adults have more diﬃculties when options diﬀer on many attributes

(e.g., [38]). Furthermore, older adults tend to search for less informa-

tion than younger adults while making consumer decisions [39] and

prefer to stick to the same brand [ ].40,41

Relatively little research has considered how older adults navigate

the online world, but the number of studies is rising with more elderly

adults using the Internet [1,2,42]. Still, older adults seem to be more

reluctant than younger adults to use e-commerce and are less familiar

with computer technology in general [1,43]. In addition, a study in

Hong Kong found that older adults perceived online purchases as less

easy than middle-aged adults [2]. Most relevant, Ma et al. [3] found

that age was negatively related to self-reported perceived beneﬁts of

consumer reviews, their persuasiveness, and use.

2.3. Aging and processing of aﬀect-rich consumer reviews

Although overall text comprehension suﬀers in old age [44], older

adults' ability to process narrative and emotional texts is well preserved

[45 46, ]. Accordingly, single consumer reviews presented in a narrative

format may present a source of information that is easily accessible for

older adults and thus exert a strong inﬂuence on their decisions, even if

the information is not representative of overall consumer opinions. Yet,

whether older adults are equally inﬂuenced by negative and positive

aﬀect-rich reviews is unclear.

Besides changes in cognitive processing, aging is also related to

changes in aﬀect and motivation, which may inﬂuence the information

older adults pay attention to. Socio-emotional selectivity theory pro-

poses to that with increasing age people focus more on maintaining

positive aﬀect and less on increasing their knowledge [24,47]. In line

with this idea, older adults have been shown to report improved psy-

chological well-being and lower levels of negative aﬀect [48]. More-

over, older adults often show a positivity eﬀect; that is, they exhibit a

preference for positive over negative information in processing in-

formation [49]. Speciﬁcally, older adults pay more attention to positive

information and remember it better than negative information [49, ].50

At face value the positivity eﬀect would suggest that older adults

will pay more attention to and consequently are more inﬂuenced by

positive reviews. However, a focus on maintaining positive aﬀect may

not always go hand in hand with a focus on positive information. In this

vein, Depping and Freund [51] proposed that to maintain positive af-

fect older adults focus on preventing losses, resulting in a higher sen-

sitivity and more attention to losses. In line with this idea, it has been

shown that in learning paradigms older adults learn better from nega-

tive than from positive consequences [52 53, ] — a bias that is not shown

by younger adults [54]. A focus on preventing losses, however, suggests

that older adults should be inﬂuenced more strongly by negative re-

views.

2.4. Presentation of options

In laboratory decision tasks, options are usually presented si-

multaneously, side by side. However, when purchasing products online

consumers often need to consider options sequentially. Although in

principle the decision task is the same, simultaneous or sequential

presentations can aﬀect the decision process. Presenting options se-

quentially can result in order eﬀects, leading often to a preference for

the ﬁrst option (e.g., [55]). Furthermore, people seem to be more sa-

tisﬁed with choices from simultaneous presented options (e.g., [ ]).56

Last but not least, decision processes may change depending on the

presentation with simultaneous presentation facilitating attribute-wise

comparisons, whereas presenting a single option may lead to more al-

ternative-wise comparisons (see, [57]). Although the inﬂuence of the

presentation type on choices is not the focus of our research, we ma-

nipulated whether products were presented sequentially or

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

simultaneously to ensure that eﬀects of average ratings and single

narrative reviews are not limited to one type of presentation.

3. Predictions and research questions

We investigated three problems. First, we wanted to know whether

older and younger adults rely on average consumer ratings and how

this depends on products' characteristics and their presentation.

Second, we examined whether single, vivid, and aﬀect-rich positive and

negative reviews can override their preferences for products with

higher average consumer ratings. Third, we inquired how these two

groups perceive the aﬀect-rich reviews. To analyze these three pro-

blems we conducted two empirical studies, one with young adults

(Study 1) and one with older adults (Study 2).

In Study 1 we expect to replicate the main ﬁndings from the lit-

erature. For one, we expect that young adults will in general prefer

options that have higher average ratings to options with lower average

ratings (e.g., [5]). Secondly, following Ziegele and Weber [18] we ex-

pect that preferences for options with higher average ratings will be

reduced when a single aﬀect-rich review favors the option with the

lower rating. In addition, we aim to examine whether a single negative

consumer review will have a stronger e ect than a single positiveﬀ

consumer review, following up on research suggesting a bias for ne-

gative information.

In Study 2 we expect that older adults will prefer options with better

attributes but that their choices will be more noisy due to the decrease

in decision making capacities in older adults [33]. Secondly, we aim to

test whether older adults will also prefer options with better average

ratings. On the one hand, Ma et al. [ ] report that older adults do not3

trust consumer ratings, indicating that they may not pay attention to

this information. On the other hand, if older adults recognize the value

of average consumer ratings, they might focus even more strongly on

this information than younger adults as older adults tend to consider

less information than younger adults [3, ].26

Thirdly, we investigate the relative inﬂuence of positive and nega-

tive, aﬀect-rich reviews on older adults' choices. For a stronger eﬀect of

positive aﬀect-rich reviews speaks the fact that older adults have been

shown to pay more attention to positive information [49]. On the other

hand, Depping and Freund [51] argued that older adults are motivated

to prevent losses. A focus on preventing losses, in turn, should result in

older adults being more strongly inﬂuenced by negative consumer re-

views.

Lastly, we examine how older adults perceive the aﬀect-rich con-

sumer reviews in comparison to the baseline reviews we used. Although

comprehension of emotional texts is fairly well preserved in older

adults [45 46, ], in general text comprehension is lower in older

adults [44] and older adults have less experience with online shopping.

Thus, it is possible that older adults will report problems in under-

standing reviews or perceive less of a diﬀerence in valence between

aﬀect-rich consumer reviews and baseline reviews.

4. Methods

During the studies participants were presented with pairs of house-

hold products (for example two vacuum cleaners) and had to indicate

for each pair which of the two options they would prefer to buy.

Products were presented on cards and described by four relevant at-

tributes (e.g., prize, power). In addition to the products' attributes, an

average consumer rating was shown for each product. All average

ratings were positive but one product was always rated somewhat

better than the other product.

We tested three between-participants conditions that varied whe-

ther a single written review was shown in addition to the average

consumer rating and the aﬀective content of this review: In the no“

single review condition”, participants only received information about

average consumer ratings. This condition allowed us to test whether

Fig. 1. Exemplary products card showing a choice between vacuum cleaners in the positive review condition. The aﬀect-rich positive review is presented for the

lower rated product (left) and the short baseline review for the higher rated product (right).

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

participants relied on average consumer rating in their choices. In the

“positive single review condition” the lower rated product was pre-

sented together with a highly positive, vivid, and aﬀect-rich review

while the higher rated product was presented with a somewhat positive

but short baseline review. In the “negative single review condition the”

higher rated product was presented with a highly negative, vivid, and

aﬀect-rich review while the lower rated product was presented with the

baseline review. Thus, in both conditions the single review was in-

consistent with the average consumer rating allowing us to test whether

it inﬂuences how frequently the higher rated product is chosen. Fig. 1

illustrates a choice in the positive single review condition.

In addition, we varied presentation-type (simultaneous vs. sequen-

tial presentation of the options) between participants resulting in a 3

(single review condition) by 2 (presentation type) design.

Studies were conducted by the Polish Japanese Academy of

Information Technology (PJAIT). They were approved by the Ethics

Committee of the Department of Psychology at the University of Basel.

The study with older adults was conducted on the premises of PJAIT

supervised by the research team. In the case of younger participants

(i.e., students of PJAIT), the study was run as an unsupervised online

survey.

4.1. Participants

Study 1 involved 154 younger adults who were students at PJAIT.

Their average age was 20.8 years (SD = 2.3) and 140 of them were

male. Study 2 involved 165 older adults who were recruited via a

LivingLab project run by PJAIT and focused on older adults [ ].58 60–

Older adults' average age was 69 years (SD = 6.8, range: 58–87 years)

and most of them were female (109 participants). Similar to the student

group, the vast majority of older adults (157 participants) had at least

secondary education.

As a compensation for taking part in the study, older participants

received a pen drive (a USB ﬂash drive) with additional materials re-

lated to the LivingLab and younger participants (students) received

extra credit points. On average, it took younger adults 8 min and older

adults 19 min to complete the study. Participants were randomly as-

signed to one of the six conditions.

4.2. Materials

4.2.1. Product cards

Participants made choices for three types of products: Vacuum

cleaners, irons, and drills. Product types were selected to ensure that

most participants would have some but not too much knowledge about

them. Each product card contained a product photo and its four attri-

butes including price.

For vacuum cleaners and irons, the attributes' values were chosen so

that it was unclear which of the two products was the better choice

because each of them was superior in at least one attribute. For drills,

one drill in the pair clearly dominated the other option because it had

better values on three attributes (it was faster, cheaper, and worked

longer on a battery) and similar values for the fourth attribute (it was

slightly heavier).

All product descriptions can be found in the online supplementary

material and on the Open Science Framework (OSF, folder materials).

4.2.2. Average consumer ratings

For each product the average consumer rating was presented as a

number of ﬁlled-in stars from a total of 5 stars, similar to the way

consumer ratings are presented on websites of online retailers, see

Fig. 1. All average ratings were positive (e.g between 3.9 and 4.7 stars)

but in each product pair one of the products was rated between 0.5 and

0.6 points higher than the other product, reﬂecting typical rating dif-

ferences found on online retail websites. Which of the two products in a

pair was presented with the better rating and which product was pre-

sented ﬁrst/on the left side of the screen was counterbalanced across

participants (within each of the six conditions) to separate the in uenceﬂ

of average ratings from the inﬂuence of product attributes on choices.

In addition to the average rating we showed the distributions of the

ratings below the average rating (see Fig. 1). The number of ratings was

kept similar across all products (around 150).

4.2.3. Single consumer reviews

Depending on the single review condition, participants received a

single narrative consumer review in addition to the product information

and the consumer ratings. The reviews were adapted from reviews of

similar products taken from a website of a large online retailer. They

were presented to subjects as randomly selected consumer reviews to

emphasize that any of the reviews for the product could have been

selected. In each pair one product received an aﬀect-rich review,

whereas the other product was presented with a baseline review. The

baseline review was a short (typically one sentence) comment that was

in general positive but lacked detail, vividness, and emotional content

such as “Not too heavy, steams well, and delivered on time. Good price

to value ratio.” (for an iron).

The aﬀect-rich single review was selected

to be of high emotional intensity and of extreme valence (i.e. highly

positive in the positive single review condition and highly negative in

the negative single review condition). They contained vivid and de-

tailed descriptions of positive/negative experiences the consumer had

made with the product to facilitate putting one self in the position of the

person writing the review (see Fig. 1). Aﬀect-rich positive and negative

reviews were selected to be of similar length, aﬀective intensity, and

detail. Neither the single aﬀect-rich reviews nor the baseline reviews

contained an explicit star-rating.

In the single positive review condition, the aﬀect-rich review was

presented with the lower rated product and the baseline review with

the higher rated product. In the negative review condition, the a ect-ﬀ

rich review was presented with the higher rated product and the

baseline review with the lower rated product.

4.2.4. Presentation type

The two products were presented simultaneously or sequentially. In

the ﬁrst case, the two product cards were shown on the same screen,

one next to the other. In the second case, they were presented on se-

parate screens. After seeing the ﬁrst option, participants had to click to

move on to the next screen to see the second option. Participants were

not allowed to go back. In both, the simultaneous and the sequential

condition, the decision itself was made on a yet separate screen that was

presented after the product cards.

4.2.5. Ratings of product attributes, consumer ratings and reviews

In addition to participants' choices we measured how they perceived

the presented information. Each choice was followed by a short survey

asking the subjects to rate the importance of the product attributes, the

average consumer rating, and the single consumer review (if applic-

able) for the decision that they had just made. In addition, they rated

the diﬃculty of the decision and their knowledge about the product

type (i.e., vacuum cleaners, irons, and drills). All ratings were made on

7-point Likert scales ranging from (1) not at all to (7) very much.

https://osf.io/3n8xw/.

We chose these statements as a comparison for the vivid emotional reviews

over completely neutral statements because they better reﬂect typical short

reviews that are found on online retailer websites and which are in general

positive [4,61,62]. Thus they provide a realistic baseline to which reviews

could be compared.

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

4.3. Procedure

After signing a consent form, participants were asked to provide

basic demographic characteristics (gender, education and age) and to

rate their experience with online shopping on a scale from (1) not at all

experienced to (7) very much experienced. Afterwards, they were in-

formed about the study and the consumer decisions they would make.

Before each decision participants received information regarding pro-

duct attributes and why they may be important while choosing between

the products. Then the two products were presented to the participants

and they had to indicate their choice. After the choice was made, par-

ticipants responded to the survey about the decision process and then

continued with the next decision. At the end of the study, participants

read all the consumer reviews used in the study (just the texts) and

rated their valence and understandability on a scale from (1) very ne-

gative/do not understand at all to (7) very positive/understand very

much respectively. The latter questionnaire was added only later for the

older adults, thus information from 41 people is missing. After the

study, participants received their reimbursement.

5. Results

In the following, we ﬁrst analyze whether product attributes, their

presentation and average consumer ratings inﬂuenced the choices.

After that we examine how single aﬀect-rich positive and negative

consumer reviews changed the frequency with which the product with

higher average rating was chosen and how participants perceived the

single reviews. Further (exploratory) analyses investigating partici-

pants' ratings are reported in the supplementary online material and on

the OSF (folder Results).

To facilitate the comparison between the age

groups, we report the results from Study 1 and Study 2 side by side in

each section.

5.1. Inﬂuence of average ratings and product attributes on choices

Our ﬁrst research questions focused on whether younger and older

adults used average consumer ratings in their decisions and if older

adults were able to reliably choose products with better attributes.

In order to test to what degree participants considered product at-

tributes, we constructed an index of product quality that indicated how

much better the attributes of one product were in comparison to the

other product in the product pair. To this end, we ﬁrst calculated for

each product attribute the percentage by which the product with higher

average rating was superior/worse than the product with lower rating

and then averaged across product attributes. A low absolute value in-

dicates that the two products are of similar quality and that making a

choice required a trade-oﬀ between the products' attributes. In contrast,

a high absolute value indicates that one product is clearly superior to

the other product and no trade-oﬀs are necessary. Although this index

can not account for subjective diﬀerences in the importance of the at-

tributes, it provides a useful index of how clearly one product in the

pair was better than the other product. As designed, for vacuum clea-

ners and irons, the two products did not diﬀer much in terms of quality

(i.e., ± 0.5), whereas for drills one product was clearly superior to the

other (i.e., ± 29.3).

In the studies, each product in a pair was presented equally often

with a higher and a lower average consumer rating. Accordingly, if

participants did not consider average ratings in their choices, the pro-

duct with the higher rating should be chosen as often as the product

with the lower rating. In contrast, if participants preferred products

with higher average ratings, the higher rated product should be chosen

more frequently. Thus, in a ﬁrst step we tested whether the probability

with which the product with the higher average rating was chosen

diﬀered from 0.5 in the no single review condition.

Overall, younger participants in Study 1 strongly followed the

average ratings. As illustrated in Fig. 2 (left panel), when no consumer

review was shown to participants, they chose the higher rated product

in 80% of the cases. The choice proportions diﬀered signiﬁcantly from

0.5 for all three products (vacuum cleaner: χ

(N = 46) = 8.10,

p = 0.004; iron: χ

(N = 46) = 11.13, p < 0.001; drill:

(N = 45) = 4.61, p = 0.032).

In contrast, for older adults we did not ﬁnd an inﬂuence of average

ratings on choices (see Fig. 2, right panel). Participants chose the higher

rated product in 58% of the choices when no review was provided. This

did not signiﬁcantly diﬀer from 50% when considering all choices to-

gether, χ

(N = 163) = 1.63, p = 0.20, nor for any of the three products

separately (in all three cases p > 0.38). This, as can be seen below, does

not mean that their choices were random.

To test for the inﬂuence of product attributes on choices, we ran

multilevel mixed eﬀects logistic regressions with random intercepts for

subjects with choice of the higher rated product as a dependent variable

and product quality (z-transformed), presentation type (0 - simulta-

neous, 1 - sequential), and order (1 - First/Left, 0 - Second/Right) and

their two-way interactions as independent variables.

All models were

implemented in R using the mixed function in the afex package [ ]63

using Likelihood Ratio Tests. Post hoc contrasts were calculated with

the lsmeans package [ ].64

For younger adults we found a strong eﬀ ect of product quality on

their choices, b = 1.58, = 0.459,SE χ

(1) = 26.92, p < 0.001. In ad-

dition, we found a signiﬁcant main eﬀect of order, b = 0.733,

SE = 0.342, χ

(1) = 5.85, p = 0.016, and an interaction of order by

product quality, χ

(1) = 5.07, = 0.024, but no ep ﬀect of presentation

type. Follow-up tests of the eﬀect of product quality separately for the

two order conditions showed a large eﬀect of product quality when the

higher rated product was second/on the right side, b = 2.04, SE = 0.54,

(1) = 21.27, < 0.001. When the higher rated product was rst/onp ﬁ

the left side, the eﬀect or product quality was smaller, but still sig-

niﬁcant, b = 0.755, = 0.369,SE χ

(1) = 5.75, p = 0.0165 (see also

Fig. 2 and Fig. 3 in the supplementary online material).

No other e ectﬀ

or interaction was signi cant.ﬁ

For older adults, the product quality index also emerged as a sig-

niﬁcant predictor of choice (b = 0.381, = 0.186,SE χ

(1) = 4.76,

p = 0.029). Increasing product quality from visibly lower quality

(quality index equal to −30) to comparable quality (quality index

equal 0) and from comparable quality to visibly better quality (quality

index equal to 30) both led to an increase in predicted probability of

choosing the higher-rated product by roughly 15% (average marginal

eﬀect for the ﬁxed part of the model). This shows that older participants

paid attention to the product attributes and were more likely to choose

a product that was clearly better on the attribute dimensions. Thus it

indicates that they were not choosing randomly (see ).Fig. 2

In addition, we found an eﬀect of presentation order for older adults

(b = −0.411, SE = 0.173, χ

(1) = 5.94, p = 0.015), suggesting that

https://osf.io/3n8xw/.

We abstain from reporting statistical comparisons between age groups to

focus on the impact of the manipulated variables. However, we report addi-

tional analyses with age groups as a factor in the supplementary online mate-

rials and on OSF.

We excluded the three-way interaction of presentation type × product

quality × order from the analyses because the sample sizes within each cell

were not suﬃcient to estimate the models. In addition, we had to exclude the

interaction of presentation type with order in the analysis for the younger

adults because the model did not converge. We did not include product type in

this model because the model became unstable when product type and the

product quality index were both included. Analyses without the quality index

indicated that choices did not diﬀer signiﬁcantly between products.

Tables reporting the choice proportions of older and younger adults by

product and product quality can be found in the supplementary online mate-

rials and on OSF (folder Results): .https://osf.io/3n8xw

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

older adults were more likely to choose the option with the higher

rating when it was presented ﬁrst or on the left-hand side of the screen.

No further signiﬁcant eﬀects were found.

In sum, both younger and older adults were more likely to chose the

better rated product when it also had better attributes. Yet, the in u-ﬂ

ence of product quality on the choices of older adults was less pro-

nounced than in the case of younger adults. This can clearly be seen

when focusing on the drills, for which by design one option was clearly

better than the other one in terms of the product's attributes. When the

better drill was also recommended by average consumer ratings, (i.e., it

dominated the other product on all relevant dimensions), younger

adults chose the better drill in 100% of the cases, showing that they

clearly recognized this dominance relationship. In contrast, older adults

chose the better drill in 76% of the cases.

5.2. Inﬂuence of single aﬀect-rich consumer reviews on choice

In the next step, we investigated our second research question, i.e.

whether the single aﬀect-rich consumer reviews inﬂuenced how fre-

quently the higher rated product was chosen (see Fig. 2). To this goal

we once again ran a multilevel mixed eﬀects logistic model predicting

how often the higher rated product was chosen, now analyzing the full

data set. In addition to the predictors in the analyses above, we in-

cluded review condition and its interactions with the other predictors in

the model.

The analysis for younger adults revealed a signiﬁcant eﬀect of re-

view condition on choices, χ

(2) = 16.79, p < 0.001 and an eﬀect of

product quality, b = 0.978, SE = 0.170, χ

(1) = 48.07, p < 0.001. No

other main eﬀect or interactions reached signiﬁcance. Post-hoc con-

trasts using the Tukey method to adjust p-values indicated that the odds

of students choosing the higher-rated product were 5.43 times smaller

when it was presented with a single, aﬀect-rich, negative review than

when no reviews were included (b = 1.691, = 0.451,SE z = 3.75,

p < 0.001). Similarly, presenting the lower rated product with a single,

aﬀect-rich, positive review reduced the odds that the option with higher

average rating would be chosen by a factor of 3.56 (b = 1.271,

SE = 0.448, z = 2.84, p = 0.013). The inﬂuence of single positive and

negative consumer reviews on choices did not diﬀer signi cantly,ﬁ

z = 1.228, p = 0.437 (see Fig. 2, left panel).

For older adults, the analyses also showed signiﬁcant main eﬀects of

review condition, χ

(2) = 30.44, p < 0.001, product quality,

b = 0.448, SE = 0.110, χ

(1) = 18.44, p < 0.001, and presentation

order, b = −0.204, SE = 0.101, χ

(1) = 4.11, p = 0.0427, but no sig-

niﬁcant interactions. Post-hoc contrasts adjusted with the Tukey

method, indicated that a single, negative, aﬀect-rich review sig-

niﬁcantly reduced the probability of choosing the higher rated product

by a factor of 3.5 compared to the no review condition, b = 1.256,

SE = 0.260, z = 4.83, p < 0.001. However, in contrast to the younger

adults, a single, aﬀect-rich, positive review of the lower rated product

did not change the likelihood of choosing the higher rated product

compared to the no review condition for older adults, b = 0.094,

SE = 0.242, z = 0.387, p = 0.921 (see Fig. 2, right panel).

5.3. Perception of the aﬀect-rich reviews

The analyses of choices had shown that for younger adults a single

aﬀect-rich review that was inconsistent with the average rating reliably

reduced how often the option with higher rating was chosen. This was

the case for positive and negative aﬀect-rich reviews. For older adults,

however, we only found an inﬂuence for negative aﬀect-rich reviews.

One reason could be that older and younger adults diﬀered in how well

they understood the aﬀect-rich reviews and as how positively/nega-

tively they perceived them compared to the baseline review presented

with the other product. To investigate this question we tested whether

participants reported that they understood the reviews and whether

participants really perceived the aﬀect-rich reviews as more positive/

negative than the baseline review.

Overall, younger and older adults reported high levels of under-

standing of the consumer reviews. Average ratings were above or close

to 5 on a seven-point scale, suggesting that both age groups understood

the reviews well (see Table 1 for an overview of participants' ratings).

To test whether the review type (aﬀect-rich positive, a ect-richﬀ

negative, and baseline reviews) diﬀered in how their valence was rated

(i.e. how positive/negative they were perceived) we run a repeated

measurement analyses of variance (ANOVA) with the valence rating as

-29.3 -0.5 +0.5 +29.3

Product quality index

0.2

0.4

0.6

0.8

Prop. of part. choosing the higher rated product

Older Adults

no single review

single positive review

single negative review

-29.3 -0.5 +0.5 +29.3

Product quality index

0.2

0.4

0.6

0.8

Prop. of part. choosing the higher rated product

Younger Adults

no single review

single positive review

single negative review

Fig. 2. Proportion of participants choosing the higher rated option by review condition and product quality for young adults (left panel) and older adults (right

panel). The product quality index indicates the quality of the higher rated option. Error bars denote 1 SE.

We did not include the four-way interaction including all predictors in the

analyses because the within-cell sample sizes were not suﬃcient to estimate the

models. For younger adults we also had to exclude the three-way interactions of

presentation type × product quality × order and review condi-

tion × order × product quality because otherwise the model did not converge.

Reducing the model further does not change the conclusions.

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

dependent variable and the type of the review as predictor.

For younger adults we found a large eﬀect of the type of review on

valence ratings (F(2,922) = 1235, < 0.001,p η

= 0.73). Contrasts

conﬁrmed that the positive reviews were perceived as more positive

than the baseline reviews (F(1,461) = 134, p < 0.001, η

= 0.23) and

the negative reviews as more negative than the baseline reviews (F

(1,461) = 1136, p < 0.001, η

= 0.71). Accordingly, participants per-

ceived the reviews as diﬀerently positive/negative but the diﬀerence in

perception was larger for the negative-baseline comparison than the

positive-baseline comparison.

Similar to the younger adults, how positively older adults perceived

the reviews depended strongly on the type of the consumer review (F

(2,736) = 349.5, p < 0.001, η

= 0.49). They perceived the positive

consumer reviews as more positive than the baseline reviews (F

(1,369) = 21.79, < 0.001,p η

= 0.06) and the negative reviews as

more negative than the baseline reviews (F(1,368) = 373.5, p < 0.001,

= 0.50). However, overall and in particular for the comparison be-

tween positive and baseline reviews, the eﬀ ect size was much smaller

than for younger adults. The same pattern of results was found for all

products, although for the drills the rating of the positive review did not

diﬀer signiﬁcantly from the rating of the baseline review (p = 0.19).

6. Discussion

We found that younger and older adults were inﬂuenced by product

attributes and aﬀect-rich negative reviews. However, whereas younger

adults strongly relied on average consumer ratings and also on a ect-ﬀ

rich positive reviews, older adults did not take them into account. These

results suggests that older adults diﬀer in how they perceive the reviews

written by other consumers and how much value they assign to this

information, which has important implications for marketing directed

at older adults. In the following we discuss the results in more details

and outline implications for designing e-commerce platforms for older

adults.

6.1. Inﬂuence of product attributes

For both younger and older participants, the quality index of pro-

ducts' attributes strongly inﬂuenced choices when no review was pre-

sented, but also when single reviews were provided. This may not be

very surprising in itself, but the results are important for two reasons.

First of all, it shows that although average consumer ratings are quite

important for younger adults, product attributes were more important

for most of them and were not overruled when they were clearly in

conﬂict with the consumer ratings. Secondly, it shows that older adults

understood the task, but struggled more with identifying a better pro-

duct just from the attributes. This is most clearly seen when considering

choices for the drills. When in the no-review condition the better drill

was also recommended by average consumer ratings, (i.e., it dominated

the other product in all relevant dimensions), older adults chose the

better drill in 76% of the cases. This demonstrates that older adults

clearly paid attention to the product attributes and were not choosing

randomly. But it also suggests that older adults had problems identi-

fying a better product based on its attributes. These results resonate

with research showing declines in decision-making abilities in new and

complex tasks in older adults [31-33]. The results are also in line with

research on consumer choice ﬁnding choice deﬁcits in older adults

(e.g., [ ]).38

6.2. Inﬂuence of average consumer ratings

Younger adults were strongly inﬂuenced by aggregated consumer

ratings. In particular, when the two products they could choose from

were similar in quality and no single positive or negative review was

presented, younger adults overwhelmingly chose the higher rated

product. The majority of younger adults chose the lower rated product

only when it had clearly better attributes than the higher rated product,

although even then a sizable minority (43%) still preferred the higher

rated product. These results dovetail with previous research reporting

the importance of consumer ratings for online purchasing decisions of

younger adults [5 7– ] and also resonate with younger adults reporting

average ratings as quite important, with a score of 4.9 on a scale from 1

to 7.

In contrast, we did not ﬁnd any evidence that older adults con-

sidered average consumer ratings in their decisions. This ﬁnding is

surprising given the prevalence of consumer reports in an online con-

text and high importance assigned to them by younger adults, but it

corresponds to ﬁndings by Ma et al. [3] suggesting that older adults

perceive consumer ratings as less relevant and helpful than younger

adults.

Why older adults did not use average ratings is less clear. One

reason could be that older adults just do not value the opinion of other

consumers as much as younger adults, which suggests that they may not

be aware of how valuable this information can be. Alternatively, the

presentation of the consumer ratings may be confusing for older adults.

Older adults have worse visual acuity making it more diﬃcult to dis-

cern small diﬀerences on the screen [27 65, ]. Most consumer ratings

tend to be positive, making diﬀerences in average ratings relatively

small. Thus it is possible that older adults had problems in realizing that

a diﬀerence in average ratings of 0.5 points carries relevant informa-

tion. Future research into this is necessary to determine the factors

underlying older adults' neglect of consumer ratings and potential ways

of making this information more accessible to them.

6.3. Inﬂuence of single aﬀect-rich reviews

Both younger and older adults' choices were strongly inﬂuenced if

the higher-rated product was accompanied by a single, vivid, negative

review. Only a minority of participants picked the option with the ne-

gative review, even though they were told that the review was selected

randomly (and thus was not necessarily representative of the reviews

the product had received). These results correspond with studies re-

porting that people are more easily inﬂuenced by anecdotal or narrative

information than by statistical information when making decisions in

medical or consumer contexts [ ].18 20–

For younger adults aﬀect-rich negative and positive single reviews

that conﬂicted with the average ratings reduced how often the higher-

rated product was chosen. The eﬀect size was somewhat larger for the

aﬀect-rich negative than for the aﬀect-rich positive reviews, but the

Table 1

Overview of rating responses in both studies. Rating scale = 1 (not at all) to 7

(very much). Since each participant made multiple evaluations (after each

choice and for each review) we use robust standard errors clustered on parti-

cipants. The question regarding the importance of the single consumer review

was only asked in the conditions including a consumer review (the positive and

negative review conditions). For the ratings of the valence and understanding of

the aﬀect-rich consumer reviews data from N = 41 older participants is

missing.

Young adults Older adults

Variables Mean SE Mean SE

Online experience 5.29 0.11 1.96 0.12

Product knowledge 3.19 0.10 3.57 0.10

Importance: product features 5.70 0.08 5.37 0.11

Importance: average rating 4.90 0.11 4.09 0.12

Importance: consumer review 3.51 0.15 4.00 0.15

Understandings: negative review 5.54 0.09 5.10 0.16

Understandings: baseline review 4.93 0.12 5.19 0.14

Understandings: positive review 5.53 0.09 5.68 0.12

Valence: negative review 1.76 0.10 2.52 0.16

Valence: baseline review 5.46 0.09 5.19 0.13

Valence: positive review 6.20 0.08 5.66 0.13

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

diﬀerence was not signiﬁcant. At ﬁrst sight, the somewhat larger e ectﬀ

of negative reviews is consistent with a negativity bias. However, when

evaluating the e ect of the aﬀ ﬀect-rich reviews one must also take into

account how the aﬀect-rich reviews were perceived compared to the

baseline review shown with the other option. Due to the nature of the

baseline reviews we chose — short but in general positive statements

selected to reﬂect typical reviews found online – the aﬀect-rich negative

reviews diﬀered more strongly in their perceived valence from the

baseline reviews than the aﬀect-rich positive reviews. Nevertheless,

younger adults' choices were inﬂuenced by the aﬀect-rich positive re-

views. Indeed, the positive aﬀect-rich reviews aﬀected their choices in a

similar way as the negative aﬀ ect-rich reviews in spite of the smaller

diﬀerence in perceived valence. A ﬁnding that is in contradiction to

research showing a negativity bias [11 13, ]. However, it is in line with

research suggesting that negativity bias may be limited to hedonic

goods [12] as the products we used were more of a utilitarian nature. In

addition, although we can’t distinguish between the features that made

the aﬀect-rich reviews more convincing than the baseline reviews in the

current study, it suggests that beside valence other features such as the

aﬀective nature and the level of detail of a review may aﬀect how much

weight participants give them in their decisions.

In contrast, older adults were strongly inﬂuenced by the a ect-richﬀ

negative reviews, but not at all by the aﬀect-rich positive reviews. At

ﬁrst glance, these results seem to be at odds with research proposing a

focus on positive information in older adults [49,50]. However, ac-

cording to socio-emotional selectivity theory older adults focus on po-

sitive information stems from a shift to emotional meaningful

goals [49]. With emotional goals in mind, avoiding products with ne-

gative reviews, and thus potential losses, seems a rational strategy as it

minimizes negative emotions that could be caused by choosing the

wrong product [51]. This suggests that in decision making tasks older

adults may be more likely to exhibit a negativity than a positivity bias.

Why older adults did not consider the aﬀect-rich positive reviews at

all is an interesting question. One possibility is that older adults did not

diﬀerentiate between the aﬀect-rich positive and the baseline reviews

as much as younger adults did. Although older adults on average rated

the aﬀect-rich positive reviews as signiﬁcantly more positive than the

baseline reviews, the diﬀerence was smaller than for the younger adults

and the diﬀerence between the aﬀect-rich negative reviews and the

baseline reviews. Accordingly, it is possible that for older adults the

perceived diﬀerences in valence was not strong enough to aﬀect their

choices. This suggests that, although in old age the understanding of

emotions in written texts and of narrative texts seems to be largely

una ,ﬀected [45 46], older adults may still have more diﬃculty in per-

ceiving nuances in emotional intensity [ ].66

6.4. Implications for a design of rating systems

We found clear indications that older adults were strongly in u-ﬂ

enced by aﬀect-rich negative consumer reviews, but not by better

average consumer ratings or aﬀect-rich positive reviews. This suggests

that social media and WOM communication directed at older adults do

not require enthusiastic and vivid descriptions whereas younger adults

can be convinced by strongly positive reviews.

The ﬁnding that older adults did not consider average ratings in our

study is intriguing. The strong eﬀect of the negative reviews suggests

that they are not completely insensitive to consumer recommendations.

Alternatively, it is possible that the diﬀerences in ratings were too small

to carry meaning for older adults. Here, it would be important to choose

designs that make it easier for older adults to recognize diﬀerences in

ratings that otherwise they may not be able to discriminate.

Our results show not only that ratings and reviews play di erentﬀ

roles in the purchasing decision process (see [67]), but also that the

importance of reviews and ratings varies between younger and older

adults. Therefore, to ensure the same level of comfort for both groups

while making decisions about purchases, the user interface should be

adapted accordingly (better visibility of negative reviews and less stress

on average ratings for older adults).

The observed diﬀerences between younger and older adults, how-

ever, have more far-reaching consequences than just regarding the

personalization of the way ratings and reviews are displayed and force

researchers to re-examine existing methods of mitigating attacks on e-

commerce ratings systems. To assure the same level of protection for

both groups of users ratings, systems should be resistant not only to

attacks on average ratings (malicious increasing or decreasing) but also

to the injection of single, fake, vivid, negative reviews.

6.5. Limitations

In this study we found strong age diﬀerences in the inﬂuence of

aggregated and single consumer reviews on choices. However, it is

important to take into account several limitations of our study.

For one, we used a cross-sectional design. Thus, it is impossible to

separate age diﬀerences from cohort eﬀects. Although we did not ndﬁ

any evidence that experience with online shopping inﬂuenced older

participants' choices (for details see supplementary online materials), it

is possible that current older adults are just less accustomed to con-

sumer ratings and reviews in general, and thus give this information

less weight. However, this may change when a new generation of more

internet-savvy individuals approaches old age.

In addition, our two samples are from a single country and di eredﬀ

not only in age but also in the gender composition, with younger adults

being mostly male and older adults mostly female. Gender partly in-

ﬂuenced product knowledge, but otherwise we did not ﬁnd an eﬀect of

gender on choices, suggesting that the results are not dependent on

gender (for details see supplementary online materials). In addition,

although we did not ﬁnd diﬀerence on education on choices, the

younger adults sample consisted of students, whereas older adults were

recruited from the community. In sum, we cannot exclude that sample

diﬀerences could have played some role and that generalizability may

be limited.

In the studies we used a set of utilitarian products with hypothetical

choices. Products were selected to be typical domestic equipment that

most people own at one point in their life, but buy only rarely, to de-

crease diﬀerences in product knowledge between participants.

However, decision processes may depend on the type of products se-

lected. For instance, the negativity bias has been shown to be limited to

hedonic goods [12]. Thus, it is possible that older adults may take

average consumer ratings more into account while making real choices

or while choosing between other types of products such as experiential

services.

To measure the inﬂuence of product attributes on choice, we used

an index of relative product quality that assumed all attributes to be

equally important. Given that subjective importance of attributes will

diﬀer between individuals, our measure of product quality most likely

underestimates the inﬂuence of product attributes on choices.

Lastly, the consumer reviews we used as a baseline comparison were

not at the (valence) midpoint between negative and positive reviews

but had a positive valence. We chose these short but positive statements

as baseline reviews because they are more typical than neutral reviews

[4 61, ,62] but still diﬀer in their valence from the strongly positive re-

views.

7. Conclusions

Our results show not only that ratings and reviews play di erentﬀ

roles in purchasing decisions (see [67]), but also that the importance of

reviews and ratings varies between younger and older adults. Whereas

students were strongly inﬂuenced by average consumer ratings and

positive aﬀect-rich reviews, the older adults in our sample gave little

importance to these types of consumer information. However, younger

and older adults were quite strongly inﬂuenced by aﬀect-rich negative

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

reviews — even if these were unrepresentative of the product reviews.

These results highlight important age diﬀerence in consumer behavior,

raising questions about the utility of consumer reviews for older adults,

as well as how consumer reviews should be presented. To ensure the

same level of comfort for both groups when making decisions about

purchases, at the very least the user interface must be adapted ac-

cordingly (better visibility of negative reviews and less stress on

average ratings for older adults).

Acknowledgements

The work was supported by a grant of the Swiss National Science

Foundation to the ﬁrst author [No. 157432]. This work was also par-

tially supported by European Union's Horizon 2020 Research and

Innovation Programme under the Marie Skłodowska-Curie grant

agreement [No. 690962]. Declarations of interest: none.

Appendix A. Supplementary materials: analyses and stimulus

materials

Supplementary data to this article can be found online at https://

doi.org/10.1016/j.dss.2018.05.006.

References

[1] J.W. Lian, D.C. Yen, Online shopping drivers and barriers for older adults: age and

gender diﬀerences, Computers in Human Behavior 37 (2014) 133–143.

[2] M. Law, M. Ng, Age and gender diﬀerences: understanding mature online users with

the online purchase intention model, Journal of Global Scholars of Marketing

Science 26 (3) (2016) 248–269.

[3] Y.J. Ma, H. Kim, H.-h. Lee, Eﬀect of individual diﬀerences on online review per-

ception and usage behavior: the need for cognitive closure and demographics,

Journal of the Korean Society of Clothing and Textiles 36 (12) (2012) 1270–1284.

[4] P.Y. Chen, S.y. Wu, J. Yoon, The impact of online recommendations and consumer

feedback on sales, ICIS 2004 Proceedings, 2004, p. 58.

[5] K. Floyd, R. Freling, S. Alhoqail, H.Y. Cho, T. Freling, How online product reviews

aﬀect retail sales: a meta-analysis, Journal of Retailing 90 (2) (2014) 217 232,–

https://doi.org/10.1016/j.jretai.2014.04.004.

[6] R.A. King, P. Racherla, V.D. Bush, What we know and don’t know about online

word-of-mouth: a review and synthesis of the literature, Journal of Interactive

Marketing 28 (3) (2014) 167–183.

[7] N. Purnawirawan, M. Eisend, P. De Pelsmacker, N. Dens, A meta-analytic in-

vestigation of the role of valence in online reviews, Journal of Interactive Marketing

31 (2015) 17–27, .https://doi.org/10.1016/j.intmar.2015.05.001

[8] Y. Liu, Word of mouth for movies: its dynamics and impact on box oﬃce revenue,

Journal of Marketing 70 (3) (2006) 74–89.

[9] E. Maslowska, E.C. Malthouse, V. Viswanathan, Do customer reviews drive pur-

chase decisions? The moderating roles of review exposure and price, Decision

Support Systems 98 (2017) 1–9.

[10] S. Karimi, F. Wang, Online review helpfulness: impact of reviewer proﬁle image,

Decision Support Systems 96 (2017) 39–48.

[11] J. Lee, D.h. Park, I. Han, The eﬀect of negative online consumer reviews on product

attitude: an information processing view, Electronic Commerce Research and

Applications 7 (2008) 341–352.

[12] S. Sen, D. Lerman, Why are you telling me this? An examination into negative

consumer reviews on the web, Journal of Interactive Marketing 21 (4) (2007)

76 94– .

[13] C. Betsch, N. Haase, F. Renkewitz, P. Schmid, The narrative bias revisited: what

drives the biasing inﬂuence of narrative information on risk perceptions? Judgment

and Decision Making 10 (3) (2015) 241–264.

[14] P. Rozin, E.B. Royzman, Negativity bias, negativity dominance, and contagion,

Personality and Social Psychology Review 5 (4) (2001) 296–320.

[15] P.F. Wu, In search of negativity bias: an empirical study of perceived helpfulness of

online reviews, Psychology & Marketing 30 (11) (2013) 971–984.

[16] BrightLocal, Local Consumer Review Survey 2016, (2016) https://www.brightlocal.

com/learn/local-consumer-review-survey/.

[17] S. Hong, H.S. Park, Computer-mediated persuasion in online reviews: statistical

versus narrative evidence, Computers in Human Behavior 28 (3) (2012) 906–919.

[18] M. Ziegele, M. Weber, Example, please! Comparing the eﬀects of single customer

reviews and aggregate review scores on online shoppers' product evaluations,

Journal of Consumer Behaviour 14 (2015) 103–114.

[19] C. Betsch, C. Ulshöfer, F. Renkewitz, T. Betsch, The inﬂuence of narrative v. sta-

tistical information on perceiving vaccination risks, Medical Decision Making 31 (5)

(2011) 742–753, .https://doi.org/10.1177/0272989X11400419

[20] P.A. Ubel, C. Jepson, J. Baron, The inclusion of patient testimonials in decision aids,

Medical Decision Making 21 (1) (2001) 60–68.

[21] A. Winterbottom, H.L. Bekker, M. Conner, A. Mooney, Does narrative information

bias individual's decision making? A systematic review, Social Science and

Medicine 67 (12) (2008) 2079–2088.

[23] P.B. Baltes, U.M. Staudinger, U. Lindenberger, Lifespan psychology: theory and

application to intellectual functioning. Annual Review of Psychology 50 (1999)

471 .–507, https://doi.org/10.1146/annurev.psych.50.1.471

[24] L.L. Carstensen, The inﬂuence of a sense of time on human development, Science

312 (5782) (2006) 1913–1915.

[25] E. Peters, T.M. Hess, D. Västfjäll, C. Auman, Adult age diﬀerences in dual in-

formation processes: implications for the role of aﬀective and deliberative processes

in older adults' decision making, Perspectives on Psychological Science 2 (1) (2007)

1 23– .

[26] R. Mata, T. Pachur, B. von Helversen, R. Hertwig, J. Rieskamp, L. Schooler,

Ecological rationality: a framework for understanding and aiding the aging decision

maker, Frontiers in Decision Neuroscience 6 (Article 19) (2012) 1–6, https://doi.

org/10.3389/fnins.2012.00019.

[27] W.A. Rogers, A.J. Stronge, A.D. Fisk, Technology and Aging, Reviews of human

factors and ergonomics 1 (1) (2005) 130–171.

[28] T.A. Salthouse, Mental exercise and mental aging: evaluating the validity of the use“

it or lose it” hypothesis, Perspectives on Psychological Science 1 (1) (2006) 68–87.

[29] T. Salthouse, Consequences of age-related cognitive declines, Annual Review of

Psychology 63 (2012) 201–226.

[30] M.L. Finucane, C.K. Mertz, P. Slovic, E.S. Schmidt, Task complexity and older

adults' decision-making competence, Psychology and Aging 20 (1) (2005) 71–84.

[31] R. Frey, R. Mata, R. Hertwig, The role of cognitive abilities in decisions from ex-

perience: age diﬀerences emerge as a function of choice set size, Cognition 142

(2015) 60–80, .https://doi.org/10.1016/j.cognition.2015.05.004

[32] R. Mata, L. Nunes, When less is enough: cognitive aging, information search, and

decision quality in consumer choice, Psychology and Aging 25 (2010) 289 298,–

https://doi.org/10.1037/a0017927.

[33] R. Mata, L.J. Schooler, J. Rieskamp, The aging decision maker: cognitive aging and

the adaptive selection of decision strategies, Psychology & Aging 22 (2007)

101037/0882 7974224796– .

[34] B. von Helversen, R. Mata, Losing a dime with a satisﬁed mind: positive a ectﬀ

predicts less search in sequential decision making, Psychology and Aging 27 (4)

(2012) 825–839, .https://doi.org/10.1037/a0027845

[35] G. Gigerenzer, P.M. Toddthe ABC Research Group, Simple Heuristics That Make Us

Smart, Oxford University Press, 1999.

[36] R. Mata, B. von Helversen, J. Rieskamp, Learning to choose: cognitive aging and

strategy selection learning in decision making, Psychology and Aging 25 (2) (2010)

299 .–309, https://doi.org/10.1037/a0018923

[37] J.A. Mikels, C.E. Löckenhoﬀ, S.J. Maglio, L.L. Carstensen, M.K. Goldstein,

A. Garber, Following your heart or your head: focusing on emotions versus in-

formation diﬀerentially inﬂuences the decisions of younger and older adults,

Journal of Experimental Psychology: Applied 16 (1) (2010) 87.

[38] C.A. Cole, S.K. Balasubramanian, Age diﬀerences in consumers' search for in-

formation: public policy implications, Journal of Consumer Research 20 (1993)

157 169– .

[39] C.M. Schaninger, D. Sciglimpaglia, The inﬂuence of cognitive personality traits and

demographics on consumer information acquisition, Journal of Consumer Research

8 (2) (1981) 208–216.

[40] R. Lambert-Pandraud, G. Laurent, E. Lapersonne, Repeat purchasing of new auto-

mobiles by older consumers: empirical evidence and interpretations, Journal of

Marketing 69 (2) (2005) 97–113.

[41] S.M. Carpenter, C. Yoon, Aging and consumer decision making, Annals of the New

York Academy of Sciences 1235 (1) (2011) 1–12.

[42] Q. Ma, K. Chen, A.H.S. Chan, P.L. Teh, Acceptance of ICTs by older adults: a review

of recent studies, International Conference on Human Aspects of IT for the Aged

Population, Springer, 2015, pp. 239–249.

[43] A. Ahmed, A.S. Sathish, Determinants of online shopping adoption: meta analysis

and review, European Journal of Social Sciences 49 (4) (2015) 483–510.

[44] G. Cohen, Language comprehension in old age, Cognitive Psychology 11 (4) (1979)

412 429– .

[45] R. De Beni, E. Borella, B. Carretti, Reading comprehension in aging: the role of

working memory and metacomprehension, Aging, Neuropsychology, and Cognition

14 (2) (2007) 189–212, .https://doi.org/10.1080/13825580500229213

[46] L.H. Phillips, R.D. MacLean, R. Allen, Age and the understanding of emotions

neuropsychological and sociocognitive perspectives, The Journals of Gerontology.

Series B, Psychological Sciences and Social Sciences 57 (6) (2002) 526–P530.

[47] L.L. Carstensen, Motivation for social contact across the life span: a theory of so-

cioemotional selectivity, Nebraska symposium on motivation, vol. 40, 1993, pp.

209 254– .

[48] S.T. Charles, L.L. Carstensen, Social and emotional aging, Annual Review of

Psychology 61 (2010) 383–409.

[49] A.E. Reed, L. Chan, J.A. Mikels, Meta-analysis of the age-related positivity e ect:ﬀ

age diﬀerences in preferences for positive over negative information, Psychology

and Aging 29 (1) (2014) 1–15.

[50] H.H. Fung, L.L. Carstensen, Sending memorable messages to the old: age di erencesﬀ

in preferences and memory for advertisements, Journal of Personality and Social

Psychology 85 (1) (2003) 163–178.

[51] M.K. Depping, A.M. Freund, Normal aging and decision making: the role of moti-

vation, Human Development 54 (6) (2011) 349–367.

[52] M.J. Frank, L. Kong, Learning to avoid in older age, Psychology and Aging 23 (2)

(2008) 392–398, .https://doi.org/10.1037/0882-7974.23.2.392

[53] D. Hämmerer, S.C. Li, V. Müller, U. Lindenberger, Life span diﬀerences in electro-

physiological correlates of monitoring gains and losses during probabilistic re-

inforcement learning, Journal of Cognitive Neuroscience 23 (3) (2011) 579–592.

[54] B. Eppinger, N.W. Schuck, L.E. Nystrom, J.D. Cohen, Reduced striatal responses to

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

reward prediction errors in older compared with younger adults, Journal of

Neuroscience 33 (24) (2013) 9905–9912.

[55] A. Mantonakis, P. Rodero, I. Lesschaeve, R. Hastie, Order in choice: eﬀects of serial

position on preferences, Psychological Science 20 (11) (2009) 1309–1312, https://

doi.org/10.1111/j.1467-9280.2009.02453.x.

[56] C. Mogilner, B. Shiv, S.S. Iyengar, Eternal quest for the best: sequential (vs. si-

multaneous) option presentation undermines choice commitment, Journal of

Consumer Research 39 (6) (2013) 1300–1312, .https://doi.org/10.1086/668534

[57] A. Dieckmann, K. Dippold, Compensatory versus noncompensatory models for

predicting consumer preferences, Judgment and Decision Making 4 (3) (2009)

200 213– .

[58] W. Kopeć, K. Skorupska, A. Jaskulska, K. Abramczuk, R. Nielek, A. Wierzbicki,

LivingLab PJAIT: towards better urban participation of seniors, Proceedings of the

International Conference on Web Intelligence, ACM, 2017, pp. 1085–1092.

[59] W. Kopeć, B. Balcerzak, R. Nielek, G. Kowalik, A. Wierzbicki, F. Casati, Older adults

and hackathons: a qualitative study, Empirical Software Engineering (2017) 1–36.

[60] W. Kopeć, K. Abramczuk, B. Balcerzak, M. Juźwin, K. Gniadzik, G. Kowalik,

R. Nielek, A location-based game for two generations: teaching mobile technology

to the elderly with the support of young volunteers, eHealth 360, Springer, 2017,

pp. 84–91.

[61] D. Godes, D. Mayzlin, Using online conversations to study word-of-mouth com-

munication, Marketing Science 23 (4) (2004) 545–560.

[62] P. Resnick, R. Zeckhauser, J. Swanson, K. Lockwood, The value of reputation on

eBay: a controlled experiment, Experimental Economics 9 (2) (2006) 79–101.

[63] H. Singmann, B. Bolker, J. Westfall, F. Aust, afex: Analysis of Factorial Experiments,

(2016) https://CRAN.R-project.org/package=afex r package version 0.16-1.

[64] R.V. Lenth, Least-squares means: the R package lsmeans, Journal of Statistical

Software 69 (1) (2016) 1–33, .https://doi.org/10.18637/jss.v069.i01

[65] F. Schieber, Human factors and aging: identifying and compensating for age-related

deﬁcits in sensory and cognitive function, Impact of technology on successful aging,

2003, pp. 42–84.

[66] L.H. Phillips, R. Allen, Adult aging and the perceived intensity of emotions in faces

and stories, Aging Clinical and Experimental Research 16 (3) (2004) 190–199.

[67] N. Hu, N.S. Koh, S.K. Reddy, Ratings lead you to the product, reviews help you

clinch it? The mediating role of online review sentiments on product sales, Decision

Support Systems 57 (2014) 42–53.

Bettina von Helversen is an assistant professor for cognitive decision psychology at the

University of Zurich. She is interested in understanding and modelling how people make

judgments and decisions. In her work she focuses on the diﬀerent cognitive strategies

people use to solve these tasks and the factors that inﬂuence strategy selection such as

task structure, memory, aﬀect or stress and how these change over the life span.

Katarzyna Abramczuk is an assistant professor for mathematical sociology at the

University of Warsaw. She is interested in the interplay between cognitive and social

processes. She designs studies and models to analyze how macro reality is built by micro

strategies and how individual choices are inﬂuenced by the macro environment. She pays

special attention to the role of ICT in this context.

Wiesław Kopeć is part of an international social informatics research team at Polish-

Japanese Academy of Information Technology (PJAIT) in Poland, doing research on ICT

technologies dedicated to older adults. He combines high-level IT and business experience

with rich academic background. Recently his interests extend to social aspects of applying

modern ICT with a special focus on HCI in the context of participatory design and ad-

vanced methods like Complex Event Processing, Machine Learnig and Big Data for sup-

porting educational and business value generation processes.

Radoslaw Nielek is a head of the research group working on ICT technologies dedicated

to older adults and an assistant professor at Polish-Japanese Academy of Information

Technology (PJAIT). He received his Ph.D. degree from PJAIT Warsaw, Poland and

Bachelor's Degree in Production Engineering and Management of Szczecin University of

Technology. His research interests include application of ICT technologies for improving

quality of life of older adults and limiting negative consequences of demographic changes.

B. von Helversen et al. Decision Support Systems 113 (2018) 1–10

Bấm Tải xuống để xem toàn bộ.

| 1/10

| | Tải xuống

Preview text:

Decision Support Systems 113 (2018) 1–10
Contents lists available at ScienceDirect Decision Support Systems
journal homepage: www.elsevier.com/locate/dss
Inﬂuence of consumer reviews on online purchasing decisions in older and younger adults
Bettina von Helversena,*, Katarzyna Abramczukb, Wiesław Kopećc, Radoslaw Nielekc
a Department of Psychology, University of Zurich, Switzerland
b Institute of Sociology, University of Warsaw, Poland
c Polish Japanese Academy of Information Technology, Poland A R T I C L E I N F O A B S T R A C T Keywords:
We investigated how product attributes, average consumer ratings, and single aﬀect-rich positive or negative Consumer decision making
consumer reviews inﬂuenced hypothetical online purchasing decisions of younger and older adults. In line with Older adults
previous research, we found that younger adults used all three types of information: they clearly preferred Consumer ratings
products with better attributes and with higher average consumer ratings. If making a choice was diﬃcult Consumer reviews
because it involved trade-oﬀs between product attributes, most younger adults chose the higher-rated product. Anecdotal evidence
The preference for the higher-rated product, however, could be overridden by a single aﬀect-rich negative or
positive review. Older adults were strongly inﬂuenced by a single aﬀect-rich negative review and also took into
consideration product attributes; however, they did not take into account average consumer ratings or single
aﬀect-rich positive reviews. These results suggest that older adults do not consider aggregated consumer in-
formation and positive reviews focusing on positive experiences with the product, but are easily swayed by
reviews reporting negative experiences. 1. Introduction
are presented simultaneously or sequentially and which product is
presented as the ﬁrst/on the left.
Understanding how people make online purchasing decisions is of
In the following, we ﬁrst review the literature on the inﬂuence of
growing importance. With an increase of 19.9% in 2016 and a fore-
consumer ratings and reviews on online purchasing decisions and on
casted growth of 17.5% for 2017, global business to consumer (B2C) e-
how decision making processes change with age. Then, we report two
commerce is now accounting for 8.7% of retail sales worldwide.1
experimental studies investigating how younger and older adults use
Overall, e-commerce is still dominated by younger and middle-aged
consumer reviews in hypothetical online purchasing decisions. Finally,
consumers, but older consumers (55-year-old and older) are increas-
we discuss the results of the studies and consequences of our ﬁndings
ingly buying goods or services online [1]. So far most research has
for designing e-commerce systems.
focused on younger adults, leaving it unclear how older adults deal with
the challenges involved in online consumer decisions (for notable ex- 2. Related work ceptions see [1–3]).
The goal of the present research is to contribute to understanding
2.1. Inﬂuence of consumer reviews on attitudes and purchasing intentions
how older adults make on-line purchasing decisions. Do they diﬀer in
their decision process from younger adults? What information do they
The eﬀect of consumer reviews on online decisions is widely re-
consider? And last but not least: how can we use this knowledge to
cognized. Numerous studies have shown that consumer ratings and
ensure better decision making on their part? We focus on how older
reviews impact people's purchasing behavior and intentions, as well as
adults use three main types of information: product attributes, average
attitudes towards products and retailers (e.g., [4–6]).
consumer ratings, and single positive and negative reviews that contain
According to recent meta-analyses, the most important features in-
an aﬀect-rich and vivid description of the reviewers' experiences. We
ﬂuencing sales and attitudes are the valence and the volume of reviews
also take into account how the products are presented i.e. whether they
[5,7]. In general, more positive reviews increase sales and attitudes,
* Corresponding author at: University of Zurich, Binzmühlestr. 14, Box 19, 8050 Zürich, Switzerland.
E-mail address: b.vonhelversen@psychologie.uzh.ch (B. von Helversen).
1 https://www.ecommercewiki.org/Prot:Global_B2C_Ecommerce_Report_2016.
https://doi.org/10.1016/j.dss.2018.05.006
Received 13 November 2017; Received in revised form 15 March 2018; Accepted 28 May 2018 Available online 18 June 2018
0167-9236/ © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/). B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
whereas negative reviews reduce them (e.g., [5,8]). Their eﬀect, how- [36,37].
ever, also depends on review exposure [9], the characteristics of the
In line with this, research in consumer contexts indicates that older
reviewer [10], and the source of the review [5].
adults have more diﬃculties when options diﬀer on many attributes
Although positive and negative reviews can sway consumers' be-
(e.g., [38]). Furthermore, older adults tend to search for less informa-
havior, some research has indicated that they diﬀer in their impact.
tion than younger adults while making consumer decisions [39] and
Purnawirawan et al. [7] reported that negative reviews had the stron-
prefer to stick to the same brand [40,41].
gest eﬀect on attitudes and usefulness, suggesting that negative reviews
Relatively little research has considered how older adults navigate
may carry more weight than positive reviews [11,12] — a ﬁnding that
the online world, but the number of studies is rising with more elderly
resonates with research in further areas of communication [13,14].
adults using the Internet [1,2,42]. Still, older adults seem to be more
However, other research has reported that with consumer reviews the
reluctant than younger adults to use e-commerce and are less familiar
negativity bias is limited to hedonic goods [12]. Furthermore, Wu [15]
with computer technology in general [1,43]. In addition, a study in
suggested that consumers may not weigh negative reviews more
Hong Kong found that older adults perceived online purchases as less
strongly per se, but perceive them as more informative because they
easy than middle-aged adults [2]. Most relevant, Ma et al. [3] found
often are rarer and of higher quality.
that age was negatively related to self-reported perceived beneﬁts of
Besides the valence of the review, the format of the information
consumer reviews, their persuasiveness, and use.
matters. Online platforms often provide consumer reviews in two for-
mats: average ratings giving an overview over the overall perceived
2.3. Aging and processing of aﬀect-rich consumer reviews
quality of the product (i.e., statistical information) and single reviews
that contain personal narratives of experiences made with a speciﬁc
Although overall text comprehension suﬀers in old age [44], older
product. The relative importance of these types of information is still
adults' ability to process narrative and emotional texts is well preserved
under debate. A recent consumer survey indicated that customers rate
[45,46]. Accordingly, single consumer reviews presented in a narrative
average ratings as most important [16]. Hong and Park [17] found that
format may present a source of information that is easily accessible for
both statistical information and narrative information are equally
older adults and thus exert a strong inﬂuence on their decisions, even if
convincing, whereas Ziegele and Weber [18] reported that although
the information is not representative of overall consumer opinions. Yet,
average ratings were considered important, single vivid narratives
whether older adults are equally inﬂuenced by negative and positive
overrode average ratings. This picture is consistent with research in the
aﬀect-rich reviews is unclear.
medical domain showing that anecdotal or narrative evidence can be
Besides changes in cognitive processing, aging is also related to
more convincing than statistical evidence of treatment quality [19-21].
changes in aﬀect and motivation, which may inﬂuence the information
The question of how strongly single reviews inﬂuence behavior is
older adults pay attention to. Socio-emotional selectivity theory pro-
particularly important because people often only read a small number
poses to that with increasing age people focus more on maintaining
of reviews before making a decision, focusing on the most recent re-
positive aﬀect and less on increasing their knowledge [24,47]. In line views [16].
with this idea, older adults have been shown to report improved psy-
In sum, research suggests that younger adults' purchasing decisions
chological well-being and lower levels of negative aﬀect [48]. More-
are strongly inﬂuenced by average consumer ratings. Average ratings of
over, older adults often show a positivity eﬀect; that is, they exhibit a
a product, however, may loose their inﬂuence on decisions if they are
preference for positive over negative information in processing in-
inconsistent with a well-written, single review [18]. Furthermore, some
formation [49]. Speciﬁcally, older adults pay more attention to positive
research indicates that negative reviews exert stronger inﬂuence than
information and remember it better than negative information [49,50].
positive ones [7] suggesting that negative single reviews may carry
At face value the positivity eﬀect would suggest that older adults
more weight than positive single reviews. In contrast, little is known
will pay more attention to and consequently are more inﬂuenced by
about how older adults make online consumer decisions and react to
positive reviews. However, a focus on maintaining positive aﬀect may consumer ratings and reviews.
not always go hand in hand with a focus on positive information. In this
vein, Depping and Freund [51] proposed that to maintain positive af-
2.2. Aging, decision making, and online purchasing
fect older adults focus on preventing losses, resulting in a higher sen-
sitivity and more attention to losses. In line with this idea, it has been
Aging is characterized by a number of changes in cognitive abilities,
shown that in learning paradigms older adults learn better from nega-
aﬀect and motivation [23-25] that impact how older adults make de-
tive than from positive consequences [52,53] — a bias that is not shown cisions (e.g., [25,26]).
by younger adults [54]. A focus on preventing losses, however, suggests
In terms of cognitive abilities, growing old is related to a decrease in
that older adults should be inﬂuenced more strongly by negative re-
ﬂuid cognitive abilities such as working memory capacity, processing views.
speed and visual processing, resulting in older adults having diﬃculties
in a number of cognitive tasks (e.g., [27–29]). This age-related decline 2.4. Presentation of options
also aﬀects the decision making process. Older adults tend to perform
worse than younger adults, in particular, if tasks are complex, demand
In laboratory decision tasks, options are usually presented si-
the processing of large amounts of information (e.g., [30,31]), or re-
multaneously, side by side. However, when purchasing products online quire learning [32,33].
consumers often need to consider options sequentially. Although in
Despite the decline of ﬂuid abilities, older adults show an increase
principle the decision task is the same, simultaneous or sequential
in crystallized abilities; that is, higher levels of declarative knowledge
presentations can aﬀect the decision process. Presenting options se-
and experience [23]. Using this knowledge and experience, older adults
quentially can result in order eﬀects, leading often to a preference for
can devise strategies to compensate for their limited ﬂuid cognitive
the ﬁrst option (e.g., [55]). Furthermore, people seem to be more sa-
abilities (e.g., [26]). Speciﬁcally, they are more selective in their in-
tisﬁed with choices from simultaneous presented options (e.g., [56]).
formation search and frequently rely on less information-intensive
Last but not least, decision processes may change depending on the
strategies [26,34]. Moreover, older adults may simplify decision pro-
presentation with simultaneous presentation facilitating attribute-wise
blems by focusing more on aﬀective cues [25]. Although these simpler
comparisons, whereas presenting a single option may lead to more al-
strategies often perform somewhat worse than more information-in-
ternative-wise comparisons (see, [57]). Although the inﬂuence of the
tensive strategies, they perform very well if they are suited to the task
presentation type on choices is not the focus of our research, we ma-
(e.g., [35]). Accordingly, the loss in decision quality can be quite small nipulated whether products were presented sequentially or 2 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
Fig. 1. Exemplary products card showing a choice between vacuum cleaners in the positive review condition. The aﬀect-rich positive review is presented for the
lower rated product (left) and the short baseline review for the higher rated product (right).
simultaneously to ensure that eﬀects of average ratings and single
less information than younger adults [3,26].
narrative reviews are not limited to one type of presentation.
Thirdly, we investigate the relative inﬂuence of positive and nega-
tive, aﬀect-rich reviews on older adults' choices. For a stronger eﬀect of
3. Predictions and research questions
positive aﬀect-rich reviews speaks the fact that older adults have been
shown to pay more attention to positive information [49]. On the other
We investigated three problems. First, we wanted to know whether
hand, Depping and Freund [51] argued that older adults are motivated
older and younger adults rely on average consumer ratings and how
to prevent losses. A focus on preventing losses, in turn, should result in
this depends on products' characteristics and their presentation.
older adults being more strongly inﬂuenced by negative consumer re-
Second, we examined whether single, vivid, and aﬀect-rich positive and views.
negative reviews can override their preferences for products with
Lastly, we examine how older adults perceive the aﬀect-rich con-
higher average consumer ratings. Third, we inquired how these two
sumer reviews in comparison to the baseline reviews we used. Although
groups perceive the aﬀect-rich reviews. To analyze these three pro-
comprehension of emotional texts is fairly well preserved in older
blems we conducted two empirical studies, one with young adults
adults [45,46], in general text comprehension is lower in older
(Study 1) and one with older adults (Study 2).
adults [44] and older adults have less experience with online shopping.
In Study 1 we expect to replicate the main ﬁndings from the lit-
Thus, it is possible that older adults will report problems in under-
erature. For one, we expect that young adults will in general prefer
standing reviews or perceive less of a diﬀerence in valence between
options that have higher average ratings to options with lower average
aﬀect-rich consumer reviews and baseline reviews.
ratings (e.g., [5]). Secondly, following Ziegele and Weber [18] we ex-
pect that preferences for options with higher average ratings will be 4. Methods
reduced when a single aﬀect-rich review favors the option with the
lower rating. In addition, we aim to examine whether a single negative
During the studies participants were presented with pairs of house-
consumer review will have a stronger eﬀect than a single positive
hold products (for example two vacuum cleaners) and had to indicate
consumer review, following up on research suggesting a bias for ne-
for each pair which of the two options they would prefer to buy. gative information.
Products were presented on cards and described by four relevant at-
In Study 2 we expect that older adults will prefer options with better
tributes (e.g., prize, power). In addition to the products' attributes, an
attributes but that their choices will be more noisy due to the decrease
average consumer rating was shown for each product. All average
in decision making capacities in older adults [33]. Secondly, we aim to
ratings were positive but one product was always rated somewhat
test whether older adults will also prefer options with better average better than the other product.
ratings. On the one hand, Ma et al. [3] report that older adults do not
We tested three between-participants conditions that varied whe-
trust consumer ratings, indicating that they may not pay attention to
ther a single written review was shown in addition to the average
this information. On the other hand, if older adults recognize the value
consumer rating and the aﬀective content of this review: In the “no
of average consumer ratings, they might focus even more strongly on
single review condition”, participants only received information about
this information than younger adults as older adults tend to consider
average consumer ratings. This condition allowed us to test whether 3 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
participants relied on average consumer rating in their choices. In the
0.6 points higher than the other product, reﬂecting typical rating dif-
“positive single review condition” the lower rated product was pre-
ferences found on online retail websites. Which of the two products in a
sented together with a highly positive, vivid, and aﬀect-rich review
pair was presented with the better rating and which product was pre-
while the higher rated product was presented with a somewhat positive
sented ﬁrst/on the left side of the screen was counterbalanced across
but short baseline review. In the “negative single review condition” the
participants (within each of the six conditions) to separate the inﬂuence
higher rated product was presented with a highly negative, vivid, and
of average ratings from the inﬂuence of product attributes on choices.
aﬀect-rich review while the lower rated product was presented with the
In addition to the average rating we showed the distributions of the
baseline review. Thus, in both conditions the single review was in-
ratings below the average rating (see Fig. 1). The number of ratings was
consistent with the average consumer rating allowing us to test whether
kept similar across all products (around 150).
it inﬂuences how frequently the higher rated product is chosen. Fig. 1
illustrates a choice in the positive single review condition.
In addition, we varied presentation-type (simultaneous vs. sequen- 4.2.3. Single consumer reviews
tial presentation of the options) between participants resulting in a 3
Depending on the single review condition, participants received a
(single review condition) by 2 (presentation type) design.
single narrative consumer review in addition to the product information
Studies were conducted by the Polish Japanese Academy of
and the consumer ratings. The reviews were adapted from reviews of
Information Technology (PJAIT). They were approved by the Ethics
similar products taken from a website of a large online retailer. They
Committee of the Department of Psychology at the University of Basel.
were presented to subjects as randomly selected consumer reviews to
The study with older adults was conducted on the premises of PJAIT
emphasize that any of the reviews for the product could have been
supervised by the research team. In the case of younger participants
selected. In each pair one product received an aﬀect-rich review,
(i.e., students of PJAIT), the study was run as an unsupervised online
whereas the other product was presented with a baseline review. The survey.
baseline review was a short (typically one sentence) comment that was
in general positive but lacked detail, vividness, and emotional content 4.1. Participants
such as “Not too heavy, steams well, and delivered on time. Good price
to value ratio.” (for an iron).3 The aﬀect-rich single review was selected
Study 1 involved 154 younger adults who were students at PJAIT.
to be of high emotional intensity and of extreme valence (i.e. highly
Their average age was 20.8 years (SD = 2.3) and 140 of them were
positive in the positive single review condition and highly negative in
male. Study 2 involved 165 older adults who were recruited via a
the negative single review condition). They contained vivid and de-
LivingLab project run by PJAIT and focused on older adults [58–60].
tailed descriptions of positive/negative experiences the consumer had
Older adults' average age was 69 years (SD = 6.8, range: 58–87 years)
made with the product to facilitate putting one self in the position of the
and most of them were female (109 participants). Similar to the student
person writing the review (see Fig. 1). Aﬀect-rich positive and negative
group, the vast majority of older adults (157 participants) had at least
reviews were selected to be of similar length, aﬀective intensity, and secondary education.
detail. Neither the single aﬀect-rich reviews nor the baseline reviews
As a compensation for taking part in the study, older participants
contained an explicit star-rating.
received a pen drive (a USB ﬂash drive) with additional materials re-
In the single positive review condition, the aﬀect-rich review was
lated to the LivingLab and younger participants (students) received
presented with the lower rated product and the baseline review with
extra credit points. On average, it took younger adults 8 min and older
the higher rated product. In the negative review condition, the aﬀect-
adults 19 min to complete the study. Participants were randomly as-
rich review was presented with the higher rated product and the
signed to one of the six conditions.
baseline review with the lower rated product. 4.2. Materials 4.2.4. Presentation type
The two products were presented simultaneously or sequentially. In 4.2.1. Product cards
the ﬁrst case, the two product cards were shown on the same screen,
Participants made choices for three types of products: Vacuum
one next to the other. In the second case, they were presented on se-
cleaners, irons, and drills. Product types were selected to ensure that
parate screens. After seeing the ﬁrst option, participants had to click to
most participants would have some but not too much knowledge about
move on to the next screen to see the second option. Participants were
them. Each product card contained a product photo and its four attri-
not allowed to go back. In both, the simultaneous and the sequential butes including price.
condition, the decision itself was made on a yet separate screen that was
For vacuum cleaners and irons, the attributes' values were chosen so
presented after the product cards.
that it was unclear which of the two products was the better choice
because each of them was superior in at least one attribute. For drills,
one drill in the pair clearly dominated the other option because it had
4.2.5. Ratings of product attributes, consumer ratings and reviews
better values on three attributes (it was faster, cheaper, and worked
In addition to participants' choices we measured how they perceived
longer on a battery) and similar values for the fourth attribute (it was
the presented information. Each choice was followed by a short survey slightly heavier).
asking the subjects to rate the importance of the product attributes, the
All product descriptions can be found in the online supplementary
average consumer rating, and the single consumer review (if applic-
material and on the Open Science Framework (OSF, folder materials).2
able) for the decision that they had just made. In addition, they rated
the diﬃculty of the decision and their knowledge about the product
4.2.2. Average consumer ratings
type (i.e., vacuum cleaners, irons, and drills). All ratings were made on
For each product the average consumer rating was presented as a
7-point Likert scales ranging from (1) not at all to (7) very much.
number of ﬁlled-in stars from a total of 5 stars, similar to the way
consumer ratings are presented on websites of online retailers, see
Fig. 1. All average ratings were positive (e.g between 3.9 and 4.7 stars)
3 We chose these statements as a comparison for the vivid emotional reviews
but in each product pair one of the products was rated between 0.5 and
over completely neutral statements because they better reﬂect typical short
reviews that are found on online retailer websites and which are in general
positive [4,61,62]. Thus they provide a realistic baseline to which reviews 2 https://osf.io/3n8xw/. could be compared. 4 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10 4.3. Procedure
with the lower rating. In contrast, if participants preferred products
with higher average ratings, the higher rated product should be chosen
After signing a consent form, participants were asked to provide
more frequently. Thus, in a ﬁrst step we tested whether the probability
basic demographic characteristics (gender, education and age) and to
with which the product with the higher average rating was chosen
rate their experience with online shopping on a scale from (1) not at all
diﬀered from 0.5 in the no single review condition.
experienced to (7) very much experienced. Afterwards, they were in-
Overall, younger participants in Study 1 strongly followed the
formed about the study and the consumer decisions they would make.
average ratings. As illustrated in Fig. 2 (left panel), when no consumer
Before each decision participants received information regarding pro-
review was shown to participants, they chose the higher rated product
duct attributes and why they may be important while choosing between
in 80% of the cases. The choice proportions diﬀered signiﬁcantly from
the products. Then the two products were presented to the participants
0.5 for all three products (vacuum cleaner: χ 2(N = 46) = 8.10,
and they had to indicate their choice. After the choice was made, par- p = 0.004; iron: χ 2(N = 46) = 11.13, p < 0.001; drill:
ticipants responded to the survey about the decision process and then
χ 2(N = 45) = 4.61, p = 0.032).
continued with the next decision. At the end of the study, participants
In contrast, for older adults we did not ﬁnd an inﬂuence of average
read all the consumer reviews used in the study (just the texts) and
ratings on choices (see Fig. 2, right panel). Participants chose the higher
rated their valence and understandability on a scale from (1) very ne-
rated product in 58% of the choices when no review was provided. This
gative/do not understand at all to (7) very positive/understand very
did not signiﬁcantly diﬀer from 50% when considering all choices to-
much respectively. The latter questionnaire was added only later for the
gether, χ 2(N = 163) = 1.63, p = 0.20, nor for any of the three products
older adults, thus information from 41 people is missing. After the
separately (in all three cases p > 0.38). This, as can be seen below, does
study, participants received their reimbursement.
not mean that their choices were random.
To test for the inﬂuence of product attributes on choices, we ran multilevel mixed e 5. Results
ﬀects logistic regressions with random intercepts for
subjects with choice of the higher rated product as a dependent variable
and product quality (z-transformed), presentation type (0 - simulta-
In the following, we ﬁrst analyze whether product attributes, their
neous, 1 - sequential), and order (1 - First/Left, 0 - Second/Right) and
presentation and average consumer ratings inﬂuenced the choices.
After that we examine how single aﬀect-rich positive and negative
their two-way interactions as independent variables.6 All models were
implemented in R using the mixed function in the afex package [63]
consumer reviews changed the frequency with which the product with
higher average rating was chosen and how participants perceived the
using Likelihood Ratio Tests. Post hoc contrasts were calculated with the lsmeans package [64].
single reviews. Further (exploratory) analyses investigating partici-
pants' ratings are reported in the supplementary online material and on
For younger adults we found a strong eﬀect of product quality on
their choices, b = 1.58, SE = 0.459, χ 2(1) = 26.92, p < 0.001. In ad-
the OSF (folder Results).4 To facilitate the comparison between the age dition, we found a signi
groups, we report the results from Study 1 and Study 2 side by side in
ﬁcant main eﬀect of order, b = 0.733,
SE = 0.342, χ 2(1) = 5.85, p = 0.016, and an interaction of order by each section.5
product quality, χ 2(1) = 5.07, p = 0.024, but no eﬀect of presentation
type. Follow-up tests of the eﬀect of product quality separately for the
5.1. Inﬂuence of average ratings and product attributes on choices
two order conditions showed a large eﬀect of product quality when the
higher rated product was second/on the right side, b = 2.04, SE = 0.54,
Our ﬁrst research questions focused on whether younger and older
χ 2(1) = 21.27, p < 0.001. When the higher rated product was ﬁrst/on
adults used average consumer ratings in their decisions and if older
the left side, the eﬀect or product quality was smaller, but still sig-
adults were able to reliably choose products with better attributes.
niﬁcant, b = 0.755, SE = 0.369, χ 2(1) = 5.75, p = 0.0165 (see also
In order to test to what degree participants considered product at-
Fig. 2 and Fig. 3 in the supplementary online material).7 No other eﬀect
tributes, we constructed an index of product quality that indicated how
or interaction was signiﬁcant.
much better the attributes of one product were in comparison to the
For older adults, the product quality index also emerged as a sig-
other product in the product pair. To this end, we ﬁrst calculated for
niﬁcant predictor of choice (b = 0.381, SE = 0.186, χ 2(1) = 4.76,
each product attribute the percentage by which the product with higher
p = 0.029). Increasing product quality from visibly lower quality
average rating was superior/worse than the product with lower rating
(quality index equal to −30) to comparable quality (quality index
and then averaged across product attributes. A low absolute value in-
equal 0) and from comparable quality to visibly better quality (quality
dicates that the two products are of similar quality and that making a
index equal to 30) both led to an increase in predicted probability of
choice required a trade-oﬀ between the products' attributes. In contrast,
choosing the higher-rated product by roughly 15% (average marginal
a high absolute value indicates that one product is clearly superior to
eﬀect for the ﬁxed part of the model). This shows that older participants
the other product and no trade-oﬀs are necessary. Although this index
paid attention to the product attributes and were more likely to choose
can not account for subjective diﬀerences in the importance of the at-
a product that was clearly better on the attribute dimensions. Thus it
tributes, it provides a useful index of how clearly one product in the
indicates that they were not choosing randomly (see Fig. 2).
pair was better than the other product. As designed, for vacuum clea-
In addition, we found an eﬀect of presentation order for older adults
ners and irons, the two products did not diﬀer much in terms of quality
(b = −0.411, SE = 0.173, χ 2(1) = 5.94, p = 0.015), suggesting that
(i.e., ± 0.5), whereas for drills one product was clearly superior to the other (i.e., ± 29.3).
In the studies, each product in a pair was presented equally often
6 We excluded the three-way interaction of presentation type × product
with a higher and a lower average consumer rating. Accordingly, if
quality × order from the analyses because the sample sizes within each cell
participants did not consider average ratings in their choices, the pro-
were not suﬃcient to estimate the models. In addition, we had to exclude the
duct with the higher rating should be chosen as often as the product
interaction of presentation type with order in the analysis for the younger
adults because the model did not converge. We did not include product type in
this model because the model became unstable when product type and the 4 https://osf.io/3n8xw/.
product quality index were both included. Analyses without the quality index
5 We abstain from reporting statistical comparisons between age groups to
indicated that choices did not diﬀer signiﬁcantly between products.
focus on the impact of the manipulated variables. However, we report addi-
7 Tables reporting the choice proportions of older and younger adults by
tional analyses with age groups as a factor in the supplementary online mate-
product and product quality can be found in the supplementary online mate- rials and on OSF.
rials and on OSF (folder Results): https://osf.io/3n8xw. 5 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10 Younger Adults Older Adults no single review 1 1 single positive review single negative review 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 no single review single positive review single negative review
rop. of part. choosing the higher rated product
rop. of part. choosing the higher rated product P 0 P 0 -29.3 -0.5 +0.5 +29.3 -29.3 -0.5 +0.5 +29.3 Product quality index Product quality index
Fig. 2. Proportion of participants choosing the higher rated option by review condition and product quality for young adults (left panel) and older adults (right
panel). The product quality index indicates the quality of the higher rated option. Error bars denote 1 SE.
older adults were more likely to choose the option with the higher
aﬀect-rich, positive review reduced the odds that the option with higher
rating when it was presented ﬁrst or on the left-hand side of the screen.
average rating would be chosen by a factor of 3.56 (b = 1.271,
No further signiﬁcant eﬀects were found.
SE = 0.448, z = 2.84, p = 0.013). The inﬂuence of single positive and
In sum, both younger and older adults were more likely to chose the
negative consumer reviews on choices did not diﬀer signiﬁcantly,
better rated product when it also had better attributes. Yet, the inﬂu-
z = 1.228, p = 0.437 (see Fig. 2, left panel).
ence of product quality on the choices of older adults was less pro-
For older adults, the analyses also showed signiﬁcant main eﬀects of
nounced than in the case of younger adults. This can clearly be seen review condition, χ 2(2) = 30.44, p < 0.001, product quality,
when focusing on the drills, for which by design one option was clearly
b = 0.448, SE = 0.110, χ2(1) = 18.44, p < 0.001, and presentation
better than the other one in terms of the product's attributes. When the
order, b = −0.204, SE = 0.101, χ 2(1) = 4.11, p = 0.0427, but no sig-
better drill was also recommended by average consumer ratings, (i.e., it
niﬁcant interactions. Post-hoc contrasts adjusted with the Tukey
dominated the other product on all relevant dimensions), younger
method, indicated that a single, negative, aﬀect-rich review sig-
adults chose the better drill in 100% of the cases, showing that they
niﬁcantly reduced the probability of choosing the higher rated product
clearly recognized this dominance relationship. In contrast, older adults
by a factor of 3.5 compared to the no review condition, b = 1.256,
chose the better drill in 76% of the cases.
SE = 0.260, z = 4.83, p < 0.001. However, in contrast to the younger
adults, a single, aﬀect-rich, positive review of the lower rated product
did not change the likelihood of choosing the higher rated product
5.2. Inﬂuence of single aﬀect-rich consumer reviews on choice
compared to the no review condition for older adults, b = 0.094,
SE = 0.242, z = 0.387, p = 0.921 (see Fig. 2, right panel).
In the next step, we investigated our second research question, i.e.
whether the single aﬀect-rich consumer reviews inﬂuenced how fre-
quently the higher rated product was chosen (see Fig. 2). To this goal
5.3. Perception of the aﬀect-rich reviews
we once again ran a multilevel mixed eﬀects logistic model predicting
how often the higher rated product was chosen, now analyzing the full
The analyses of choices had shown that for younger adults a single
data set. In addition to the predictors in the analyses above, we in-
aﬀect-rich review that was inconsistent with the average rating reliably
cluded review condition and its interactions with the other predictors in
reduced how often the option with higher rating was chosen. This was the model.8
the case for positive and negative aﬀect-rich reviews. For older adults,
The analysis for younger adults revealed a signiﬁcant eﬀect of re-
however, we only found an inﬂuence for negative aﬀect-rich reviews.
view condition on choices, χ 2(2) = 16.79, p < 0.001 and an eﬀect of
One reason could be that older and younger adults diﬀered in how well
product quality, b = 0.978, SE = 0.170, χ 2(1) = 48.07, p < 0.001. No
they understood the aﬀect-rich reviews and as how positively/nega-
other main eﬀect or interactions reached signiﬁcance. Post-hoc con-
tively they perceived them compared to the baseline review presented
trasts using the Tukey method to adjust p-values indicated that the odds
with the other product. To investigate this question we tested whether
of students choosing the higher-rated product were 5.43 times smaller
participants reported that they understood the reviews and whether
when it was presented with a single, aﬀect-rich, negative review than
participants really perceived the aﬀect-rich reviews as more positive/
when no reviews were included (b = 1.691, SE = 0.451, z = 3.75,
negative than the baseline review.
p < 0.001). Similarly, presenting the lower rated product with a single,
Overall, younger and older adults reported high levels of under-
standing of the consumer reviews. Average ratings were above or close
to 5 on a seven-point scale, suggesting that both age groups understood
8 We did not include the four-way interaction including all predictors in the
the reviews well (see Table 1 for an overview of participants' ratings).
analyses because the within-cell sample sizes were not suﬃcient to estimate the
To test whether the review type (a
models. For younger adults we also had to exclude the three-way interactions of
ﬀect-rich positive, aﬀect-rich presentation type × product quality × order and review condi-
negative, and baseline reviews) diﬀered in how their valence was rated
tion × order × product quality because otherwise the model did not converge.
(i.e. how positive/negative they were perceived) we run a repeated
Reducing the model further does not change the conclusions.
measurement analyses of variance (ANOVA) with the valence rating as 6 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10 Table 1
for most of them and were not overruled when they were clearly in
Overview of rating responses in both studies. Rating scale = 1 (not at all) to 7
conﬂict with the consumer ratings. Secondly, it shows that older adults
(very much). Since each participant made multiple evaluations (after each
understood the task, but struggled more with identifying a better pro-
choice and for each review) we use robust standard errors clustered on parti-
duct just from the attributes. This is most clearly seen when considering
cipants. The question regarding the importance of the single consumer review
choices for the drills. When in the no-review condition the better drill
was only asked in the conditions including a consumer review (the positive and
was also recommended by average consumer ratings, (i.e., it dominated
negative review conditions). For the ratings of the valence and understanding of
the other product in all relevant dimensions), older adults chose the
the aﬀect-rich consumer reviews data from N = 41 older participants is
better drill in 76% of the cases. This demonstrates that older adults missing.
clearly paid attention to the product attributes and were not choosing Young adults Older adults
randomly. But it also suggests that older adults had problems identi-
fying a better product based on its attributes. These results resonate Variables Mean SE Mean SE
with research showing declines in decision-making abilities in new and Online experience 5.29 0.11 1.96 0.12
complex tasks in older adults [31-33]. The results are also in line with Product knowledge 3.19 0.10 3.57 0.10
research on consumer choice ﬁnding choice deﬁcits in older adults Importance: product features 5.70 0.08 5.37 0.11 (e.g., [38]). Importance: average rating 4.90 0.11 4.09 0.12 Importance: consumer review 3.51 0.15 4.00 0.15
Understandings: negative review 5.54 0.09 5.10 0.16
6.2. Inﬂuence of average consumer ratings
Understandings: baseline review 4.93 0.12 5.19 0.14
Understandings: positive review 5.53 0.09 5.68 0.12
Younger adults were strongly inﬂuenced by aggregated consumer Valence: negative review 1.76 0.10 2.52 0.16
ratings. In particular, when the two products they could choose from Valence: baseline review 5.46 0.09 5.19 0.13 Valence: positive review 6.20 0.08 5.66 0.13
were similar in quality and no single positive or negative review was
presented, younger adults overwhelmingly chose the higher rated
product. The majority of younger adults chose the lower rated product
dependent variable and the type of the review as predictor.
only when it had clearly better attributes than the higher rated product,
For younger adults we found a large eﬀect of the type of review on
although even then a sizable minority (43%) still preferred the higher
valence ratings (F(2,922) = 1235, p < 0.001, η2= 0.73). Contrasts
rated product. These results dovetail with previous research reporting
conﬁrmed that the positive reviews were perceived as more positive
the importance of consumer ratings for online purchasing decisions of
than the baseline reviews (F(1,461) = 134, p < 0.001, η2 = 0.23) and
younger adults [5–7] and also resonate with younger adults reporting
the negative reviews as more negative than the baseline reviews (F
average ratings as quite important, with a score of 4.9 on a scale from 1
(1,461) = 1136, p < 0.001, η2 = 0.71). Accordingly, participants per- to 7.
ceived the reviews as diﬀerently positive/negative but the diﬀerence in
In contrast, we did not ﬁnd any evidence that older adults con-
perception was larger for the negative-baseline comparison than the
sidered average consumer ratings in their decisions. This ﬁnding is positive-baseline comparison.
surprising given the prevalence of consumer reports in an online con-
Similar to the younger adults, how positively older adults perceived
text and high importance assigned to them by younger adults, but it
the reviews depended strongly on the type of the consumer review (F
corresponds to ﬁndings by Ma et al. [3] suggesting that older adults
(2,736) = 349.5, p < 0.001, η2 = 0.49). They perceived the positive
perceive consumer ratings as less relevant and helpful than younger
consumer reviews as more positive than the baseline reviews (F adults.
(1,369) = 21.79, p < 0.001, η2 = 0.06) and the negative reviews as
Why older adults did not use average ratings is less clear. One
more negative than the baseline reviews (F(1,368) = 373.5, p < 0.001,
reason could be that older adults just do not value the opinion of other 2
η = 0.50). However, overall and in particular for the comparison be-
consumers as much as younger adults, which suggests that they may not
tween positive and baseline reviews, the eﬀect size was much smaller
be aware of how valuable this information can be. Alternatively, the
than for younger adults. The same pattern of results was found for all
presentation of the consumer ratings may be confusing for older adults.
products, although for the drills the rating of the positive review did not
Older adults have worse visual acuity making it more diﬃcult to dis-
diﬀer signiﬁcantly from the rating of the baseline review (p = 0.19).
cern small diﬀerences on the screen [27,65]. Most consumer ratings
tend to be positive, making diﬀerences in average ratings relatively 6. Discussion
small. Thus it is possible that older adults had problems in realizing that
a diﬀerence in average ratings of 0.5 points carries relevant informa-
We found that younger and older adults were inﬂuenced by product
tion. Future research into this is necessary to determine the factors
attributes and aﬀect-rich negative reviews. However, whereas younger
underlying older adults' neglect of consumer ratings and potential ways
adults strongly relied on average consumer ratings and also on aﬀect-
of making this information more accessible to them.
rich positive reviews, older adults did not take them into account. These
results suggests that older adults diﬀer in how they perceive the reviews
6.3. Inﬂuence of single aﬀect-rich reviews
written by other consumers and how much value they assign to this
information, which has important implications for marketing directed
Both younger and older adults' choices were strongly inﬂuenced if
at older adults. In the following we discuss the results in more details
the higher-rated product was accompanied by a single, vivid, negative
and outline implications for designing e-commerce platforms for older
review. Only a minority of participants picked the option with the ne- adults.
gative review, even though they were told that the review was selected
randomly (and thus was not necessarily representative of the reviews
6.1. Inﬂuence of product attributes
the product had received). These results correspond with studies re-
porting that people are more easily inﬂuenced by anecdotal or narrative
For both younger and older participants, the quality index of pro-
information than by statistical information when making decisions in
ducts' attributes strongly inﬂuenced choices when no review was pre-
medical or consumer contexts [18–20].
sented, but also when single reviews were provided. This may not be
For younger adults aﬀect-rich negative and positive single reviews
very surprising in itself, but the results are important for two reasons.
that conﬂicted with the average ratings reduced how often the higher-
First of all, it shows that although average consumer ratings are quite
rated product was chosen. The eﬀect size was somewhat larger for the
important for younger adults, product attributes were more important
aﬀect-rich negative than for the aﬀect-rich positive reviews, but the 7 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
diﬀerence was not signiﬁcant. At ﬁrst sight, the somewhat larger eﬀect
adapted accordingly (better visibility of negative reviews and less stress
of negative reviews is consistent with a negativity bias. However, when
on average ratings for older adults).
evaluating the eﬀect of the aﬀect-rich reviews one must also take into
The observed diﬀerences between younger and older adults, how-
account how the aﬀect-rich reviews were perceived compared to the
ever, have more far-reaching consequences than just regarding the
baseline review shown with the other option. Due to the nature of the
personalization of the way ratings and reviews are displayed and force
baseline reviews we chose — short but in general positive statements
researchers to re-examine existing methods of mitigating attacks on e-
selected to reﬂect typical reviews found online – the aﬀect-rich negative
commerce ratings systems. To assure the same level of protection for
reviews diﬀered more strongly in their perceived valence from the
both groups of users ratings, systems should be resistant not only to
baseline reviews than the aﬀect-rich positive reviews. Nevertheless,
attacks on average ratings (malicious increasing or decreasing) but also
younger adults' choices were inﬂuenced by the aﬀect-rich positive re-
to the injection of single, fake, vivid, negative reviews.
views. Indeed, the positive aﬀect-rich reviews aﬀected their choices in a
similar way as the negative aﬀect-rich reviews in spite of the smaller 6.5. Limitations
diﬀerence in perceived valence. A ﬁnding that is in contradiction to
research showing a negativity bias [11,13]. However, it is in line with
In this study we found strong age diﬀerences in the inﬂuence of
research suggesting that negativity bias may be limited to hedonic
aggregated and single consumer reviews on choices. However, it is
goods [12] as the products we used were more of a utilitarian nature. In
important to take into account several limitations of our study.
addition, although we can’t distinguish between the features that made
For one, we used a cross-sectional design. Thus, it is impossible to
the aﬀect-rich reviews more convincing than the baseline reviews in the
separate age diﬀerences from cohort eﬀects. Although we did not ﬁnd
current study, it suggests that beside valence other features such as the
any evidence that experience with online shopping inﬂuenced older
aﬀective nature and the level of detail of a review may aﬀect how much
participants' choices (for details see supplementary online materials), it
weight participants give them in their decisions.
is possible that current older adults are just less accustomed to con-
In contrast, older adults were strongly inﬂuenced by the aﬀect-rich
sumer ratings and reviews in general, and thus give this information
negative reviews, but not at all by the aﬀect-rich positive reviews. At
less weight. However, this may change when a new generation of more
ﬁrst glance, these results seem to be at odds with research proposing a
internet-savvy individuals approaches old age.
focus on positive information in older adults [49,50]. However, ac-
In addition, our two samples are from a single country and diﬀered
cording to socio-emotional selectivity theory older adults focus on po-
not only in age but also in the gender composition, with younger adults
sitive information stems from a shift to emotional meaningful
being mostly male and older adults mostly female. Gender partly in-
goals [49]. With emotional goals in mind, avoiding products with ne-
ﬂuenced product knowledge, but otherwise we did not ﬁnd an eﬀect of
gative reviews, and thus potential losses, seems a rational strategy as it
gender on choices, suggesting that the results are not dependent on
minimizes negative emotions that could be caused by choosing the
gender (for details see supplementary online materials). In addition,
wrong product [51]. This suggests that in decision making tasks older
although we did not ﬁnd diﬀerence on education on choices, the
adults may be more likely to exhibit a negativity than a positivity bias.
younger adults sample consisted of students, whereas older adults were
Why older adults did not consider the aﬀect-rich positive reviews at
recruited from the community. In sum, we cannot exclude that sample
all is an interesting question. One possibility is that older adults did not
diﬀerences could have played some role and that generalizability may
diﬀerentiate between the aﬀect-rich positive and the baseline reviews be limited.
as much as younger adults did. Although older adults on average rated
In the studies we used a set of utilitarian products with hypothetical
the aﬀect-rich positive reviews as signiﬁcantly more positive than the
choices. Products were selected to be typical domestic equipment that
baseline reviews, the diﬀerence was smaller than for the younger adults
most people own at one point in their life, but buy only rarely, to de-
and the diﬀerence between the aﬀect-rich negative reviews and the
crease diﬀerences in product knowledge between participants.
baseline reviews. Accordingly, it is possible that for older adults the
However, decision processes may depend on the type of products se-
perceived diﬀerences in valence was not strong enough to aﬀect their
lected. For instance, the negativity bias has been shown to be limited to
choices. This suggests that, although in old age the understanding of
hedonic goods [12]. Thus, it is possible that older adults may take
emotions in written texts and of narrative texts seems to be largely
average consumer ratings more into account while making real choices
unaﬀected [45,46], older adults may still have more diﬃculty in per-
or while choosing between other types of products such as experiential
ceiving nuances in emotional intensity [66]. services.
To measure the inﬂuence of product attributes on choice, we used
6.4. Implications for a design of rating systems
an index of relative product quality that assumed all attributes to be
equally important. Given that subjective importance of attributes will
We found clear indications that older adults were strongly inﬂu-
diﬀer between individuals, our measure of product quality most likely
enced by aﬀect-rich negative consumer reviews, but not by better
underestimates the inﬂuence of product attributes on choices.
average consumer ratings or aﬀect-rich positive reviews. This suggests
Lastly, the consumer reviews we used as a baseline comparison were
that social media and WOM communication directed at older adults do
not at the (valence) midpoint between negative and positive reviews
not require enthusiastic and vivid descriptions whereas younger adults
but had a positive valence. We chose these short but positive statements
can be convinced by strongly positive reviews.
as baseline reviews because they are more typical than neutral reviews
The ﬁnding that older adults did not consider average ratings in our
[4,61,62] but still diﬀer in their valence from the strongly positive re-
study is intriguing. The strong eﬀect of the negative reviews suggests views.
that they are not completely insensitive to consumer recommendations.
Alternatively, it is possible that the diﬀerences in ratings were too small 7. Conclusions
to carry meaning for older adults. Here, it would be important to choose
designs that make it easier for older adults to recognize diﬀerences in
Our results show not only that ratings and reviews play diﬀerent
ratings that otherwise they may not be able to discriminate.
roles in purchasing decisions (see [67]), but also that the importance of
Our results show not only that ratings and reviews play diﬀerent
reviews and ratings varies between younger and older adults. Whereas
roles in the purchasing decision process (see [67]), but also that the
students were strongly inﬂuenced by average consumer ratings and
importance of reviews and ratings varies between younger and older
positive aﬀect-rich reviews, the older adults in our sample gave little
adults. Therefore, to ensure the same level of comfort for both groups
importance to these types of consumer information. However, younger
while making decisions about purchases, the user interface should be
and older adults were quite strongly inﬂuenced by aﬀect-rich negative 8 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
reviews — even if these were unrepresentative of the product reviews.
Medicine 67 (12) (2008) 2079–2088.
These results highlight important age diﬀerence in consumer behavior,
[23] P.B. Baltes, U.M. Staudinger, U. Lindenberger, Lifespan psychology: theory and
application to intellectual functioning. Annual Review of Psychology 50 (1999)
raising questions about the utility of consumer reviews for older adults,
471–507, https://doi.org/10.1146/annurev.psych.50.1.471.
as well as how consumer reviews should be presented. To ensure the
[24] L.L. Carstensen, The inﬂuence of a sense of time on human development, Science
same level of comfort for both groups when making decisions about 312 (5782) (2006) 1913–1915.
[25] E. Peters, T.M. Hess, D. Västfjäll, C. Auman, Adult age diﬀerences in dual in-
purchases, at the very least the user interface must be adapted ac-
formation processes: implications for the role of aﬀective and deliberative processes
cordingly (better visibility of negative reviews and less stress on
in older adults' decision making, Perspectives on Psychological Science 2 (1) (2007)
average ratings for older adults). 1–23.
[26] R. Mata, T. Pachur, B. von Helversen, R. Hertwig, J. Rieskamp, L. Schooler,
Ecological rationality: a framework for understanding and aiding the aging decision Acknowledgements
maker, Frontiers in Decision Neuroscience 6 (Article 19) (2012) 1–6, https://doi. org/10.3389/fnins.2012.00019.
The work was supported by a grant of the Swiss National Science
[27] W.A. Rogers, A.J. Stronge, A.D. Fisk, Technology and Aging, Reviews of human
factors and ergonomics 1 (1) (2005) 130–171.
Foundation to the ﬁrst author [No. 157432]. This work was also par-
[28] T.A. Salthouse, Mental exercise and mental aging: evaluating the validity of the “use
tially supported by European Union's Horizon 2020 Research and
it or lose it” hypothesis, Perspectives on Psychological Science 1 (1) (2006) 68–87.
Innovation Programme under the Marie Skłodowska-Curie grant
[29] T. Salthouse, Consequences of age-related cognitive declines, Annual Review of
Psychology 63 (2012) 201–226.
agreement [No. 690962]. Declarations of interest: none.
[30] M.L. Finucane, C.K. Mertz, P. Slovic, E.S. Schmidt, Task complexity and older
adults' decision-making competence, Psychology and Aging 20 (1) (2005) 71–84.
Appendix A. Supplementary materials: analyses and stimulus
[31] R. Frey, R. Mata, R. Hertwig, The role of cognitive abilities in decisions from ex-
perience: age diﬀerences emerge as a function of choice set size, Cognition 142 materials
(2015) 60–80, https://doi.org/10.1016/j.cognition.2015.05.004.
[32] R. Mata, L. Nunes, When less is enough: cognitive aging, information search, and
Supplementary data to this article can be found online at https://
decision quality in consumer choice, Psychology and Aging 25 (2010) 289 2 – 98,
doi.org/10.1016/j.dss.2018.05.006.
https://doi.org/10.1037/a0017927.
[33] R. Mata, L.J. Schooler, J. Rieskamp, The aging decision maker: cognitive aging and
the adaptive selection of decision strategies, Psychology & Aging 22 (2007) References 101037/0882–7974224796.
[34] B. von Helversen, R. Mata, Losing a dime with a satisﬁed mind: positive aﬀect
predicts less search in sequential decision making, Psychology and Aging 27 (4)
[1] J.W. Lian, D.C. Yen, Online shopping drivers and barriers for older adults: age and
(2012) 825–839, https://doi.org/10.1037/a0027845.
gender diﬀerences, Computers in Human Behavior 37 (2014) 133–143.
[35] G. Gigerenzer, P.M. Toddthe ABC Research Group, Simple Heuristics That Make Us
[2] M. Law, M. Ng, Age and gender diﬀerences: understanding mature online users with
Smart, Oxford University Press, 1999.
the online purchase intention model, Journal of Global Scholars of Marketing
[36] R. Mata, B. von Helversen, J. Rieskamp, Learning to choose: cognitive aging and
Science 26 (3) (2016) 248–269.
strategy selection learning in decision making, Psychology and Aging 25 (2) (2010)
[3] Y.J. Ma, H. Kim, H.-h. Lee, Eﬀect of individual diﬀerences on online review per-
299–309, https://doi.org/10.1037/a0018923.
ception and usage behavior: the need for cognitive closure and demographics,
[37] J.A. Mikels, C.E. Löckenhoﬀ, S.J. Maglio, L.L. Carstensen, M.K. Goldstein,
Journal of the Korean Society of Clothing and Textiles 36 (12) (2012) 1270–1284.
A. Garber, Following your heart or your head: focusing on emotions versus in-
[4] P.Y. Chen, S.y. Wu, J. Yoon, The impact of online recommendations and consumer
formation diﬀerentially inﬂuences the decisions of younger and older adults,
feedback on sales, ICIS 2004 Proceedings, 2004, p. 58.
Journal of Experimental Psychology: Applied 16 (1) (2010) 87.
[5] K. Floyd, R. Freling, S. Alhoqail, H.Y. Cho, T. Freling, How online product reviews
[38] C.A. Cole, S.K. Balasubramanian, Age diﬀerences in consumers' search for in-
aﬀect retail sales: a meta-analysis, Journal of Retailing 90 (2) (2014) 217–232,
formation: public policy implications, Journal of Consumer Research 20 (1993)
https://doi.org/10.1016/j.jretai.2014.04.004. 157 1 – 69.
[6] R.A. King, P. Racherla, V.D. Bush, What we know and don’t know about online
[39] C.M. Schaninger, D. Sciglimpaglia, The inﬂuence of cognitive personality traits and
word-of-mouth: a review and synthesis of the literature, Journal of Interactive
demographics on consumer information acquisition, Journal of Consumer Research
Marketing 28 (3) (2014) 167–183. 8 (2) (1981) 208–216.
[7] N. Purnawirawan, M. Eisend, P. De Pelsmacker, N. Dens, A meta-analytic in-
[40] R. Lambert-Pandraud, G. Laurent, E. Lapersonne, Repeat purchasing of new auto-
vestigation of the role of valence in online reviews, Journal of Interactive Marketing
mobiles by older consumers: empirical evidence and interpretations, Journal of
31 (2015) 17–27, https://doi.org/10.1016/j.intmar.2015.05.001.
Marketing 69 (2) (2005) 97–113.
[8] Y. Liu, Word of mouth for movies: its dynamics and impact on box oﬃce revenue,
[41] S.M. Carpenter, C. Yoon, Aging and consumer decision making, Annals of the New
Journal of Marketing 70 (3) (2006) 74–89.
York Academy of Sciences 1235 (1) (2011) 1–12.
[9] E. Maslowska, E.C. Malthouse, V. Viswanathan, Do customer reviews drive pur-
[42] Q. Ma, K. Chen, A.H.S. Chan, P.L. Teh, Acceptance of ICTs by older adults: a review
chase decisions? The moderating roles of review exposure and price, Decision
of recent studies, International Conference on Human Aspects of IT for the Aged
Support Systems 98 (2017) 1–9.
Population, Springer, 2015, pp. 239–249.
[10] S. Karimi, F. Wang, Online review helpfulness: impact of reviewer proﬁle image,
[43] A. Ahmed, A.S. Sathish, Determinants of online shopping adoption: meta analysis
Decision Support Systems 96 (2017) 39–48.
and review, European Journal of Social Sciences 49 (4) (2015) 483–510.
[11] J. Lee, D.h. Park, I. Han, The eﬀect of negative online consumer reviews on product
[44] G. Cohen, Language comprehension in old age, Cognitive Psychology 11 (4) (1979)
attitude: an information processing view, Electronic Commerce Research and 412 4 – 29.
Applications 7 (2008) 341–352.
[45] R. De Beni, E. Borella, B. Carretti, Reading comprehension in aging: the role of
[12] S. Sen, D. Lerman, Why are you telling me this? An examination into negative
working memory and metacomprehension, Aging, Neuropsychology, and Cognition
consumer reviews on the web, Journal of Interactive Marketing 21 (4) (2007)
14 (2) (2007) 189–212, https://doi.org/10.1080/13825580500229213. 76 9 – 4.
[46] L.H. Phillips, R.D. MacLean, R. Allen, Age and the understanding of emotions
[13] C. Betsch, N. Haase, F. Renkewitz, P. Schmid, The narrative bias revisited: what
neuropsychological and sociocognitive perspectives, The Journals of Gerontology.
drives the biasing inﬂuence of narrative information on risk perceptions? Judgment
Series B, Psychological Sciences and Social Sciences 57 (6) (2002) 526–P530.
and Decision Making 10 (3) (2015) 241–264.
[47] L.L. Carstensen, Motivation for social contact across the life span: a theory of so-
[14] P. Rozin, E.B. Royzman, Negativity bias, negativity dominance, and contagion,
cioemotional selectivity, Nebraska symposium on motivation, vol. 40, 1993, pp.
Personality and Social Psychology Review 5 (4) (2001) 296–320. 209 2 – 54.
[15] P.F. Wu, In search of negativity bias: an empirical study of perceived helpfulness of
[48] S.T. Charles, L.L. Carstensen, Social and emotional aging, Annual Review of
online reviews, Psychology & Marketing 30 (11) (2013) 971–984.
Psychology 61 (2010) 383–409.
[16] BrightLocal, Local Consumer Review Survey 2016, (2016) https://www.brightlocal.
[49] A.E. Reed, L. Chan, J.A. Mikels, Meta-analysis of the age-related positivity eﬀect:
com/learn/local-consumer-review-survey/.
age diﬀerences in preferences for positive over negative information, Psychology
[17] S. Hong, H.S. Park, Computer-mediated persuasion in online reviews: statistical
and Aging 29 (1) (2014) 1–15.
versus narrative evidence, Computers in Human Behavior 28 (3) (2012) 906–919.
[50] H.H. Fung, L.L. Carstensen, Sending memorable messages to the old: age diﬀerences
[18] M. Ziegele, M. Weber, Example, please! Comparing the eﬀects of single customer
in preferences and memory for advertisements, Journal of Personality and Social
reviews and aggregate review scores on online shoppers' product evaluations,
Psychology 85 (1) (2003) 163–178.
Journal of Consumer Behaviour 14 (2015) 103–114.
[51] M.K. Depping, A.M. Freund, Normal aging and decision making: the role of moti-
[19] C. Betsch, C. Ulshöfer, F. Renkewitz, T. Betsch, The inﬂuence of narrative v. sta-
vation, Human Development 54 (6) (2011) 349–367.
tistical information on perceiving vaccination risks, Medical Decision Making 31 (5)
[52] M.J. Frank, L. Kong, Learning to avoid in older age, Psychology and Aging 23 (2)
(2011) 742–753, https://doi.org/10.1177/0272989X11400419.
(2008) 392–398, https://doi.org/10.1037/0882-7974.23.2.392.
[20] P.A. Ubel, C. Jepson, J. Baron, The inclusion of patient testimonials in decision aids,
[53] D. Hämmerer, S.C. Li, V. Müller, U. Lindenberger, Life span diﬀerences in electro-
Medical Decision Making 21 (1) (2001) 60–68.
physiological correlates of monitoring gains and losses during probabilistic re-
[21] A. Winterbottom, H.L. Bekker, M. Conner, A. Mooney, Does narrative information
inforcement learning, Journal of Cognitive Neuroscience 23 (3) (2011) 579–592.
bias individual's decision making? A systematic review, Social Science and
[54] B. Eppinger, N.W. Schuck, L.E. Nystrom, J.D. Cohen, Reduced striatal responses to 9 B. von Helversen et al.
Decision Support Systems 113 (2018) 1–10
reward prediction errors in older compared with younger adults, Journal of
[67] N. Hu, N.S. Koh, S.K. Reddy, Ratings lead you to the product, reviews help you
Neuroscience 33 (24) (2013) 9905–9912.
clinch it? The mediating role of online review sentiments on product sales, Decision
[55] A. Mantonakis, P. Rodero, I. Lesschaeve, R. Hastie, Order in choice: eﬀects of serial
Support Systems 57 (2014) 42–53.
position on preferences, Psychological Science 20 (11) (2009) 1309–1312, https://
doi.org/10.1111/j.1467-9280.2009.02453.x.
Bettina von Helversen is an assistant professor for cognitive decision psychology at the
[56] C. Mogilner, B. Shiv, S.S. Iyengar, Eternal quest for the best: sequential (vs. si-
University of Zurich. She is interested in understanding and modelling how people make
multaneous) option presentation undermines choice commitment, Journal of
judgments and decisions. In her work she focuses on the diﬀerent cognitive strategies
Consumer Research 39 (6) (2013) 1300–1312, https://doi.org/10.1086/668534.
people use to solve these tasks and the factors that inﬂuence strategy selection such as
[57] A. Dieckmann, K. Dippold, Compensatory versus noncompensatory models for
task structure, memory, aﬀect or stress and how these change over the life span.
predicting consumer preferences, Judgment and Decision Making 4 (3) (2009) 200–213.
Katarzyna Abramczuk is an assistant professor for mathematical sociology at the
[58] W. Kopeć, K. Skorupska, A. Jaskulska, K. Abramczuk, R. Nielek, A. Wierzbicki,
LivingLab PJAIT: towards better urban participation of seniors, Proceedings of the
University of Warsaw. She is interested in the interplay between cognitive and social
International Conference on Web Intelligence, ACM, 2017, pp. 1085
processes. She designs studies and models to analyze how macro reality is built by micro –1092.
strategies and how individual choices are in
[59] W. Kopeć, B. Balcerzak, R. Nielek, G. Kowalik, A. Wierzbicki, F. Casati, Older adults
ﬂuenced by the macro environment. She pays
and hackathons: a qualitative study, Empirical Software Engineering (2017) 1
special attention to the role of ICT in this context. –36.
[60] W. Kopeć, K. Abramczuk, B. Balcerzak, M. Juźwin, K. Gniadzik, G. Kowalik,
R. Nielek, A location-based game for two generations: teaching mobile technology
Wiesław Kopeć is part of an international social informatics research team at Polish-
to the elderly with the support of young volunteers, eHealth 360, Springer, 2017,
Japanese Academy of Information Technology (PJAIT) in Poland, doing research on ICT pp. 84–91.
technologies dedicated to older adults. He combines high-level IT and business experience
[61] D. Godes, D. Mayzlin, Using online conversations to study word-of-mouth com-
with rich academic background. Recently his interests extend to social aspects of applying
munication, Marketing Science 23 (4) (2004) 545–560.
modern ICT with a special focus on HCI in the context of participatory design and ad-
[62] P. Resnick, R. Zeckhauser, J. Swanson, K. Lockwood, The value of reputation on
vanced methods like Complex Event Processing, Machine Learnig and Big Data for sup-
eBay: a controlled experiment, Experimental Economics 9 (2) (2006) 79–101.
porting educational and business value generation processes.
[63] H. Singmann, B. Bolker, J. Westfall, F. Aust, afex: Analysis of Factorial Experiments,
(2016) https://CRAN.R-project.org/package=afex r package version 0.16-1.
Radoslaw Nielek is a head of the research group working on ICT technologies dedicated
[64] R.V. Lenth, Least-squares means: the R package lsmeans, Journal of Statistical
to older adults and an assistant professor at Polish-Japanese Academy of Information
Software 69 (1) (2016) 1–33, https://doi.org/10.18637/jss.v069.i01.
Technology (PJAIT). He received his Ph.D. degree from PJAIT Warsaw, Poland and
[65] F. Schieber, Human factors and aging: identifying and compensating for age-related
Bachelor's Degree in Production Engineering and Management of Szczecin University of
deﬁcits in sensory and cognitive function, Impact of technology on successful aging,
Technology. His research interests include application of ICT technologies for improving 2003, pp. 42–84.
quality of life of older adults and limiting negative consequences of demographic changes.
[66] L.H. Phillips, R. Allen, Adult aging and the perceived intensity of emotions in faces
and stories, Aging Clinical and Experimental Research 16 (3) (2004) 190–199. 10

Inﬂuence of consumer reviews on online purchasing decisions in older andyounger adults - Tài liệu tham khảo | Đại học Hoa Sen

Thông tin :

Tài liệu liên quan:

Bạo lực học đường (RE) - Tài liệu tham khảo | Đại học Hoa Sen

Tiểu luận giữa kì môn Triết học Nhóm Triết - Tài liệu tham khảo | Đại học Hoa Sen

Triết học trong cuộc sống - Tài liệu tham khảo | Đại học Hoa Sen

Four Methods of Surgical Correction of Abomasal Displacement in t- Tài liệu tham khảo | Đại học Hoa Sen

Bài thu hoạch tthcm - bài thu hoạch đi bảo tàng - Tài liệu tham khảo | Đại học Hoa Sen