You are on page 1of 25

Egocentric Processing: The Advantages of

Person-Related Features in Consumers’


Product Decisions

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


LIAD WEISS

This article explores the possibility that product features may resonate differently
with different consumers based on how consumers classify the product in relation
to their selves. Prior research has shown that relating products to a consumer’s
self affects product memory, judgment, and choice. Here we identify a novel way
in which the self contextualizes consumers’ product decisions: egocentric proc-
essing. We introduce a theoretical distinction between two types of product fea-
tures based on relative applicability to people versus products: person-related
(e.g., toughness) and product-related (e.g., durability). Seven experiments dem-
onstrated that consumers use self-categorization cues, such as ownership or
brand, to classify products in relation to the category of self. Consumers then use
the category of self, to which person-related features neatly apply, to process in-
formation about in-self products. Person-related features thus gain three advan-
tages in consumer decisions about in-self (vs. out-self) products: greater consider-
ation, faster processing, and higher importance. We see these advantages
especially when (1) similar advantages are present in self-judgment, (2) consum-
ers are self-focused, and (3) the self-categorization cue is self-defining. Our find-
ings both open up new ways for marketers to increase the appeal of products for
specific consumer segments and demonstrate ways to identify and target these
segments.

Keywords: self-concept, categorization, information processing, product features,


ownership, brands

I am so excited to tell you all about iPhone XS . . . .


It is the most beautiful iPhone we have ever made . . .
It’s also tougher.

Liad Weiss (liad.weiss@wisc.edu) is an assistant professor of market- Philip W. Schiller, Senior Vice President of
ing, University of Wisconsin–Madison, 975 University Ave., Madison, Worldwide Marketing, Apple Event Keynote
WI 53706, USA. Please address correspondence to Liad Weiss. The au- (2018)
thor thanks Anat Keinan, Dan Bartels, Gita V. Johar, Joann Peck, Robin
Tanner, and the entire review team for their helpful comments and sugges- Once a year, in a carefully orchestrated event, Apple
tions; Stephen Spiller and Markus Brauer for their data analysis advice;
and Szu-Chi Huang for her help with validating the product-dependent/ reveals the features of the newest version of its flagship de-
product-independent construct. Support for this research was provided by vice, the iPhone. This event targets people who identify
the Office of the Vice Chancellor for Research and Graduate Education at closely with Apple and even see their iPhones as part of
the University of Wisconsin–Madison with funding from the Wisconsin their identity, such as tech-geeks, fans, and long-time
Alumni Research Foundation. Supplementary materials are included in
the web appendix accompanying the online version of this article. iPhone users (vs. people who do not feel connected to the
brand or own an iPhone). In seeking to maximize the
Editors: Darren W. Dahl, Amna Kirmani, and Andrew T. Stephen
phone’s appeal (i.e., preference, choice, and willingness to
Associate Editor: Leonard Lee pay) for this specific audience, did Schiller choose his
words wisely? Did framing the iPhone’s ability to with-
Advance Access publication December 14, 2021
stand shocks as “toughness” rather than “durability” and its
C The Author(s) 2021. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc.
V
All rights reserved. For permissions, please e-mail: journals.permissions@oup.com  Vol. 49  2022
https://doi.org/10.1093/jcr/ucab070

288
WEISS 289

visual appeal in terms of “beauty” rather than “aesthetics” “may increase the accessibility of all attributes of a good”
increase its appeal for fans? Can we reliably predict which (Morewedge and Giblin 2015, 343), we theorize and dem-
product features will appeal to which consumers? This ar- onstrate that categorizing an item as being inside the cate-
ticle explores the possibility that product features may res- gory of self actually creates advantages for a specific
onate differently with different consumers based on how subset of the item’s features.
consumers classify the product in relation to their selves. Which item features will benefit from egocentric proc-
We argue that framing the iPhone in terms of person- essing? Since we use the self to process information about
related features, like toughness and beauty, helped Apple persons (Rogers et al. 1977), both ourselves and others, we
maximize the phone’s appeal to consumers who classify predict that the features of an item that apply to persons,
the iPhone as being inside the category of self. akin to the features that we routinely use the self-concept
Prior research has found that we view products as more to process, should gain advantages when processing is ego-

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


or less self-attached, linked to our identity, and associated centric. To test this prediction, we introduce a new theoret-
with our self or part of our self, based on factors such as ical distinction between two types of features based on
whether we own the item or feel a connection to its brand whether they mainly apply to people or to products. Our
(Belk 1988; Escalas and Bettman 2003; Kirmani 2009). pretests found that English speakers perceive some fea-
From a cognitive perspective, we can understand these tures, like toughness and beauty, to be more appropriate for
relationships between products and people as different describing people than products. We call the set of features
ways of expressing how a product is classified in relation that meets this criterion person-related. In contrast, we
to the self. While we have long understood the self as a car- found that English speakers perceive other features, like
dinal reference category for segmenting, organizing, and durability and aesthetics, to be more appropriate for de-
understanding our environment (Rogers, Kuiper, and scribing products than people. We call this set of features
Kirker 1977), especially our social environment (Brewer product-related. The higher applicability of person-related
1991), recent research has begun to explore the ways the features to the category of self (vs. of products) is predicted
self is also used to classify and understand material objects. to boost their consideration, processing, and importance in
Just as people can be classified into in-groups and out- decisions about items that people classify as being inside
groups, products can also be classified as being inside or (vs. outside of) the category of self. Emphasizing the
outside of the category of self, based on ownership (Weiss iPhone’s beauty and toughness, then, should increase its
and Johar 2013). These ongoing acts of classification—di- appeal with Apple fans (vs. nonfans) since we expect these
viding people into “us” and “them” and objects into “me” person-related features to be more important and easier to
and “not me”—allow us to meet our basic and ongoing consider and process for people who classify the iPhone as
need to identify and separate the “self” from the “not self” being inside the category of self.
(Goffman 1972). Marketers regularly make choices about which features
Relating items to the self has been shown to boost mem- to emphasize (e.g., durability or aesthetics) and what to
ory (Symons and Johnson 1997) and to impact product call a given feature (e.g., durable or tough); knowing which
judgment (Weiss and Johar 2013), valuation (Dommer and features will resonate with which audience can be a power-
Swaminathan 2013), and choice (White and Dahl 2007). ful tool in targeted marketing communications. Table 1
Adding to this research, we propose that classifying an provides examples of several features that marketers can
item as being inside the category of self allows people to describe using person-related or product-related terms
use this category (in addition to the “default” category of (appendix).
products) to process information about the item. Following Our findings have important marketing implications:
Weiss and Johar’s (2013, 2016) theory of egocentric cate- they open up new ways for marketers to tailor their mes-
gorization, we call this egocentric processing because the sage to specific consumer segments. We demonstrate not
consumer uses the self as a central category and processes only that marketers can increase the appeal of high margin
item information based on the self. While previous re- products for consumers who classify these products as be-
search has asserted that relating a good to a person’s self ing inside the category of self, by using person-related

TABLE 1

EXAMPLES OF PERSON-RELATED/PRODUCT-RELATED TERMS

Feature Person-related term Product-related term


Ability to withstand shocks Tough, resilient Durable, sturdy
Visual appeal Beautiful, glamorous Aesthetically pleasing, dazzling
Cross-situation functionality Adaptable Adjustable
Fit for actual use Pragmatic Practical
290 JOURNAL OF CONSUMER RESEARCH

features to target these consumers, but also (and crucially Bodenhausen, and Becker 2007), product memory
from a practical standpoint) how marketers can identify improves (Symons and Johnson 1997) and the accessibility
and target these consumers based on their brand use (study of positive product information is boosted
1), affiliation (study 2), or stage in the purchase process (Nayakankuppam and Mishra 2005). Owning (vs. not own-
(studies 3 and 7). For example, study 1 shows that among ing) a product thus improves its valuation and preference
Apple users, ad click for the iPhone 11 (vs. Galaxy 10) was for us, creating an “endowment effect” (Thaler 1980).
higher when the ad described the phone in person-related We can also feel psychological ownership for an item
terms (strong and beautiful), but not when the ad described without legally owning it (Pierce, Kostova, and Dirks
the phone in product-related terms (durable and dazzling). 2003), for example, because we used to own it (Strahilevitz
We also show that even when consumers do not classify and Loewenstein 1998) or anticipate owning it in the future
the marketer’s product as being inside the category of self, (Ariely and Simonson 2003). Even simply imagining that

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


marketers can still benefit from using person-related fea- we own an item increases its association with the self
tures by relating such features to a relevant product the (Peck and Shu 2009), leading to better item memory
consumer already classifies this way. Study 6 demonstrates (Cunningham et al. 2008) and a distinct pattern of brain ac-
this approach to the sale of add-ons for a product the con- tivity (Turk et al. 2011).
sumer already owns—a case for the consumer’s phone. We also feel connected to brands that we have used and
The study shows that relating the case’s features to the con- had experience with (Kirmani 2009) or that are associated
sumer’s phone rather than to the case itself (i.e., a case that with an in-group (Escalas and Bettman 2003). Connection
“makes your phone resilient” vs. “is resilient”) increased to a brand (e.g., Apple) can transfer to its products (e.g., an
willingness to pay (WTP) for the case when the features iPhone), thereby improving valuation and preference for
were described in person-related terms (resilient and beau- the brand’s products (MacInnis and Folkes 2017; Park,
tiful), but not when the features were described in product- MacInnis, and Priester 2006).
related terms (durable and aesthetically pleasing).
Across six lab studies and one field experiment, using a THE SELF AS AN ORGANIZING
variety of products and features, we found robust evidence CATEGORY FOR PRODUCTS
for the advantages of person-related features in consumer
decisions about in-self (vs. out-self) items. We document Some researchers have taken the product-self link fur-
three advantages of person-related features: greater consid- ther, arguing that products can extend (Belk 1988), be in-
eration (as shown through free feature listing), faster proc- corporated into (Escalas and Bettman 2003), overlap with
essing (as shown through response time), and higher or become included in (Park, Eisingerich, and Park 2013),
importance (as shown through ad click-through rate, WTP, or become part of (MacInnis and Folkes 2017) the self.
and choice). Theory-driven mediators and moderators im- From a cognitive perspective, we can understand these
plicate egocentric processing as the driver of these effects. relationships as instances of a process of egocentric catego-
We specifically find that the advantages of person-related rization (Weiss and Johar 2013, 2016): different relation-
features for in-self (vs. out-self) items are bigger when sim- ships to products map onto differences in how consumers
ilar advantages are present in self-judgment, when consum- categorize or classify products in relation to the self, with
ers are self-focused, and when the categorization cue self-linked products classified by the consumer as being in-
(brand, ownership) is essential to how people define the side the category of self.
self. We discuss these theory-driven boundary conditions If we understand product-self links as instances of ego-
in detail following our report on studies 1–3. We next re- centric categorization, then we can understand the circum-
view the relevant literature on egocentric categorization, stances that facilitate these links as cues that help people
ownership, and branding and offer theoretical support for classify items in relation to the self. We call these circum-
our predictions. stances self-categorization cues, or self-cues for short–
mental or sensory markers that guide people in their classi-
THE PRODUCT-SELF LINK IN fication of products as being inside or outside of the cate-
CONSUMER BEHAVIOR gory of self. Here we look at two self-cues: ownership and
branding. When a product’s self-cue indicates to consum-
Extant research concludes that products and brands are ers that the product is inside the category of self (because
central to the consumer self. Products we own become they own it or feel connected to its brand), we will refer to
self-linked because we use them to construct, maintain, the product as an in-self item. When a product’s self-cue
and enhance the self (Beggan 1992; Sirgy 1982), which indicates to consumers that the product is outside of the
gives products personal meaning (Rochberg-Halton 1984) category of self (because they neither own it nor feel con-
and creates emotional attachment (Ahuvia 2005). As prod- nected to its brand), we will refer to the product as an out-
ucts become associated with our selves (Gawronski, self item.
WEISS 291

ITEM CATEGORIZATION AND THE related (product-related)? Prior research in cognitive psy-
ADVANTAGES OF SPECIFIC FEATURES chology has noted that “the attributes used to describe per-
sons differ substantially from those used to describe
What are the consequences of categorizing an object in- inanimate objects” (Mitchell, Heatherton, and Neil Macrae
side the category of the self for how information about the 2002, 15238), and that people tend to associate different
object is processed and understood? Research on categories attributes with objects and with people (Barton and
and concepts has shown that when people classify an ob- Komatsu 1989; Mitchell, Macrae, and Banaji 2005). But
ject within a focal category (e.g., a laptop as a gaming the ways people understand attributes in terms of their ap-
[working] device), it is easier for them to think about the plicability to people versus products and the implications
object in terms of attributes that are more applicable to this of this distinction for marketing have not yet been studied.
category (e.g., as fun [efficient]) (Higgins 1996; Medin Ample research has established the close link between

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


1989). The stereotyping literature provides a similar con- applicability and usage frequency: terms that are more ap-
clusion (Tajfel 1969); when judging members within a so- plicable to a noun are more frequently used to describe it
cial category, people consider (Miller and Turnbull 1986) (for a review see Higgins 1996). Thus, as a real-world indi-
and heavily weigh (Bodenhausen and Lichtenstein 1987) cator of whether features vary in their applicability to peo-
information about attributes that are applicable to that so- ple versus products we examined whether certain English
cial category. terms are used more or less frequently to describe people
If we think of ownership and branding as self-cues that versus products. To calculate the frequency with which
guide categorization of an item as being inside or outside commonly used adjectives referred to person- and product-
of the self, we can begin to explore the possibility that ego- nouns, we used secondary data from two comprehensive
centric processing of an item may create advantages for datasets: the English Web Treebank and WordNet. The
some rather than all (Morewedge and Giblin 2015 on own- results revealed substantial variation in the relative fre-
ership) of the item’s features. Which item features will quency with which 60 adjectives were used to describe per-
benefit from egocentric processing of item information? If sons and products (web appendix A).
item features that are more (vs. less) applicable to an item’s Frequency of use, while a valuable indicator of applica-
bility, must be calculated through a complex and laborious
category gain advantages, then we can expect that classify-
process and is available only for a limited number of adjec-
ing an item as being inside the category of self will create
tives that are sufficiently common in available data sets
advantages for those features that are more applicable to
(web appendix A). In order to make our methods more ap-
the self. We use the self-concept to process information
proachable and to broaden the scope, we tested whether
about persons; this includes, for example, how slender,
people actually perceive adjectives that are more (less) fre-
beautiful, or smart we consider ourselves to be (Rogers
quently used to describe people (products) as being more
et al. 1977), but also how slender, beautiful, or smart we
applicable to people (products). Ninety-five MTurk work-
consider others (Dunning and Hayes 1996). We propose,
ers were asked about the extent to which the 60 adjectives
then, that item features that are conveyed in person-related for which we had usage frequency data applied to people
terms (e.g., how slender, beautiful, or smart—or tough—an and products. In line with the high correspondence between
iPhone is), akin to the features that we routinely use the relative frequency of use and perceived applicability, the
self-concept to process, should gain advantages from ego- frequency–applicability correlation was positive and sig-
centric processing. nificant (r ¼ .77, p < .0001); this high correlation validated
that the direct measure of perceived applicability (a mea-
PERSON-RELATED VERSUS PRODUCT- sure that is easily used and can be applied to any adjective)
RELATED FEATURES captures a meaningful real-world variation.
In further pretesting among 121 MTurk workers, we
The types of features that people use to describe prod- used the perceived applicability measure to classify fea-
ucts vary along various dimensions, such as tangibility, tures for our studies (see appendix for the list of relevant
concreteness, or familiarity (Carpenter, Glazer, and features). In our studies we used features that were rated as
Nakamoto 1994; Trope and Liberman 2003). To explore more (less) applicable to people than to products to a de-
the possibility that the way consumers classify an item in gree that was statistically significant as person-related
relation to their selves can create advantages for specific (product-related) features. Features that were rated as simi-
features, we propose a novel distinction between feature larly appropriate for describing people and products were
types, based on relative applicability to people versus considered neutral.
products. Notably, person-relatedness can vary both within fea-
Are there terms in the English language that are per- tures (across different labels of the same feature, e.g.,
ceived as more applicable to or appropriate for describing strong vs. durable) and between features (across different
person (product) traits, and, in that sense, are more person- features, e.g., honest vs. portable). Our studies look at both
292 JOURNAL OF CONSUMER RESEARCH

types of variation. Study 4 tests the discriminant validity of through rate), processing (as shown through response
our typology by showing its better ability to account for time), and consideration (as shown through free feature
the results compared to other distinctions (abstract vs. con- listing). We can understand these advantages as creating a
crete, intangible vs. tangible). This typology adds to the in- more prominent role for person-related features in con-
fluential work that has generalized and standardized brand sumer decisions about in-self (vs. out-self) products. Stated
personality research (Grohmann 2009; Johar, Sengupta, formally,
and Aaker 2005). While the brand personality typology
H1:Person-related features gain advantages in decisions
asks which personality traits associated with products and
about in-self (vs. out-self) items.
brands resonate with consumers in general (Aaker 1997),
we use a broader set of traits (including those not related to H2:Egocentric processing drives the advantages of person-
related features in decisions about in-self (vs. out-self)
personality, e.g., beauty) to ask which product traits reso-

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


items.
nate with which consumers.
How does the distinction between person-related and If consumers use egocentric processing to make deci-
product-related features apply to the world of marketing? sions about an item (i.e., if they process information about
Marketers can be expected to use product-related features to the item based on the category of self), then we can expect
promote products and brands, and they do. For example, on to find two complementary lines of process evidence: (1)
their US websites, various car manufacturers describe their item inclusion in the category of self and (2) situational use
cars in terms of features that our pretests classified as product- of the self (rather than of other relevant categories) as the
related: Mercedes-Benz describes the 2020 EQC as focal category for objects. We will use both lines of evi-
“aesthetic” and “compact,” and BMW portrays its 8 Series as dence to implicate egocentric processing as the underlying
“smooth” and “convenient.” However, any observer of con- psychological driver for the predicted effect.
temporary advertising can see that marketers also use person- Next, studies 1–3 test the importance advantage of
related features to position and advertise products and brands, person-related features (hypothesis 1) across advantage
as shown in the opening quote from Apple. For example, on indicators (click-through rate, study 1; WTP, study 2;
their US websites, various car manufacturers also describe choice, study 3) and self-cues (brand, studies 1–2; owner-
their cars in terms of features that our pretests classified as ship, study 3). To examine egocentric processing as a
person-related but here are used to describe products: driver (hypothesis 2), we test whether item inclusion in the
Mercedes-Benz portrays the 2018 E-class Sedan as category of self (study 2) or use of self as the focal cate-
“intelligent” and “daring,” and BMW depicts its 7 Series as gory for objects (study 3) statistically mediate the predicted
“intuitive” and its 5 Series as “elegant.” Thus, while marketers effect. We then go on to introduce moderators (hypothesis
already use both person- and product-related features to posi- 3) and test their effect (studies 4–7). Following Meyvis and
tion their products and brands, we study how they might make Van Osselaer (2018) we winsorized variables that can be
strategic choices between these features based on the specific impacted by outliers (i.e., WTP and RT); we report robust-
consumer segment they wish to target. ness checks for this and for variable transformation in web
Next, we outline our predicted effect (hypothesis 1) and appendix B.
its psychological driver (hypothesis 2) and present findings
from studies 1 to 3. We then introduce theory-driven mod- Study 1: Importance Advantage in the
erators (hypothesis 3) and use studies 4–7 to test them. Likelihood to Click on a Smartphone Ad on
Figure 1 presents an overview of the conceptual model Facebook
and the supporting evidence provided by each study.
As a first test for the importance advantage of person-
related features in decisions about in-self (vs. out-self)
THE ADVANTAGES OF PERSON- products (hypothesis 1), we used a field experiment, where
RELATED FEATURES IN DECISIONS multiple cues exist and people act in their natural environ-
ABOUT IN-SELF (VS. OUT-SELF) ment. An actual expression of purchase interest—the likeli-
PRODUCTS hood to click on an ad on Facebook—was used as indicator
for feature importance. Features were considered more im-
When making decisions about in-self items, consumers portant if using them to describe the product in the ad in-
may use the category of self to guide their decisions about creased the likelihood to click on the ad. Following from
the items. Since we use the self-concept to process infor- the example that opened the article, we specifically exam-
mation about persons, we predict that processing item in- ined whether consumers would be more likely to click on
formation based on the category of self will create an ad for an in-self (vs. out-self) smartphone when it is de-
advantages for the item’s person-related features. We mea- scribed using a person-related (but not a product-related)
sure the advantages of person-related features in terms of feature. Self-cue was manipulated by changing whether the
importance (as shown through WTP, choice, and ad click- advertised smartphone was by a brand the audience had
WEISS 293

FIGURE 1

CONCEPTUAL MODEL AND SUMMARY OF EVIDENCE

Egocentric processing

Decisions about in-self Item inclusion in the category of Advantages of person-


Study
self
(vs. out-self) items related features
Mediation (H2) 2
Advantage
Evident Mediation (H2) 5 Study
Self-cue (H1)
Study from:

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


(in vs. out)
Moderation by extent to which 6–7 Consideration 5
Brand 1–2 self-cue is self-defining (H3b)
Processing 4
Ownership 3–7 Use of self as the focal category for
Study Importance 1–3, 6–7
objects

Mediation (H2) 3
Evident
from:
Moderation by self-focus (H3a) 5

extensive usage experience with (Kirmani 2009). (See web to 220,801 impressions (i.e., total number of exposures)
appendix C for the full stimuli for all studies.) over the duration of the campaign.

Method. Study 1 was conducted in the fall of 2019 on


Facebook. The objective was to increase the click-through Results and Discussion. Altogether, 1,457 Facebook
rate of an advertisement for a smartphone among US con- users (of the 156,510 users reached) clicked on the ad they
sumers between the ages of 18 and 65 who access saw, giving a total click-through rate of .931% (the average
Facebook using Apple products (based on targeting infor- CTR for Facebook ads is 0.90%; Irvine 2020). A logistic
mation available through Facebook). The ad, once clicked, regression of ad click (1 ¼ click, 0 ¼ no click) on cue (in-
led to the official selling page of the smartphone. The study self ¼ 1, out-self ¼ 1), feature type (person-related ¼ 1,
utilized Facebook split testing, an A/B testing service that product-related ¼ 1), and their interaction showed no ef-
randomly divides audience exposure across each version of fect of feature type (b ¼ .026, SE ¼ .027, v2 ¼ .94, p ¼ .33),
an ad to test which version performs better, while ensuring a significant effect of self-cue (b ¼ .07, SE ¼ .027,
audiences are evenly split and statistically comparable. v2 ¼ 6.8, p ¼ .009), and importantly a significant two-way
The study used a 2 (cue: in-self vs. out-self)  2 (feature interaction (b ¼ .08, SE ¼ .027, v2 ¼ 9.66, p ¼ .002).
type: person-related vs. product-related) between-subject Planned contrast supported the predicted importance ad-
design. Our target market was composed of people who ac- vantage of person-related features in decisions about in-
cess Facebook using an Apple device; their usage experi- self (vs. out-self) items (hypothesis 1): ad click was higher
ence with Apple rendered the Apple brand (a non-Apple for the ad for the in-self (vs. out-self) phone when it was
brand) an in-self (out-self) cue. Our target audience for the described as strong and beautiful (Min-self ¼ 1.1% vs. Mout-
2
in-self (out-self) condition was therefore presented with an self ¼ .82%, v ¼ 16.68, p < .0001, odds ratio ¼ 1.35), but
ad for iPhone 11 (Galaxy 10), the most recent smartphone not when it was described as durable and dazzling (Min-self-
by Apple (Samsung) at the time of the experiment. Feature ¼ .89% vs. Mout-self ¼ .91%, v2 ¼ .12, p ¼ .72, odds
type was manipulated through two features listed on the ratio ¼ .97).
ad: for people in the person-related (product-related) fea- As stated earlier, person-relatedness can vary both
ture ad, the smartphone was described as strong and beauti- within feature (across different labels of the same feature,
ful (durable and dazzling), as in figure 2. e.g., strong vs. durable) and between features (across dif-
A daily budget of $216 per day for five days, evenly ferent features, e.g., honest vs. portable). To generalize
split between the four conditions, was set based on a 90% study 1’s within feature findings, a separate study repli-
power assessment done by Facebook. The budget led to a cated these results using a between feature design. We re-
total reach (i.e., different people exposed) of 156,510 and port these results in web appendix D.
294 JOURNAL OF CONSUMER RESEARCH

FIGURE 2

FACEBOOK ADS, STUDY 1

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


In-self, person-related In-self, product-related

Out-self, person-related Out-self, product-related

Study 2: Importance Advantage in Incentive- manipulated through the mug’s description: for participants
Compatible Valuation of a Mug in the person-related (product-related) feature condition,
the mug was described as strong (durable). As a DV, par-
Study 2 further tested the predicted importance advan- ticipants were asked for the maximum $ amount they were
tage of person-related features in decisions about in-self willing to pay (WTP) for the mug using an open-ended re-
(vs. out-self) items (hypothesis 1) using an incentive- sponse box (Atasoy and Morewedge 2018). To render the
compatible product valuation (Atasoy and Morewedge
WTP decision incentive-compatible, participants were in-
2018) rather than click-through rate as indicator for feature
formed that they may actually be required to pay the
importance. We also examined egocentric processing as a
amount they stated and receive the mug.
driver (hypothesis 2) by testing whether the predicted ef- Two items (adapted from Weiss and Johar 2013) then
fect is mediated by item inclusion in the category of self. measured the mug’s inclusion in the category of self
Self-cue was manipulated by changing the brand of the fo-
(1 ¼ “not at all,” and 7 ¼ “very much so”). The average of
cal product, a coffee mug, to reflect either an in-group or
these items, specifically the extent to which participants
an out-group of the study’s student population (Dommer
felt that the mug was “part of you” and “somehow integral
and Swaminathan 2013).
to who you are” (r ¼ .84, M ¼ 2.08, SD ¼ 1.28), served as
Method. One hundred and ninety-two students in an in- the predicted mediator. To test alternative processes, we
troductory marketing class in a large midwestern university measured participants’ perception that the mug “was like
participated in a short online study for course credit. Three a person” (M ¼ 1.48, SD ¼ 1.05) and “was like a friend”
failed an attention check and were excluded from further (M ¼ 1.44, SD ¼ .92) (1 ¼ “not at all,” and 7 ¼ “very
analysis, leading to a sample size of 189 participants. much so”; adapted from Aggarwal and McGill 2007). A
Participants were informed that they would enter a draw four-item Importance to Identity scale (Luhtanen and
for $35 in cash as an additional reward. After acknowledg- Crocker 1992) then measured the importance of the uni-
ing their understanding of this compensation, participants versity identity (1 ¼ “not at all,” and 7 ¼ “very much so”)
were presented with a picture and a short description of a with statements like “Being a [student at the focal univer-
mug—the focal product category. The study employed a 2 sity] is an important reflection of who I am.” Identity im-
(cue: in-self vs. out-self)  2 (feature type: person-related portance (a ¼ .84) did not vary across self-cue conditions
vs. product-related) between-subject design. For partici- (in-self: M ¼ 4.99, SD ¼ 1.22; out-self: M ¼ 5.11,
pants in the in-self (out-self) cue condition, the mug had a SD ¼ 1.39; t (1851) ¼ .65, NS). Participants then provided
logo of the participants’ own (a different regional) univer- demographic information and were debriefed and
sity, relating the mug to their in-group (out-group)
(Dommer and Swaminathan 2013). Feature type was 1 Two participants did not provide a response to this measure.
WEISS 295

thanked. One participant was selected at random to re- Alternative Explanations. One may wonder whether
ceive the stated prize. the effect can be explained by the anthropomorphism of
Results. WTP (winsorized at 1% and 99%) was entered the mug (McConnell et al. 2011) or its perception as a
untransformed (skewness ¼ 1.87 < 1.96) into a 2 (cue: in- friend (Chandler and Schwarz 2010), but this was not the
self vs. out-self)  2 (feature type: person-related vs. case. The same mediation analysis with three mediators
product-related) analysis of variance (ANOVA). The showed that only mug inclusion in the self (.9709, 95% CI
results showed a significant effect of self-cue (F(1, [.1194–2.2831]), but not the extent to which the mug was
185) ¼ 5.8, p ¼ .017, gp2 ¼ .0304), no effect of feature type like a person (.0902, 95% CI [.8633 to .3244]) or like a
(F(1, 185) ¼ 1.49, NS), and a significant two-way interac- friend (.0414, 95% CI [.2254 to .6692]) mediated the ef-
tion (F(1, 185) ¼ 5.32, p ¼ .022, gp2 ¼ .028). Planned con- fect. The 2  2 means and ANOVA results of mug as per-

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


trasts supported the predicted importance advantage of son and mug as friend are shown in tables 2 and 3.
person-related features in decisions about in-self (vs. out-
self) items (hypothesis 1): WTP was higher for the mug Study 3: Importance Advantage in a Choice
with participants’ own (vs. a different) university logo Between Headphones
when the mug was described as strong (in-self: M ¼ $9.36,
Study 3 further tested the predicted importance advan-
SD ¼ 7.42; out-self: M ¼ $5.07, SD ¼ 4.16, F(1,
tage of person-related features in decisions about in-self
185) ¼ 10.61, p ¼ .0013, gp2 ¼ .054), but not when it was
(vs. out-self) items (hypothesis 1) using choice (rather than
described as durable (in-self: M ¼ $6.15, SD ¼ 4.93; out-
ad click-through rate or WTP) as indicator for feature im-
self: M ¼ $6.06, SD ¼ 7.45, F(1, 185)< 1, NS).
portance. Study 3 used item ownership (rather than brand)
Mediation. In support of egocentric processing as the as self-cue. While studies 1 and 2 held ownership of the
underlying process (hypothesis 2), a bootstrapping analysis considered item constant in order to focus on brand, study
with 10,000 samples (PROCESS model 14; Hayes 2013) 3 did not provide brand information for the considered
confirmed the mediating role of mug inclusion in the self item in order to focus on ownership. Using ownership as
(.8425, 95% CI [.023–2.0545]). Mug inclusion in the self self-cue allowed us to limit the presence of confounding
significantly mediated the effect of self-cue on WTP when factors like brand associations (knowledge, experience, or
the mug was described as strong (1.1037, 95% CI [.3731– familiarity) and appearance (e.g., logo), which using brand
2.2452]), but not when it was described as durable (.2612, as a self-cue may introduce. In order to hold these factors
95% CI, [.1628 to .8791]). The 2  2 means and constant in testing our theory, studies 3 through 7 use own-
ANOVA results for mug inclusion in the self are shown in ership as self-cue.
tables 2 and 3. Mediation details for all studies are pro- Study 3 also further tested egocentric processing as a
vided in web appendix E. driver for our effect (hypothesis 2). Whereas study 2 did so

TABLE 2

MEANS BY FEATURE TYPE AND SELF-CUE, STUDY 2

Feature type Self-cue Mug in self Mug as person Mug as friend

Person-related (strong) In (n ¼ 47) 2.55 1.68 1.7


Out (n ¼ 43) 1.53 1.33 1.19
Product-related (durable) In (n ¼ 47) 2.49 1.34 1.38
Out (n ¼ 52) 1.73 1.54 1.46

TABLE 3

ANOVA FOR COMPETING MEDIATORS, STUDY 2

Mug in self Mug as person Mug as friend

F gp2 F gp2 F gp2

Self-cue 25.32** 0.1204 0.26 0.0014 2.73† 0.0146


Feature type 0.14 0.0008 0.17 0.0009 0.03 0.0001
Self-cue  Feature type 0.54 0.0029 3.25† 0.0172 5.05* 0.0266

p.1, *p <.05, **p<.0001.
296 JOURNAL OF CONSUMER RESEARCH

by checking the mediating role of including the item in the Of the six features, three were person-related (marked
category of self, study 3 did so by checking the mediating with an asterisk in figure 3) and three product-related. To
role of using the self as the focal category for objects. Our generalize the test for the predicted effect, we created two
theory suggests that while decisions about in-self items choice-sets; we used choice-set as a second (between-sub-
may involve use of the self as the focal category, decisions ject) factor. The choice of the headphones that were rated
about out-self items do not. Prior research has shown that better on two person-related features (indicated by a bold
using (vs. not using) a focal category in one task increases outline in figure 3) served as the dependent variable.
the use of the category in a later task (Srull and Wyer Participants chose between headphones before moving to
1979). We thus tested (i) whether making a choice about part 2 of the study.
in-self (vs. out-self) items increases the use of the self as To capture the use of self as an organizing category for
the focal category for objects in a subsequent task and (ii) objects, participants were asked to enter the first seven

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


whether greater use of the self as the focal category for objects “that come to your mind right now. Enter any prod-
objects predicts greater importance of person-related fea- uct or object as soon as it pops up.” After a short filler task,
tures in product choice. they were shown the list of seven items that they had gen-
To measure the extent to which people use the self as erated (presented in random order). They were asked to
the focal category for objects, we used a method from drag the items into a box in an order that reflected a rank-
Weiss and Johar (2013; study 1). As an indicator of greater ing of the degree to which they saw these items as being
situational use of the self as the focal organizing category, part of the self, with items they classify as more (less)
we looked for greater consistency (or positive correlation) “me” further at the top (bottom). Input from these entry
between (i) the order in which items come to people’s and ranking tasks served to construct the predicted media-
mind and (ii) the classification of these items as part of the tor. Two items (adapted from Peck and Shu 2009) then
self. Finding that in-self items come to mind earlier than tested the self-cue manipulation by measuring the psycho-
out-self items reflects an implicit reliance on the self as an logical ownership that participants felt for the headphones
organizing category for objects. Low consistency (or no (r ¼ .89; 1 ¼ “strongly disagree,” and 7 ¼ “strongly
correlation) between item retrieval order and item classifi- agree”). Results indicated a successful manipulation
cation in the self reflects no use of the self as an organizing (owned: M ¼ 4.21, SD ¼ 1.94 vs. unowned: M ¼ 2.08,
category for objects. SD ¼ 1.27, t (242) ¼ 10.25, p < .0001). Participants then
provided demographic information and were debriefed and
Method. Two hundred and forty-nine students in an in-
thanked.
troductory marketing class in a large midwestern university
participated in a short lab study for course credit. Five par- Results. Headphones choice (better rated person-
ticipants failed an attention check and were excluded from related features ¼ 1, better rated product-related
further analysis, resulting in a sample size of 244 partici- features ¼ 0) was submitted to a logistic regression with
pants. Prior research has found that directing attention to cue (in-self ¼ 1 vs. out-self ¼ 1) as a single factor. The
the self facilitates the use of the self as a category for prod- analysis controlled for which of the two choice-sets the
ucts (Weiss and Johar 2013; hypothesis 3b). To direct the participants responded to (b ¼ .74, v2 ¼ 27.06, p < .0001,
attention of all participants to the self in the context of the odds ratio ¼ 4.41). In further support of the importance ad-
focal product category of headphones, participants read a vantage of person-related features in decisions about in-
definition of the relevant identity of audiophile and were self (vs. out-self) items (hypothesis 1), more people chose
asked to “write two things about your self that reflect your the headphones with better person-related features when
audiophile side.” Cue (in-self vs. out-self) was the sole fac- the decision was about owned (vs. unowned) headphones
tor and was manipulated between subjects using ownership (b ¼ .3, v2 ¼ 4.41, p ¼ .036, odds ratio ¼ 1.815; 50% or 29/
(Weiss and Johar 2013) by asking participants to imagine a 58 vs. 38% or 23/61 and 83% or 50/60 vs. 71% or 46/65, in
choice between two items either before or after owning choice-sets 1 and 2, respectively).
them. Participants in the in-self (out-self) products condi- The results also supported the prediction that making a
tion were asked to imagine that they “buy (consider) two choice about in-self (vs. out-self) items increases the use of
high quality headphones, with the intent to keep (acquire) the self as an organizing category for objects, as shown
and use only one of them.” They were then shown a table through greater consistency between (i) the order in which
with the star rating of each device along six attributes and items come to people’s mind and (ii) the classification of
were asked, “Based on how each of the headphones you these items as part of the self. A within-subject rank-order
bought (considered) is rated along the following features, (Spearman) correlation between these two variables was
choose which one you prefer to keep (buy).” (For discus- calculated for each participant, Fisher transformed, and
sion of the theoretical and practical relevance of comparing submitted to an ANOVA with self-cue as a single factor. In
decisions before and after ownership, see the General line with predictions, participants making decisions about
Discussion.) in-self (vs. out-self) items had significantly higher
WEISS 297

FIGURE 3

TWO CHOICE-SETS BETWEEN HEADPHONES, STUDY 3

Attribute Headphones A Headphones B Attribute Headphones A Headphones B

Outdoorsiness * Compactness

Compactness Outdoorsiness *

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


Dependability * Adjustability

Adjustability Dependability *
Honesty * Smoothness

Smoothness Honesty *

NOTE.– Asterisks indicate person-related features. Bold outline indicates which headphones in each choice-set are rated better on two person-related features.

individual correlations (in-self: M ¼ .28, SE ¼ .059 vs. out- However, we wanted to go beyond these findings to ex-
self: M ¼ .08, SE ¼ .057, F(1, 242) ¼ 5.98, p ¼ .015, plore what role the variability of the self may play in the
gp2 ¼ .0241). effect. Would we see this effect more strongly for some
features or for some people? When would the effect attenu-
Mediation. In further support of egocentric processing ate? To better understand the psychological process and to
as the underlying process (hypothesis 2), a bootstrapping advance theory, we go on to study boundary conditions.
analysis (PROCESS model 4; Hayes 2013) with 10,000 These boundary conditions suggest ways that marketers
samples confirmed the mediating role of the use of the self could refine and add precision to their targeting (as demon-
as the focal category for objects (as captured through the strated in studies 6 and 7). It is important to note, though,
rank-order correlations) (.0441, SE ¼ .0316, 95% CI that the overall effect shown in studies 1–3, and at an aver-
[.0035, .1304]). age level of each moderator in studies 4–7, suggests the po-
tential for applications that are broader and easier to
implement. We next introduce three factors that should
THEORY-DRIVEN BOUNDARY moderate the effect only if, as our theory holds, egocentric
CONDITIONS processing drives the effect.
First, while person-related features can be expected to
Studies 1–3 show that the effect of self-cues on the
benefit from egocentric processing overall, we would not
advantages of person-related features is robust enough to expect all person-related features to benefit equally. If we
be seen overall; studies 2 and 3 also implicate the role of use the self to process information about in-self items, fea-
egocentric processing in the effect. Finding that egocentric tures that we process faster when judging the self should
processing creates an overall advantage for person-related be processed faster when judging in-self items. Since fre-
features—regardless of the specific feature or the charac- quent processing of information increases processing speed
teristics of the individual respondent—may seem surpris- (Higgins 1996), we would expect that egocentric process-
ing. After all, the self is the category for a specific person ing will benefit the set of features we frequently use to
and is therefore highly variable. But this overall effect is judge ourselves, including self-features we care about
consistent with foundational findings in the self-reference (Bargh and Pratto 1986) and features we often use to judge
literature that relating (vs. not relating) to the self a list of others where the self is used as a comparison standard
adjectives “deemed appropriate for a self-description task” (Mussweiler and Bodenhausen 2002). If we see that judg-
improved recall of the adjectives overall, not just the ones ments about in-self (vs. out-self) items are faster for fea-
with relevance to the self (Rogers et al. 1977, 680). This re- tures that are processed faster in self-judgment, then this
call effect is likely explained by the role the self-concept will show that this advantage is driven by the use of the
plays in processing information about others (Dunning and self to process information about in-self items (hypothesis
Hayes 1996). 3a).
298 JOURNAL OF CONSUMER RESEARCH

Second, while the results of studies 1–3 indicate that H3:The advantages of person-related features for in-self (vs.
people generally classify items with an in-self cue as being out-self) items will be bigger when (a) similar advantages
inside the category of self, some people may not use the are present in self-judgment, (b) consumers are self-focused,
self as the focal organizing category and, instead, classify and (c) the self-categorization cue is self-defining. These
items (even ones with an in-self cue) based on other rele- advantages will appear overall, for an average level of each
moderator, and be smaller or even absent when these moder-
vant categories (e.g., the category of smartphones for an
ators are low.
iPhone). Prior research finds that people’s chronic or situa-
tional tendency to focus on themselves increases the use of Next, we present the results of study 4, which used real
the self as a focal category for what we call in-self items behavior (i.e., response time) to test for a processing ad-
across what we call different self-cues (brand: Dagogo- vantage in consumer decisions about in-self (vs. out-self)
Jack and Forehand 2018; ownership: Weiss and Johar items (hypothesis 1) and for moderation by whether a simi-

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


2013). When self-focus is low, people’s attention is di- lar advantage exists in self-judgment (hypothesis 3a).
rected outwards, away from the self (Duval and Wicklund Study 5 used free feature listing to test for a consideration
1972; Fenigstein, Scheier, and Buss 1975). Outward fo- advantage (hypothesis 1) and for moderation by self-focus
cused people are thus less likely to use the self (vs. other (hypothesis 3b). Studies 6 and 7 further tested for an im-
categories) as an organizing class for objects. If we see that portance advantage (hypothesis 1; shown also in studies 1–
our effect is stronger for self-focused people, then this will 3) and for moderation by whether the self-cue is self-
show that the effect is driven by the use of the self as a cat- defining (hypothesis 3c).
egory for in-self items (hypothesis 3b).
Third, different self-cues or circumstances that can fos- Study 4: Processing Advantage in Judging a
ter a link between a product and the consumer’s self (own- Mug—Moderated by Advantage in Self-
ership, brand) may be more or less essential to how Judgment
different consumers define their selves (Belk 1988; Study 4 had two main goals. First, we tested the predic-
Ferraro, Escalas, and Bettman 2011). When these self-cues tion that person-related (but not product-related or neutral)
are more self-defining, they unambiguously mark the features gain a processing advantage in decisions about in-
boundaries of the self for a consumer, and they clearly self (vs. out-self) products (hypothesis 1). The processing
guide classification of an item as being inside or outside of advantage was captured using response time to judge a
the self (Burris and Branscombe 2005) and a consumer’s traveling mug on 27 different features. Second, we tested
subsequent response to the item. For example, while con- the prediction that the processing advantage of in-self (vs.
sumers were found to be generally more likely to recycle out-self) items is moderated by whether a similar advan-
rather than trash items that our framework defines as in- tage also exists in self-judgment (hypothesis 3a). If egocen-
self (vs. out-self) (e.g., a Coke vs. Pepsi can for Coke tric processing drives the effect of self-cue, the effect can
drinkers, a cup with the US vs. UK flag image for be expected to be bigger for those features that have a
Americans), the effect of self-cues (i.e., brand, flag image) processing advantage in self-judgment, but smaller or even
was stronger for people who found the focal self-cue self- absent for features that do not have a processing advantage
defining (i.e., those who strongly identified themselves as in self-judgment.
Coke drinkers or as Americans) (Trudel, Argo, and Meng Method. One hundred and eighty-eight students in an
2016). A similar effect has been shown with an ownership introductory marketing class in a large midwestern univer-
self-cue based on individual differences about whether sity participated in a short lab study for course credit.
people perceive what they own (“mine”) as defining who Three participants failed an attention check and were ex-
they are (“me”) (Weiss and Johar 2013). People who do cluded from further analysis, resulting in a sample size of
not perceive “mine” as “me” do not use ownership to clas- 185 participants. Study 4 used three feature types: person-
sify an item as being inside or outside of the self and in- related, product-related, and neutral (appendix). We in-
stead assign to owned and unowned items the same levels cluded neutral features (those that were rated as similarly
of “me-ness.” If we see that decisions about in-self (vs. appropriate for describing people and products) in order to
out-self) items following an ownership self-cue have a account for the full spectrum of features. This resulted in a
stronger effect on people who use ownership to define the 2 (cue: in-self vs. out-self, between-subject)  3 (feature
boundary of the self (than on people who do not), then this type: person-related vs. neutral vs. product-related, within-
will show that the effect is driven by the classification of subject) mixed design. On each lab table, we placed a trav-
in-self items as being inside the category of self (hypothe- eling mug that we asked participants to evaluate. Self-cue
sis 3c). was manipulated using ownership: participants in the in-
These predicted boundary conditions lead to the follow- self condition learned that they get to keep the mug they
ing hypothesis: evaluate; participants in the out-self condition learned that
WEISS 299

they get to keep a different mug that they will receive at full model did not converge, following Brauer and Curtin
the end of the study (and thus did not own the mug they (2018, 404), we removed the by-attribute RTself random
were to evaluate). After verifying that participants under- slope and all covariances among random effects to obtain
stood whether they owned the focal mug and guiding them convergence.
through its inspection, we asked them to place the mug on We first report the omnibus (three feature-types) analy-
the table. sis and then test our hypotheses based on contrasts with the
Participants were then told that they will be presented two focal types, person-related and product-related. RTself
with several short general descriptions. They were asked to (F(1, 40.7) ¼ 8.16, p ¼ .007) was a significant predictor.
press “s” on a keyboard if the description reflects the mug Self-cue (F(1, 178) ¼ .24, p ¼ .62), the omnibus effect of
and “o” if not, and to do so as quickly as possible to reflect feature-type (F(2, 24.1) ¼ 1.46, p ¼ .25), and their interac-
their actual evaluation of the mug. Participants first prac- tion (F(2, 24) ¼ 2.04, p ¼ .15) as well as the omnibus inter-

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


ticed the task on five objective mug attributes (e.g., black, action of feature-type with RTself (F(2, 44.3) ¼ 1.28,
metallic) and then moved on to respond to the focal set of p ¼ .29) and the interaction of self-cue with RTself (F(1,
words. This focal set of words (web appendix F), presented 41) ¼ .58, p ¼ .45) were not significant. The omnibus
in randomized order, was composed of nine person-related three-way interaction (F(2, 44.6) ¼ 3.12, p ¼ .054) was
features (e.g., creative, resilient), nine neutral features marginally significant.
(e.g., stylish, reliable), and nine product-related features Importantly, consistent with overall support for hypothe-
(e.g., smooth, compact), all selected from the list of pre- sis 1, the interaction contrast between self-cue and the two
tested features in the appendix. The time (in ms) it took focal feature-types was negative and marginally significant
each participant to respond to each word served as the de- (b ¼ .86, SE ¼ .43, t (24.1) ¼ 2.01, p ¼ .056). In line
pendent variable. with hypothesis 3a, this two-way interaction contrast was
Participants then repeated the same task of responding at qualified by RTself, resulting in a significant three-way in-
maximal speed to the same features using the same keys teraction contrast (b ¼ .17, SE ¼ .07, t (42.3) ¼ 2.37,
with the self as the target of judgment (i.e., indicated for p ¼ .023; for untransformed means see figure 4).
every feature whether it reflects them). The time (in ms) it We used spotlight analysis (Fitzsimons 2008) to inter-
took each participant to respond to each word with regard pret this three-way interaction contrast. The interaction be-
to themselves served as the predicted moderator. tween self-cue and the two focal feature-types was
Participants then provided their demographic information, negative and significant at fast RTself (i.e., at 1 SD below
received either the mug they evaluated or a different mug the mean of RTself, b ¼ 1.76, SE ¼ .57, t (55.1) ¼ 3.08,
(based on the experimental condition), and were debriefed p ¼ .0032; the interaction remains significant at p ¼ .05 for
and thanked. RTself as slow as .025 SDs below the mean). Planned con-
trasts showed that participants in the in-self (vs. out-self)
Results. We reorganized the data to include 27 rows mug condition responded faster to the mug’s person-
per participant, one for each feature. Response time (in ms) related features (Min-self ¼ 1,124 vs. Mout-self ¼ 1,227, t
to a feature when rating the mug (or RTmug) (winsorized at (205) ¼ 2.1, p ¼ .037), but not to its product-related fea-
5% and 95%, square root transformed) was entered to a lin- tures (Min-self ¼ 1,263 vs. Mout-self ¼ 1,229, t(178) ¼ .67,
ear mixed-effects model (using proc mixed in SAS 9.4). p ¼ .51) at fast RTself. Moreover, in line with hypothesis
The model included two (cue: in-self vs. out-self, between 3a, a similar analysis showed no effect at slow RTself (i.e.,
subject), feature-types (person-related vs. neutral vs. at 1 SD above the mean of RTself; ps > .51).
product-related, within subject), feature response time
when rating the self (or RTself, winsorized at 5% and 95%, Alternative Explanations. One might ask whether the
square root transformed, within subject), and all their two- results reflect a general fluency effect, whereby feature flu-
way and three-way interactions. Note that the data contain ency expedited judgment of all targets, but this was not the
two random factors, subject and feature, that are crossed case. We found that the observed advantage (i) is specific
with each other. Since RTself is continuous and varies both to in-self (but not out-self) items and (ii) appears only for
within subject and within feature, it was adaptively cen- person-related (but not product-related) features. These
tered (Pier et al. 2018). To account for nonindependence in findings are inconsistent with a mere fluency account and
the data (Brauer and Curtin 2018), we included the appro- consistent with our egocentric processing account.
priate random effects, specifically a by-participant random Discussion. Using real behavior and a wide array of
effect with an intercept and slopes for feature-type, RTself, features, the results of study 4 supported the predicted
and their interaction, as well as a by-attribute random ef- processing advantage of person-related (but not product-
fect with an intercept and slopes for self-cue, RTself, and related) features in decisions about in-self (vs. out-self)
their interaction, plus all possible covariances. items (hypothesis 1). The results also supported the predic-
Satterthwaite approximation was used to estimate the tion that the processing advantage for in-self items is big-
degrees of freedom (Brauer and Curtin 2018). Since the ger (smaller or even absent) when it exists (does not exist)
300 JOURNAL OF CONSUMER RESEARCH

FIGURE 4

FASTER TO JUDGE THE PERSON-RELATED FEATURES OF AN IN-SELF (VS. OUT-SELF) MUG WHEN FAST TO JUDGE THE SELF,
STUDY 4

RT mug (ms)
1500
Out-Self In-Self

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


1200

900
Person- Neutral Product- Person- Neutral Product- Person- Neutral Product-
Related Features Related Related Features Related Related Features Related
Features Features Features Features Features Features
Fast Avg. Slow
RT Self

NOTE.– Fast/Avg./Slow RTself represents (1 SD below the mean/the mean/1 SD above the mean) time to judge the self.

in self-judgment (hypothesis 3a). Finding a bigger (smaller Method. Three hundred and sixty-two students in an
or even no) processing advantage for in-self (vs. out-self) introductory marketing class in a large midwestern univer-
items for the features on which a person is faster (slower) sity participated in a short study for class credit. Seven
to judge the self supports our theory that this advantage is failed an attention check and were excluded from further
driven by egocentric processing (hypothesis 2). Finding a analysis, leading to a sample size of 355 participants. The
marginally significant two-way interaction between self- study employed a cue (in-self vs. out-self) by self-focus
cue and the two feature-types (i.e., person- and product- (measured) between-subjects design. Self-cue was manipu-
related) in addition to a three-way interaction with RTself lated between subjects using ownership by asking partici-
supports our prediction that although the presence of an ad- pants to imagine making a choice between two headphones
vantage in self-judgment is a moderator, it is not a neces- either before or after owning them. Those in the in-self
sary condition for the advantages of person-related features (out-self) headphones condition read: “Imagine that you
(see also studies 1–3). are in the market for a high-quality set of headphones.
After narrowing down the options based on price, quality,
Study 5: Consideration Advantage in and brand, you eventually buy (consider) two sets of head-
Headphones Choice—Moderated by Self-Focus phones, with the intent to keep (buy) and use only one of
them.” We informed participants that the two items were
Study 5 had three goals. First, we tested the predicted selected based on price, quality, and brand in order to limit
consideration advantage of person-related features in deci- any difference in the inferences that participants could
sions about in-self (vs. out-self) items (hypothesis 1) by have made in each self-cue condition about the criteria for
comparing the person-relatedness of the factors listed as the initial item selection. This follows previous research
important in choice between owned versus choice between (Chernev 2001; Kardes et al. 1993) that looks at a later
unowned headphones. Second, we further examined stage of the purchasing decision, after consumers have al-
whether egocentric processing drives the advantage of ready narrowed down the choice-set based on initial crite-
person-related features (hypothesis 2) by testing whether ria like price and quality.
the predicted consideration advantage of person-related All participants were then asked to list four features
features is mediated by item inclusion in the category of (other than price, quality, and brand) that are important to
self. Third, we tested the prediction that the effect of self- them in choosing between the headphones. Then, as a de-
cue is moderated by self-focus (hypothesis 3b). If egocen- pendent variable, participants were asked to rate the extent
tric processing drives the effect, then the effect can be to which each feature they listed is applicable to people
expected to be bigger (smaller or even absent) when self- (1 ¼ “Not at all” and 5 ¼ “Very much so”). Among the fea-
focus is high (low). tures participants listed, features such as design,
WEISS 301

portability, and durability were rated as less person-related, FIGURE 5


whereas features such as style, appearance, and reliability
were rated as more person-related. Next, participants an- HIGHER PERSON-RELATEDNESS OF SPONTANEOUSLY
CONSIDERED FEATURES FOR IN-SELF (VS. OUT-SELF)
swered questions to test alternative explanations (regarding HEADPHONES WHEN SELF-CONSCIOUSNESS IS HIGH,
the applicability of pre-selected features to people and fea- STUDY 5
ture intangibility and abstractness). We then measured the
predicted mediator—headphones inclusion in the category
of self—using the items from study 2 (r ¼ .62, M ¼ 2.55, Person-
relatedness of
SD ¼ 1.43). Participants then answered more questions to spontaneously
test alternative explanations (about anthropomorphism). considered
Following Carver and Scheier (1978) we next used the features

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


20
self-consciousness scale (Fenigstein et al. 1975) to capture
participants’ degree of self-focus (1 ¼ “extremely In-Self
uncharacteristic” and 5 ¼ “extremely characteristic”; Out-Self
a ¼ .61, M ¼ 3.58, SD ¼ .48). Participants then responded 12
to the manipulation check from study 3 of the focal self-
cue, ownership (r ¼ .9). Results were consistent with a suc-
cessful manipulation (owned: M ¼ 4.12, SD ¼ 1.87 vs.
unowned: M ¼ 2.66, SD ¼ 1.59, t (353) ¼ 7.97, p < .0001). 4
The participants provided demographic information and 2 3 4 5
Self-Consciousness
were debriefed and thanked.
NOTE.– Feature person-relatedness for in- (vs. out-) self headphones is signifi-
Results. The person-relatedness of the free-listed fea-
cantly higher when self-consciousness is 3.38 or above.
tures was summed into a person-relatedness index (min-
¼ 4, max ¼ 20, M ¼ 8.39, SD ¼ 3.64) and regressed on cue
(out-self ¼ 1, in-self ¼ 1), self-consciousness (standard- Discussion. Using free-listed features, the results of
ized), and their interaction. In line with overall support for study 5 supported the predicted consideration advantage of
hypothesis 1, the results showed that person-relatedness of person-related features in decisions about in-self (vs. out-
features was higher for in-self (vs. out-self) items (b ¼ .63, self) items (hypothesis 1). Using moderation by an individ-
SE ¼ .19, t (351) ¼ 3.34, p ¼ .0009). Self-consciousness ual difference on self-focus (hypothesis 3b) and statistical
was a significantly positive predictor (b ¼ .38, SE ¼ .19, t mediation by inclusion of the item in the category of self,
(351) ¼ 2.03, p ¼ .043). In line with hypothesis 3b, the in- study 5’s results also supported the prediction that the ob-
teraction with self-consciousness was positive and signifi- served effect of cue (in-self vs. out-self) is driven by ego-
cant (b ¼ .56, SE ¼ .19, t (351) ¼ 2.97, p ¼ .0032; figure 5). centric processing (hypothesis 2). Finding a significant
The interaction between self-cue and self-consciousness main effect of self-cue in addition to an interaction with
was analyzed using the Johnson–Neyman floodlight tech- self-consciousness supports our prediction that although
nique (Spiller et al. 2013). In line with hypothesis 3b, the self-focus is a moderator, it is not a necessary condition for
results revealed significantly higher feature person- the advantage of person-related features (see also studies
relatedness (z ¼ 1.96, p ¼ .05) for the in-self (vs. out-self) 1–4). See web appendix G for analyses that rule out the
items for levels of self-consciousness as low as 3.38 (.41 possibility that the results reflect a change in the mere rat-
SDs below the mean of 3.58). ing of (vs. actual) person-relatedness of the listed features
or in feature abstractness/intangibility. Further analyses
Mediation. The results supported the prediction that rule out the perception of the products as people or friends
egocentric processing drives the advantages of person- (vs. egocentric processing) as process (see the results of re-
related features in decisions about in-self (vs. out-self) gression analyses for these alternative accounts in table 4).
items (hypothesis 2) by showing that headphones inclusion
in the category of self-mediated the effect of cue (in-self Study 6: Importance Advantage in the Context of
vs. out-self) on the person-relatedness of the features peo-
ple spontaneously considered important for headphones
Add-Ons—Moderated by the Extent to Which
choice. A bootstrapping analysis with 10,000 samples Self-Cue Is Self-Defining
(PROCESS model 4; Hayes 2013) confirmed the mediation Study 6 further tested the predicted importance advan-
effect (.1180, 95% CI [.0314, .2695]). The results of a re- tage of person-related features (hypothesis 1) in the context
gression analysis for headphones inclusion in the self are of the sale of add-ons (Bertini, Ofek, and Ariely 2009).
shown in table 4. Since add-ons provide benefit only if used with a matching
302 JOURNAL OF CONSUMER RESEARCH

TABLE 4

REGRESSION ANALYSIS FOR COMPETING MEDIATORS, STUDY 5

Headphones in self Headphones as person Headphones as friend

b t b t b t

Self-cue .22* 3.01 .03 .42 .01 0.16


Self-consciousness .22* 2.90 .00 .01 .11 1.49
Self-cue  Self- .08 1.04 .05 .70 .09 1.11
consciousness
*p <.005.

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


base product (Guiltinan 1987), the features of an add-on then asked to enter the maximum dollar amount they would
(e.g., a cellphone case) can be described in reference to ei- pay to purchase this case using an open-ended response
ther the add-on itself (e.g., the case’s own visual appeal) or box (as in study 2). Four Mine-Me scale items then mea-
its matching product (e.g., the visual appeal of the cell- sured participants’ tendency to perceive the self-cue of
phone when using the case). This allowed us to manipulate ownership as self-defining (1 ¼ “strongly disagree” and
self-cue using long-standing ownership (vs. newly formed 7 ¼ “strongly agree”) (a ¼ .91, M ¼ 3.95, SD ¼ 1.55). The
ownership as in studies 3–5). If person-related features items were: “I feel that objects I own are part of my self;”
gain advantages in decisions about in-self (vs. out-self) “When I own a product, I see it as part of me;” “Once I get
items (hypothesis 1; as studies 1–5 show), then relating the to own something, I feel that it is more part of myself;” “If
add-on features to the already owned (in-self) matching I think of the self as a category, things that I own are part
product, rather than to the (out-self) add-on can be of this category, while things that I don’t own are external
expected to increase the importance of person-related fea- to it.” Participants were then debriefed and thanked.
tures in decisions about the add-on.
Results and Discussion. WTP (winsorized at 1%, 99%)
Study 6 also tested the prediction that the effect of self-
was regressed untransformed (skewness ¼ 1.77 < 1.96) on
cue is moderated by whether people view the self-cue as
cue (in-self ¼ 1 vs. out-self ¼ 1), feature type (person
self-defining (hypothesis 3c). If item self-categorization
related ¼ 1 vs. product related ¼ 1), Mine-Me scale
drives the effect, we can expect that the effect will be big-
(standardized), and their interactions. The results showed a
ger (smaller or even absent) for people who see (do not
positive and significant effect of self-cue (b ¼ 1.7,
see) the focal self-cue as central to the way they define
SE ¼ .74, t (365) ¼ 2.3, p ¼ .02) and a negative and signifi-
themselves. Perception of the focal self-cue, ownership, as
cant effect of feature type (b ¼ 1.51, SE ¼ .74, t
self-defining was measured using a “Mine-Me” scale
(365) ¼ 2.04, p ¼ .04). Mine-Me level was a positive and
adapted from prior measures (Weiss and Johar 2013, 2016)
significant predictor (b ¼ 2.23, SE ¼ .74, t (365) ¼ 3.02,
to capture how much participants consider owned items
p ¼ .003); its interaction with self-cue was positive and
part of themselves.
marginally significant (b ¼ 1.42, SE ¼ .74, t (365) ¼ 1.92,
Method. Three hundred and seventy-nine MTurk work- p ¼ .055) and its interaction with feature type was not sig-
ers participated in a short study for a nominal fee. Six nificant (b ¼ .96, SE ¼ .74, t (365) ¼ 1.29, p ¼ .2).
failed an attention check and were excluded from further Consistent with overall support for hypothesis 1, the inter-
analysis, leading to a sample size of 373 participants. The action between self-cue and feature type was positive and
study used a 2 (cue: in-self vs. out-self)  2 (feature type: marginally significant (b ¼ 1.39, SE ¼ .74, t (365) ¼ 1.88,
person-related vs. product-related)  Mine-Me scale p ¼ .06). In line with hypothesis 3c, this two-way interac-
between-subject design. All participants were asked to tion was qualified by Mine-Me level, resulting in a signifi-
imagine looking online for a cell phone case and finding cant three-way interaction (b ¼ 1.92, SE ¼ .74, t
one that seems like a good fit. Self-cue and feature type (365) ¼ 2.6, p ¼ .0097; figure 6).
were both manipulated by modifying the caption that de- Spotlight analysis showed that the interaction between
scribed the case. Participants in the out-self (in-self) item self-cue and feature type was positive and significant at
condition who received the person-related features descrip- high levels of Mine-Me (i.e., at 1 SD above the mean,
tion read: “A case that is (makes your phone) not only re- b ¼ 13.25, SE ¼ 4.18, t (365) ¼ 3.17, p ¼ .0017; the interac-
silient but also beautiful.” For participants who received tion remains significant at p ¼ .05 for participants with lev-
the corresponding product-related features description, the els of Mine-Me as low as 4, or .03 SDs above the mean of
features were replaced with “durable” and “aesthetically 3.95). Planned contrasts showed that focusing the decision
pleasing.” As the dependent variable, participants were on an in-self (vs. out-self) item increased WTP when the
WEISS 303

FIGURE 6

HIGHER WTP FOR A CASE DESCRIBED USING PERSON-RELATED FEATURES WHEN THE FEATURES REFER TO AN IN-SELF (VS.
OUT-SELF) ITEM AND MINE-ME IS HIGH, STUDY 6

WTP for
cellphone case
$40
Out-self In-self

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


$20

$0
Person- Product- Person- Product- Person- Product-
related related related related related related
Low Avg. High
Mine-Me

NOTE.– Low/Avg./High Mine-Me reflects (1 SD below the mean/the mean/1 SD above the mean) of the Mine-Me scale.

features were person-related (in-self: M ¼ $29.63, marginally significant interaction between self-cue and
SE ¼ 2.22 vs. out-self: M ¼ $16.77, SE ¼ 2.01, b ¼ 12.86, feature-type in addition to a three-way interaction with
SE ¼ 2.99, t (365) ¼ 4.29, p < .0001), but not when the fea- Mine-Me level supports our prediction that although the
tures were product-related (in-self: M ¼ $24.11, SE ¼ 2.01 degree to which the self-cue is self-defining is a moderator,
vs. out-self: M ¼ $24.5, SE ¼ 2.12, b ¼ .39, SE ¼ 2.92, t it is not a necessary condition for the effect (see also stud-
(365) ¼ .13, p ¼ .89) at high Mine-Me levels. In line with ies 1–5).
hypothesis 3c, a similar analysis showed no effect at low
Mine-Me levels (i.e., at 1 SD below the mean; ps > .58). Study 7: Importance Advantage in the Context of
Using a different product category and long-standing in- Product Naming—Moderated by the Extent to
stead of newly formed ownership as the in-self cue, the Which Self-Cue Is Self-Defining
results of study 6 further supported the importance advan-
tage of person-related features in decisions about in-self Study 7 further tested the predicted importance advan-
(vs. out-self) items (hypothesis 1). The results of study 6 tage of person-related features (hypothesis 1) and the mod-
also supported the prediction that the effect of self-cue is erating effect of whether the self-cue is self-defining
moderated by whether the focal self-cue is self-defining (hypothesis 3c) in the context of a second marketing appli-
(hypothesis 3c). Finding that the effect of self-cue is bigger cation: product naming. Marketers sometimes give prod-
(smaller or absent) among people for whom ownership is ucts an informative name to convey a differentiating
(is not) self-defining (i.e., those high [low] on the Mine- feature (e.g., Adidas ZX Flux Smooth, Sketchers Burst
Me scale) supports our theory that this effect is driven by Daring). Hypothesis 1 suggests that a product name that
egocentric processing (hypothesis 2). includes a person-related feature would increase the appeal
The results of study 6 suggest that marketers can use of in-self (vs. out-self) products. We tested this prediction
long-standing product ownership to create advantages for in an incentive-compatible study that offered participants a
person-related features and, in doing so, expand the cir- consequential choice between two real flash drives. To
cumstances in which using person-related features creates keep feature relevance high, the actual drives included a
advantages. These results also offer novel theoretical and person-related or a product-related feature (based on pre-
practical insights specifically to the sales of add-ons, a do- testing; appendix) in their names: the “Intelligent Stick”
main with practical importance that has been understudied and the “Compact Attache.” (To reduce unnecessary varia-
in the field of marketing (Bertini et al. 2009). Finding a tion in product name and curb participants’ ability to look
304 JOURNAL OF CONSUMER RESEARCH

up the products online, the second word of both names was FIGURE 7
modified to “Drive.”)
CHOOSING AN ITEM WITH A PERSON-RELATED NAME FROM
Method. Two hundred and eight students in an intro- IN-SELF (VS. OUT-SELF) ITEMS WHEN MINE-ME LEVEL IS
ductory marketing class in a large midwestern university HIGH, STUDY 7
participated in a short online study for course credit. Two
participants failed an attention check and were excluded
from further analysis, leading to a sample size of 206 par- Choice
probability of
ticipants. Participants were presented with an Amazon- item with
style online shopping-cart that included two real flash drive person-related
name
products with slightly modified names: the “Intelligent 0.9

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


Drive” and the “Compact Drive.” To mimic naturalistic
shopping cart settings, participants received the name of
each product and the same three pieces of information
0.5
about each product (color picture, capacity, and shipping
provider), counterbalanced across products.
The study employed a cue (in-self vs. out-self) by Mine- In-Self Out-Self
Me level between-subject design. Self-cue was manipu- 0.1
1 2 3 4 5 6 7
lated between subjects using ownership by asking partici-
Mine-Me Level (measured)
pants to imagine making a choice between the two
products either before (out-self) or after (in-self) purchase NOTE.– Johnson–Neyman region of significance when Mine-Me level is
(as in study 3). In addition to using study 3’s wording, self- greater than 4.27.

cue was also manipulated by varying the top banner of the


cart: participants in the out-self (in-self) items condition level was also positive and significant (b ¼ .32, v2 ¼ 4.43,
saw a standard (modified) top banner that stated “shopping p ¼ .035; figure 7).
cart” (“owned items cart”) and, right below, “items to buy The interaction between the self-cue and Mine-Me level
now” (“items you already acquired”). was analyzed using the Johnson–Neyman floodlight tech-
Participants were then informed that “by responding to nique. In line with hypothesis 3c, results revealed a signifi-
this survey, you have been entered into a lottery. We will cantly greater choice probability of the item with person-
randomly select one tenth of participants in this study, and related name (z ¼ 1.96, p ¼ .05) for the in-self (vs. out-self)
these winners will receive their chosen flash drive.” This items for participants with levels of Mine-Me as low as
rendered their choice incentive compatible. Participants 4.27 (.21 SDs below the mean of 4.5).
then chose the flash drive they wanted to buy (keep). This Using a real (i.e., consequential) choice of actual prod-
choice served as the study’s dependent variable. ucts, a different product category, a familiar choice inter-
Participants then responded to the same Mine-Me scale face, and an implicit indication for the product’s features
from study 6 to measure the extent to which the self-cue of based on the product’s name, the results of study 7 further
ownership was self-defining (a ¼ .79, M ¼ 4.5, SD ¼ 1.08) supported the importance advantage of person-related fea-
and to the ownership manipulation check from study 2 tures in decisions about in-self (vs. out-self) items (hypoth-
(r ¼ .73). Results were consistent with a successful owner- esis 1). The results of study 7 also further supported the
ship manipulation (owned: M ¼ 4.59, SD ¼ 1.17 vs. prediction that the effect of self-cue is moderated by
unowned: M ¼ 2.75, SD ¼ .79, t (204) ¼ 13.1, p < .0001). whether the self-cue is self-defining (hypothesis 3c).
Participants were then debriefed and thanked. Twenty-one Finding that the effect of self-cue is bigger (smaller or ab-
of them were subsequently selected at random and given sent) among people for whom ownership is (is not) self-
the flash drive they had chosen. defining (i.e., people who are high [low] on the Mine-Me
scale) supports our theory that this effect is driven by ego-
Results and Discussion. Participants’ chosen flash centric processing (hypothesis 2). Finding that participants
drive (intelligent drive ¼ 1, compact drive ¼ 0) was entered were more likely to choose the product with a person-
into a logistic regression with cue (out-self ¼ 1, in-self- related feature in its name when the choice was between
¼ 1), Mine-Me level (standardized), and their two-way in- in-self (vs. between out-self) items in a real (i.e., conse-
teraction as predictors. Mine-Me level was not a quential) choice between real products further supports the
statistically significant predictor (b ¼ .07, v2 ¼ .18, p ¼ .7). generalizability and robustness of our effects. Finding a
Consistent with overall support for hypothesis 1, an in-self significant main effect of self-cue in addition to an interac-
(out-self) cue significantly increased choice of the intelli- tion with Mine-Me levels suggests that whether the self-
gent drive (b ¼ .36, v2 ¼ 6.01, p ¼ .014). Moreover, in line cue is self-defining is a moderator but not a necessary con-
with hypothesis 3c, the interaction effect with Mine-Me dition for the effect (see also studies 1–6).
WEISS 305

GENERAL DISCUSSION faster response to the same features when judging the self
(study 4) and that (ii) an in-self (vs. out-self) cue does not
This article departs from prior research by showing that favor person-related features for less self-focused people
the way consumers classify an item relative to themselves (who are less likely to use the self as a category for objects;
can create advantages for a subset of the item’s features. study 5). Mediation analysis showed that making decisions
We test and support the prediction that egocentric process- about in-self (vs. out-self) items increases use of the self as
ing drives the advantages of person-related features in a category for objects which, in-turn, predicts more choice
decisions about in-self (vs. out-self) items. In egocentric of items with better rated person-related features (study 3).
processing, people use self-cues such as ownership and Various steps ruled out feature relevance to the product
brand to classify items as being inside or outside of the cat- category or feature familiarity as confounding factors in
egory of self; they then use this category, to which person- the observed effects (web appendix F).

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


related features neatly apply, to process information about
in-self items. Person-related features thus gain advantages Contributions
in consideration, processing, and importance in decisions
about in-self (vs. out-self) items. The advantages are bigger This article integrates insights from a variety of bodies
when similar advantages exist in self-judgment, when con- of research on the relationships between consumers and
sumers are self-focused, and when the self-cue is self- products. We build on and further substantiate prior work
defining; the advantages are present at an average level of on egocentric categorization, and specifically its central as-
each moderator and are smaller or even absent when these sertion that people use the self as an organizing category
moderators are low. for objects (Weiss and Johar 2013, 2016). In line with prior
research (Dagogo-Jack and Forehand 2018; Weiss and
Johar 2013), we implicate the underlying role of the self by
Main Findings using established or adapted mediators (item inclusion in
The results obtained from seven studies are in line with the category of self, studies 2 and 5; use of self as an orga-
the hypothesis that person-related features gain advantages nizing category for objects, study 3) and moderators (self-
in decisions about in-self (vs. out-self) items (hypothesis focus, study 5; Mine-Me, studies 6 and 7). We also identify
1). The results were replicated across experimental para- egocentric processing as a distinct component of egocen-
digms, different advantages, multiple advantage measures, tric categorization and uniquely study its implications for
diverse product categories, and two self-cues. Within the the consideration, processing, and importance of a subset
studies employing branding as a self-cue, the results held of product features.
across different operationalizations of connection to a Importantly, the present research also extends the ego-
brand. Within the studies employing ownership as self-cue, centric categorization framework, which initially conceptu-
the results held across different ownership manipulations. alized ownership as the only criterion for self-
The results were robust across different levels of interac- categorization (Johar, Chung, and Weiss 2019; Weiss and
tion with the target item. Johar 2013, 2016, 2018). Generalizing across two different
Process measures and theory-driven moderators sup- product circumstances that may lead to the classification of
ported our theorized process. We provided two lines of evi- items relative to the self (ownership and brands), we intro-
dence to support our prediction that the greater advantages duce self-cue as an organizing concept for such circum-
of person-related features for in-self (vs. out-self) items are stances. This allows us to integrate insights from different
driven by egocentric processing (hypothesis 2): (i) item in- literatures and improves our ability to understand and pre-
clusion in the category of self and (ii) use of self as the fo- dict how people will classify items in relation to them-
cal category for objects. Item inclusion in the category of selves in a variety of situations. The various self-cues
self was evident from finding that an in-self (vs. out-self) consumers use and their effects, the ways consumers re-
cue does not favor person-related features among people solve inconsistencies across self-cues (e.g., an unowned
who do not perceive the focal self-cue as self-defining (and product [out cue] of an in-group brand [in cue]), and other
thus are unlikely to categorize items with an in-self cue in unexplored implications of self-categorization cues are
the self; studies 6 and 7). Further, mediation analysis worthwhile directions for future research.
showed that an in-self (vs. out-self) cue increases item in- There are competing theories about how the endowment
clusion in the category of self which, in-turn, increases effect works (Morewedge and Giblin 2015;
both WTP for a product that is described using person- Nayakankuppam and Mishra 2005), but they all agree that
related features (study 2) and the person-relatedness of the ownership increases the accessibility of all the positive and
features considered important in item choice (study 5). Use desirable attributes of a product equally. Presenting a more
of self as a category for objects was evident from finding nuanced theory, we show that the enhancing effect of own-
that (i) faster response to person-related features when ership is seen for a specific subset of the item’s features—
judging an in-self (vs. out-self) item was predicted by its person-related features. This finding advances our
306 JOURNAL OF CONSUMER RESEARCH

ability to predict when the known effects of ownership on consumer already owns (e.g., a voice activated speaker that
valuation and choice will strengthen (or weaken). makes your home smart vs. one that is smart on its own).
One factor that may explain why our findings differ Our research also suggests that an on-line retailer can im-
from the endowment effect literature is the choice context. prove the appeal of a high-margin product by describing it
Many researchers have explained the endowment effect by differently to consumers at different stages of their purchase
positing a status quo bias (Thaler 1980), whereby people decision, emphasizing product-related features in the pre-
try to reaffirm the existing state of being owners purchase stage (through product information available on-
(Morewedge and Giblin 2015) by seeking, attending to, line), but emphasizing person-related features in the post-
and remembering data about the positive aspects of an purchase stage (through product information attached to the
owned product (Johnson, Haubl, and Keinan 2007). Yet purchased product) so as to appeal specifically to product
some of the choices we examine in the present research owners. Marketers may similarly benefit from highlighting

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


have no status quo option: choices about which of two person- (vs. product-)related features when they engage in
items one owns to keep or choices about which of two marketing practices that induce consumers to feel product
items one does not own to buy (studies 3, 5, and 7). In con- ownership before purchase, such as encouraging customers
trast to endowment choices about whether to own an item, to customize or test-drive products prior to purchase (Peck
these choices are about which item to own and therefore and Shu 2009). BMW may thus benefit from describing an
have no status-quo. Unlike the endowment literature, our 8 Series car as “smooth” in ads that encourage people to
egocentric processing framework is well positioned to offer visit a dealership for the first time, but as “intuitive” in
predictions about such product decisions. materials given to consumers after a test drive.
In recent decades, with product return polices becoming Notably, future research and market applications of our
more lenient and the rise of online shopping (Bernard 2010), typology should keep some limitations and unanswered
product decisions where there is no status-quo option are in- questions in mind. The extent to which a feature is perceived
creasingly relevant, as consumers choose which of two (or to describe products versus persons is likely dynamic, and
more) products bought online to keep and which to return, for may vary across contexts. To use the distinction between
example. To the best of our knowledge, this article is the first product- and person-related features effectively, researchers
to compare and contrast choices between owned and between and marketers should test how features are perceived among
unowned items. We believe that in the current consumer land- their target population close to the time of use. Our pretests
scape this can be a fruitful direction for future research. provide a template for how this might be done. We have not
Prior brand research has shown that connection to a brand tested whether the effect we observe generalizes to negative
(Dommer and Swaminathan 2013) and the specific features features, like those used in product testimonials or reviews.
of a brand (Chan, Berger, and Van Boven 2012) both affect This area for future research will likely be informative, as
preference for the brand’s products. Our categorization per- greater attention to or slower processing of negative (vs.
spective allows us to look for the first time at the relation- positive) information (Unkelbach et al. 2008) may alter or
ship between these two aspects of branding. Specifically, we even eliminate the effect for negative features.
argue that the type of feature used to describe the brand acts The insights from this research may have more wide-
as a moderator for the effect of brand-self link on prefer- ranging implications. The decisions about in-self items fol-
ence. We can thus ask whether the greater preference for an lowing ownership in studies 3, 5, and 7 of this article are
in-self (vs. out-self) item is seen more when the item is de- characterized by a one-time choice that leads consumers to
scribed in terms of one type of feature versus another. keep their selected option (e.g., one of two sets of head-
Relatedly, while brand personality research views, for exam- phones bought online) but disown (send back) the unse-
ple, honest and genuine as two facets of the personality di- lected option. However, the scope of decisions among in-
mension “sincerity” (Aaker 1997), we show that different self items following ownership can be broadened to in-
labels for qualitatively similar attributes can have meaning- clude repeated choices among redundant products that the
ful and systematic implications for product preference. chooser owns, such as which of the multiple jackets, bags,
or electronic devices a consumer owns to use on a given
day. Such decisions are made repeatedly, often on a daily
Implications
basis, and typically are not followed by disowning the un-
Our research suggests various ways that brands and mar- selected items. Such repeated decisions are consequential,
keters can tailor their message to specific consumer seg- as they are likely to affect consumers’ product replacement
ments. As discussed earlier, a brand can benefit from using choices (Okada 2001). Further, some products, such as
person-related features to describe its high margin products payment devices (e.g., credit cards), often employ per-use
to its fans, as Apple did in its iPhone launch and we did in pricing, whereby marketers rely on the device to be repeat-
our Facebook field study. Person-related features can even edly selected over competing devices in order to be profit-
increase product appeal to nonfans, for example, by describ- able. Does the increased importance of person-related
ing the features in relation to a complementary item that the features, as identified here in the context of a narrowly
WEISS 307

defined range of choices among in-self items, also apply to 2019. Data for an additional study reported in web appen-
broadly defined, repeated choices among in-self items? dix C were collected on Facebook in the spring of 2021.
This and other related questions about the implications of Data for studies 2 (spring 2019), 3 (fall 2017), 4 (fall
egocentric processing await further research. 2015), 5 (fall 2019), and 7 (spring 2013 and fall 2014)
were collected using a student panel at the University of
DATA COLLECTION INFORMATION Wisconsin–Madison. Data for study 6 were collected using
Amazon Mechanical Turk in the spring of 2019. The au-
The author collected data (studies 1, 2, 3, 5, 6, and 7) or thor analyzed the data for all studies. The data are currently
supervised the collection of data (study 4) for all studies. stored in a project directory on the Open Science
Data for study 1 were collected on Facebook in the fall of Framework.

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


308 JOURNAL OF CONSUMER RESEARCH

APPENDIX: FEATURE APPLICABILITY TO PEOPLE VERSUS PRODUCTS

Original Honest*
Steady Sincere*
Unique Cheerful*
Slim Intelligent*
Loud Responsible*
Rugged Friendly*

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


Authenc Spirited*
Complex Daring*
Stable Creave*
Handy Smart*
Thin Romanc*
Stylish Charming*
Weighty* Imaginave*
Efficient* Genuine*
Dazzling* Intuive*
Effecve* Outdoorsy*
Versale* Glamourous*
Delicate* Pragmac*
Speedy* Resilient*
Praccal* Strong*
Smooth* Beauful*
Useful* Adaptable*
Adjustable* Elegant*
Precise* Quick*
Thick* Slender*
Aesthec* Responsive*
Sturdy* Dependable*
Durable* Innovave
Convenient* Tough*
Compact* Flexible
Programmable* Reliable
Portable* Handy

1 2 3 4 5 1 2 3 4 5

Applicable to describe a product Applicable to describe a product


Applicable to describe a person Applicable to describe a person

Note: See web appendix A for details.


*Statistically significant difference between the feature’s applicability to describe people and products (p  .05).
WEISS 309

and Uniqueness Motives in Choice,” Journal of Consumer


Research, 39 (3), 561–73.
REFERENCES Chandler, Jesse and Norbert Schwarz (2010), “Use does not Wear
Aaker, Jennifer L. (1997), “Dimensions of Brand Personality,” Ragged the Fabric of Friendship: Thinking of Objects as
Journal of Marketing Research, 34 (3), 347–56. Alive Makes People Less Willing to Replace Them,” Journal
Aggarwal, Pankaj and Ann L. McGill (2007), “Is that Car Smiling of Consumer Psychology, 20 (2), 138–45.
at Me? Schema Congruity as a Basis for Evaluating Chernev, Alexander (2001), “The Impact of Common Features on
Anthropomorphized Products,” Journal of Consumer Consumer Preferences: A Case of Confirmatory Reasoning,”
Research, 34 (4), 468–79. Journal of Consumer Research, 27 (4), 475–88.
Ahuvia, Aaron C. (2005), “Beyond the Extended Self: Loved Cunningham, Sheila J., David J. Turk, Lynda M. Macdonald, and
Objects and Consumers’ Identity Narratives,” Journal of C. Neil Macrae (2008), “Yours or Mine? Ownership and
Consumer Research, 32 (1), 171–84. Memory,” Consciousness and Cognition, 17 (1), 312–18.

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


Ariely, Dan and Itamar Simonson (2003), “Buying, Bidding, Dagogo-Jack, Sokiente W. and Mark R. Forehand (2018),
Playing, or Competing? Value Assessment and Decision “Egocentric Improvement Evaluations: Change in the Self as
Dynamics in Online Auctions,” Journal of Consumer an Anchor for Brand Improvement Judgments,” Journal of
Psychology, 13 (1), 113–23. Marketing Research, 55 (6), 934–50.
Atasoy, Ozgun and Carey K. Morewedge (2018), “Digital Goods Dommer, Sara L. and Vanitha Swaminathan (2013), “Explaining
Are Valued Less than Physical Goods,” Journal of Consumer the Endowment Effect through Ownership: The Role of
Research, 44 (6), 1343–57. Identity, Gender, and Self-Threat,” Journal of Consumer
Bargh, John A. and Felicia Pratto (1986), “Individual Construct Research, 39 (5), 1034–50.
Accessibility and Perceptual Selection,” Journal of Dunning, David and Andrew F. Hayes (1996), “Evidence for
Experimental Social Psychology, 22 (4), 293–311. Egocentric Comparison in Social Judgment,” Journal of
Barton, Michelle E. and Lloyd K. Komatsu (1989), “Defining Personality and Social Psychology, 71 (2), 213–29.
Features of Natural Kinds and Artifacts,” Journal of Duval, T. Shelly and Robert A. Wicklund. (1972), A Theory of
Psycholinguistic Research, 18 (5), 433–47. Objective Self-Awareness, New York: Academic Press.
Beggan, James K. (1992), “On the Social Nature of Nonsocial Escalas, Jennifer Edson and James R. Bettman (2003), “You are
Perception—The Mere Ownership Effect,” Journal of What They Eat: The Influence of Reference Groups on
Personality and Social Psychology, 62 (2), 229–37. Consumers’ Connections to Brands,” Journal of Consumer
Belk, Russell W. (1988), “Possessions and the Extended Self,” Psychology, 13 (3), 339–48.
Journal of Consumer Research, 15 (2), 139–68. Fenigstein, Allan, Michael F. Scheier, and Arnold H. Buss (1975),
Bernard, Tara S. (2010), “Naughty and Nice’ Retailer Return “Public and Private Self-Consciousness: Assessment and
Policies,” The New York Times, December 1. Theory,” Journal of Consulting and Clinical Psychology, 43
Bertini, Marco, Elie Ofek, and Dan Ariely (2009), “The Impact of (4), 522–27.
Add-on Features on Consumer Product Evaluations,” Journal Ferraro, Rosellina., Jennifer E. Escalas, and James R. Bettman
of Consumer Research, 36 (1), 17–28. (2011), “Our Possessions, Our Selves: Domains of Self-
Bodenhausen, Galen V. and Meryl Lichtenstein (1987), “Social Worth and the Possession-Self Link,” Journal of Consumer
Stereotypes and Information-Processing Strategies: The
Psychology, 21 (2), 169–77.
Impact of Task Complexity,” Journal of Personality and
Fitzsimons, Gavan J. (2008), “Death to Dichotomizing,” Journal
Social Psychology, 52 (5), 871–80.
of Consumer Research, 35 (1), 5–8.
Brauer, Markus and John J. Curtin (2018), “Linear Mixed-Effects
Gawronski, bertram, Galen V. Bodenhausen, and Andrew P.
Models and the Analysis of Nonindependent Data: A Unified
Becker (2007), “I like It, Because I like Myself: Associative
Framework to Analyze Categorical and Continuous
Independent Variables that Vary Within-Subjects and/or Self-Anchoring and Post-Decisional Change of Implicit
Within-Items,” Psychological Methods, 23 (3), 389–411. Evaluations,” Journal of Experimental Social Psychology, 43
Brewer, Marilynn B. (1991), “The Social Self—On being the (2), 221–32.
Same and Different at the Same Time,” Personality and Goffman, Erving (1972), Relations in Public, New York: Harper
Social Psychology Bulletin, 17 (5), 475–82. and Row.
Burris, Christopher T. and Nyla R. Branscombe (2005), “Distorted Grohmann, Bianca (2009), “Gender Dimensions of Brand
Distance Estimation Induced by a Self-Relevant National Personality,” Journal of Marketing Research, 46 (1), 105–19.
Boundary,” Journal of Experimental Social Psychology, 41 Guiltinan, Joseph P. (1987), “The Price Bundling of Services: A
(3), 305–12. Normative Framework,” Journal of Marketing, 51 (2),
Carpenter, Gregory S., Rashi Glazer, and Kent Nakamoto (1994), 74–85.
“Meaningful Brands from Meaningless Differentiation: The Hayes, Andrew F. (2013), Introduction to Mediation, Moderation,
Dependence on Irrelevant Attributes,” Journal of Marketing and Conditional Process Analysis, New York: Guilford.
Research, 31 (3), 339–50. Higgins, E. Tory (1996), “Knowledge Activation: Accessibility,
Carver, Charles S. and Michael F. Scheier (1978), “Self-Focusing Applicability and Salience,” in Social Psychology:
Effects of Dispositional Self-Consciousness, Mirror Handbook of Basic Principles, ed. E. Tory Higgins and Arie
Presence, and Audience Presence,” Journal of Personality W. Kruglanski, New York: Guilford Press, 133–68.
and Social Psychology, 36 (3), 324–32. Irvine, Mark (2020), “Facebook Ad Benchmarks for Your
Chan, Cindy, Jonah Berger, and Leaf Van Boven (2012), Industry [Data],” https://www.wordstream.com/blog/ws/
“Identifiable but Not Identical: Combining Social Identity 2017/02/28/facebook-advertising-benchmarks.
310 JOURNAL OF CONSUMER RESEARCH

Johar, Gita V., Jaideep Sengupta, and Jennifer L. Aaker (2005), Park, C. Whan, Deborah J. MacInnis, and Joseph Priester (2006),
“Two Roads to Updating Brand Personality Impressions: “Brand Attachment: Constructs, Consequences, and Causes,”
R
Trait versus Evaluative Inferencing,” Journal of Marketing Foundations and TrendsV in Marketing, 1 (3), 191–230.
Research, 42 (4), 458–69. Peck, Joann and Suzanne B. Shu (2009), “The Effect of Mere
Johar, Gita Venkataramani, Jaeyeon Chung, and Liad Weiss Touch on Perceived Ownership,” Journal of Consumer
(2019), “Ownership and Identity: A Cognitive Perspective,” Research, 36 (3), 434–47.
in Handbook of Research on Identity Theory in Marketing, Pier, Elizabeth L., Markus Brauer, Amarette Filut, Anna Kaatz,
ed. Americus Reed and Mark Forehand, Cheltenham: Joshua Raclaw, Mitchell J. Nathan, Cecilia E. Ford, and
Edward Elgar Publishing, 141–57. Molly Carnes (2018), “Low Agreement among Reviewers
Johnson, Eric J., Gerald Haubl, and Anat Keinan (2007), “Aspects Evaluating the Same Nih Grant Applications,” Proceedings
of Endowment: A Query Theory of Value Construction,” of the National Academy of Sciences of the United States of
Journal of Experimental Psychology. Learning, Memory, and America, 115 (12), 2952–57.
Cognition, 33 (3), 461–74.

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


Pierce, Jon L., Tatiana Kostova, and Kurt T. Dirks (2003), “The
Kardes, Frank R., Gurumurthy Kalyanaram, Murali State of Psychological Ownership: Integrating and Extending
Chandrashekaran, and Ronald J. Dornoff (1993), “Brand a Century of Research,” Review of General Psychology, 7
Retrieval, Consideration Set Composition, Consumer (1), 84–107.
Choice, and the Pioneering Advantage,” Journal of Rochberg-Halton, Eugene (1984), “Object Relations, Role
Consumer Research, 20 (1), 62–75. Models, and Cultivation of the Self,” Environment and
Kirmani, Amna (2009), “The Self and the Brand,” Journal of Behavior, 16 (3), 335–68.
Consumer Psychology, 19 (3), 271–75. Rogers, Timothy B., Nicholas A. Kuiper, and W. S. Kirker (1977),
Luhtanen, Riia and Jennifer Crocker (1992), “A Collective Self- “Self-Reference and Encoding of Personal Information,”
Esteem Scale: Self-Evaluation of One’s Social Identity,” Journal of Personality and Social Psychology, 35 (9),
Personality and Social Psychology Bulletin, 18 (3), 302–18. 677–88.
McConnell, Allen R., Christina M. Brown, Tonya M. Shoda, Laura E. Sirgy, M. Joseph (1982), “Self-Concept in Consumer Behavior: A
Stayton, and Colleen E. Martin (2011), “Friends with Benefits: Critical Review,” Journal of Consumer Research, 9 (3),
On the Positive Consequences of Pet Ownership,” Journal of 287–300.
Personality and Social Psychology, 101 (6), 1239–52. Spiller, Stephen A., Gavan J. Fitzsimons, John G. Lynch, and
MacInnis, Deborah J. and Valerie S. Folkes (2017), “Humanizing
Gary H. McClelland (2013), “Spotlights, Floodlights, and the
Brands: When Brands Seem to be like Me, Part of Me, and in
Magic Number Zero: Simple Effects Tests in Moderated
a Relationship with Me,” Journal of Consumer Psychology,
Regression,” Journal of Marketing Research, 50 (2), 277–88.
27 (3), 355–74.
Srull, Thomas K. and Robert S. Wyer Jr. (1979), “Role of
Medin, Douglas L. (1989), “Concepts and Conceptual Structure,”
Category Accessibility in the Interpretation of Information
The American Psychologist, 44 (12), 1469–81.
Meyvis, Tom and Stijn M. J. Van Osselaer (2018), “Increasing the about Persons—Some Determinants and Implications,”
Power of Your Study by Increasing the Effect Size,” Journal Journal of Personality and Social Psychology, 37 (10),
of Consumer Research, 44 (5), 1157–73. 1660–72.
Miller, Dale T. and William Turnbull (1986), “Expectancies and Strahilevitz, Michal A. and George Loewenstein (1998), “The
Interpersonal Processes,” Annual Review of Psychology, 37 Effect of Ownership History on the Valuation of Objects,”
(1), 233–56. Journal of Consumer Research, 25 (3), 276–89.
Mitchell, Jason P., Todd F. Heatherton, and C. Neil Macrae Symons, Cynthia S. and Blair T. Johnson (1997), “The Self-
(2002), “Distinct Neural Systems Subserve Person and Reference Effect in Memory: A Meta-Analysis,”
Object Knowledge,” Proceedings of the National Academy Psychological Bulletin, 121 (3), 371–94.
of Sciences of the United States of America, 99 (23), Tajfel, Henri (1969), “Cognitive Aspects of Prejudice,” Journal of
15238–43. Social Issues, 25 (4), 79–97.
Mitchell, Jason P., C. Neil Macrae, and Mahzarin R. Banaji Thaler, Richard (1980), “Toward a Positive Theory of Consumer
(2005), “Forming Impressions of People versus Inanimate Choice,” Journal of Economic Behavior and Organization, 1
Objects: Social-Cognitive Processing in the Medial (1), 39–60.
Prefrontal Cortex,” NeuroImage, 26 (1), 251–57. Trope, Yaacov and Nira Liberman (2003), “Temporal Construal,”
Morewedge, Carey K. and Colleen E. Giblin (2015), Psychological Review, 110 (3), 403–21.
“Explanations of the Endowment Effect: An Integrative Trudel, Remi, Jennifer J. Argo, and Matthew D. Meng (2016),
Review,” Trends in Cognitive Sciences, 19 (6), 339–48. “The Recycled Self: Consumers’ Disposal Decisions of
Mussweiler, Thomas and Galen V. Bodenhausen (2002), “I Know Identity-Linked Products,” Journal of Consumer Research,
You Are, but What Am I? Self-Evaluative Consequences of 43 (2), 246–64.
Judging in-Group and out-Group Members,” Journal of Turk, David J., Kim van Bussel, Gordon D. Waiter, and C. Neil
Personality and Social Psychology, 82 (1), 19–32. Macrae (2011), “Mine and Me: Exploring the Neural Basis of
Nayakankuppam, Dhananjay and Himanshu Mishra (2005), “The Object Ownership,” Journal of Cognitive Neuroscience, 23
Endowment Effect: Rose-Tinted and Dark-Tinted Glasses,” (11), 3657–68.
Journal of Consumer Research, 32 (3), 390–95. Unkelbach, Christian, Klaus Fiedler, Myriam Bayer, Martin
Okada, Erica Mina (2001), “Trade-Ins, Mental Accounting, and Stegmüller, and Daniel Danner (2008), “Why Positive
Product Replacement Decisions,” Journal of Consumer Information is Processed Faster: The Density Hypothesis,”
Research, 27 (4), 433–46. Journal of Personality and Social Psychology, 95 (1), 36–49.
WEISS 311

Weiss, Liad and Gita V. Johar (2013), “Egocentric Categorization ——— (2018), “Psychological Ownership in Egocentric
and Product Judgment: Seeing Your Traits in What You Own Categorization Theory,” in Psychological Ownership and
(and Their opposite in What You Don’t),” Journal of Consumer Behavior, ed. Joann Peck and Suzanne Shu,
Consumer Research, 40 (1), 185–201. Cham: Springer, 33–51.
——— (2016), “Products as Self-Evaluation Standards: When White, Katherine and Darren W. Dahl (2007), “Are All out-
Owned and Unowned Products Have opposite Effects on Groups Created Equal? Consumer Identity and
Self-Judgment,” Journal of Consumer Research, 42 (6), Dissociative Influence,” Journal of Consumer Research,
915–30. 34 (4), 525–36.

Downloaded from https://academic.oup.com/jcr/article/49/2/288/6460825 by Jnls Cust Serv on 27 July 2022


© 2022 Journal of Consumer Research Inc. Copyright of Journal of Consumer Research is the
property of Oxford University Press / USA and its content may not be copied or emailed to
multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.

You might also like