Individual Differences in Haptic Information Processing: The “Need for Touch” Scale

Author(s): Joann Peck and Terry L. Childers

Source: Journal of Consumer Research , Vol. 30, No. 3 (December 2003), pp. 430-442
Published by: Oxford University Press
Stable URL: https://www.jstor.org/stable/10.1086/378619

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Consumer Research

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
Individual Differences in Haptic Information
Processing: The “Need for Touch” Scale
JOANN PECK
TERRY L. CHILDERS*

This research details the development of the “Need for Touch” (NFT) scale de-
signed to measure individual differences in preference for haptic (touch) infor-
mation. The 12-item NFT scale consists of autotelic and instrumental dimensions.
Results are reported that support the scale’s hypothesized internal structure as
well as its reliability, convergent, discriminant, and nomological validity. Individual
differences in chronic accessibility to haptic information across groups varying in
NFT were also found in two experiments. Additionally, NFT moderated the rela-
tionship between direct experience and confidence in judgment.

F or centuries, people have recognized the importance of

the sense of touch. Aristotle believed that touch me-
diates every type of sense perception, even vision (Siegel
Internet may affect the way consumers search for infor-
mation and their subsequent decision making. We argue that
the nature and use of touch can affect these aspects of online
1970). Closer to marketing, interpersonal touch has been shopping behavior. In addition, studying touch may lead to
found to affect both attitudes and behavior and Hornik insights regarding brand judgments and choice preferences,
(1992) has called for more research on the role of touch. In leisure satisfaction, information search, and product attribute
consumer behavior, evidence has been found for individual importance as well as the appreciation and acquisition of
differences in terms of preference for sensory forms of in- treasured possessions. For instance, individuals’ confidence
formation (see Heckler, Childers, and Houston 1993 for a in product judgments may be affected by whether or not
discussion of visual versus verbal information processing). they can touch a product during evaluation. Attitude toward
Casual observation reveals that individuals differ greatly in a product may also differ depending on whether a shopper
the amount of touch they exhibit while shopping. Whereas has the opportunity to touch a product and experience pleas-
some consumers touch products to simply place them in urable sensory feedback (e.g., rub a soft leather coat) before
shopping carts, other consumers spend more time exploring purchase. How consumption environments structure and en-
products with their hands before ultimately making a pur- able the acquisition and utilization of haptic information, or
chase decision. It seems likely that some individuals would prohibit it, may in turn lead to the differential utilization of
prefer information available through the sense of touch. To available haptic attributes.1 Some consumers are likely to
date, however, the preference for information obtained become frustrated by their inability to acquire this infor-
through the sense of touch has not been explored. mation, causing them to forgo certain nontouch shopping
With the growth of various forms of nontouch media (e.g., environments (e.g., online shopping). Thus, assessing the
catalog and Internet shopping), this individual difference is differential role of haptic information among consumers can
important to conceptualize and measure. Burke (1997) notes contribute to a better understanding of consumer behavior
that shopping from home often does not provide the same across a broad range of domains.
level of product information or entertainment as physical As a first step in exploring haptic information processing,
stores. Peterson, Balasubramanian, and Bronnenberg (1997) this article develops and validates a measure of this indi-
emphasize that information-presentation mechanisms on the vidual difference in the “Need for Touch” (NFT). In the
following sections we elaborate on the nature of touch and
*Joann Peck is assistant professor of marketing at the University of haptic information and develop a set of hypotheses. These
Wisconsin—Madison, 3114 Grainger Hall, 975 University Avenue, Mad- hypotheses are tested by conducting seven studies. In the
ison, WI 53706-1323; email: [email protected]; Terry L. Childers is the
Gatton Endowed Chair in Electronic Marketing, Gatton College of Business 1
and Economics, University of Kentucky, Lexington, KY 40506-0034; Gibson (1966) adopted the term “haptics” to refer to the functionally
email: [email protected]. This research was funded, in part, by a grant discrete system involved in the seeking and the pickup of information by
to the first author from the Carlson School of Management at the University the hand. Touching with the hands (or the haptic system) has been reported
of Minnesota. The authors would like to thank the JCR editor, the associate to be particularly adept at encoding the material object properties corre-
editor, and the three reviewers for their helpful and constructive comments. sponding to texture, hardness, temperature, and weight information
(Klatzky and Lederman 1992, 1993; Lederman and Klatzky 1987).

430

䉷 2003 by JOURNAL OF CONSUMER RESEARCH, Inc. ● Vol. 30 ● December 2003

All rights reserved. 0093-5301/2004/3003-0008$10.00

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 431

final section we discuss the implications of our findings mension of NFT refers to those aspects of prepurchase touch
about this individual difference in the NFT for the under- that reflect outcome-directed touch with a salient purchase
standing and study of consumer behavior. goal. Contained within the domain of this form of touch are
goal-driven evaluative outcomes related to the consumer
(e.g., comfort and certainty in their judgments) as well as
TOUCH AND HAPTIC INFORMATION the target product (e.g., quality or worth). These instru-
mental judgments would be expected to focus on properties
The Motivation Component: NFT. The NFT is con- attuned to haptic utilization that reflect a product’s texture,
ceptually defined as a preference for the extraction and util- hardness, temperature, or weight. The image of the con-
ization of information obtained through the haptic system. sumer involved in instrumental touch is that of a problem
As a preference, NFT is based on motivational versus ability solver consciously engaged in the goal-directed activities of
differences among individuals (Johansson 1978; Spreen and searching for information and arriving at a final product
Strauss 1991). This need to examine products haptically can judgment. For instance, picking up a notebook computer
be driven by motivations associated with what Holbrook and holding it in one’s hand to assess its weight and deriving
and Hirschman (1982) describe in terms of either consumer an inference with respect to its portability would be an ex-
problem solvers or consumers seeking fun, fantasy, arousal, ample of an instrumentally driven haptic product evaluation.
sensory stimulation, and enjoyment. This dichotomy has NFT: The Autotelic Factor. In contrast, the autotelic
been represented in the retail context by the themes of shop- dimension of NFT relates to touch as an end in and of itself.
ping as work (Sherry, McGrath, and Levy 1993) versus the Autotelic touch involves a hedonic-oriented response seek-
festive perspective of shopping as fun (Babin, Darden, and ing fun, arousal, sensory stimulation, and enjoyment (Hol-
Griffin 1994; Sherry 1990). In the utilitarian view, consum- brook and Hirschman 1982). Autotelic touch corresponds
ers are concerned with purchasing products in an efficient to the sensory aspects of product touch, with no purchase
and timely manner to achieve their goals with a minimum goal necessarily salient, but with spontaneous investigation
of irritation. In contrast, as one consumer noted, “I enjoy of multisensory psychophysical product relationships (Hol-
looking around and imagining what one day, I would ac- brook and Hirschman 1982). Thus, central to defining the
tually have money to buy. Shopping . . . is an adventure” domain of autotelic touch are its hedonics (e.g., enjoyment
(Sherry 1990, p. 27). This adventure reflects shopping’s and affect) and the compulsive or irresistible need to engage
potential entertainment value and the enjoyment that is part in exploratory variety seeking via touch (e.g., lack of control
of the experience versus the achievement of any prespecified and indiscriminate processing). Evidence for the apprecia-
end goal (Holbrook and Hirschman 1982). This dichotomy tion of this experiential aspect of consumer behavior is found
is also consistent with the psychological literature on in- in museums that offer multisensory environments including
dividual differences in motivation relating to, for example, music and hands-on displays of sculpture (Fiore, Moreno,
the need to achieve or the need for power (McClelland, and Kimle 1996).
Koestner, and Weinberger 1989). To operationalize this individual difference in NFT, we
As with prior research on scales that assess individual employed scale-development procedures following the
needs (e.g., Need for Cognition [NFC; Cacioppo and Petty guidelines suggested by Churchill (1979) and Anderson and
1982] and Need to Evaluate [NTE; Jarvis and Petty 1996]) Gerbing (1988). A series of studies are reported that assess
we do not assume a biological basis for this individual dif- the scale’s internal structure, reliability, and relationship to
ference, but NFT is consistent with the implicit versus self- a variety of consumer-behavior-related constructs. We then
attributed dual motivation model advocated by McClelland conduct two additional studies designed to provide a better
et al. (1989). In this dichotomy, self-attributed motives cor- understanding of how differences in chronic accessibility to
responding to the instrumental dimension of NFT are char- haptic information relate to preferences in the acquisition
acterized by organized analytic thought that is initiated by and utilization of haptic information. Following this, we
an explicit goal that drives behavior. In contrast, more im- report on an experiment where NFT is shown to moderate
plicit motives represented by autotelic touch reflect com- the relationship between direct experience and confidence
pulsive and affective themes intrinsic to an activity that are in judgment.
not elicited by reference to unmet goals. These differences
are similar to distinctions made between the conscious goal-
setting nature of episodic driven motives versus those de- DEVELOPMENT OF THE NFT SCALE
rived from semantic memory, which more automatically
influence behavior without conscious effort (McClelland et Generation of Items
al. 1989). This dual characterization of NFT from both the
Undergraduate students (n p 135) were provided with
retail as well as the psychological literature on motivations
conceptual definitions of the NFT construct and its dimen-
is consistent with our perspective on the NFT as a multi-
sions and were asked to submit items relating to NFT, similar
dimensional construct with two underlying factors, instru-
to the process utilized by Richins and Dawson (1992). The
mental and autotelic touch, as elaborated on next.
authors then edited the items for appropriate wording and
NFT: The Instrumental Factor. The instrumental di- deleted very similar items, resulting in a 50-item scale.

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
432 JOURNAL OF CONSUMER RESEARCH

EXHIBIT 1

THE TWO DIMENSIONS OF NEED FOR TOUCH AND THE SCALE ITEMS

1. When walking through stores, I can’t help touching all kinds of products. (A)
2. Touching products can be fun. (A)
3. I place more trust in products that can be touched before purchase. (I)
4. I feel more comfortable purchasing a product after physically examining it. (I)
5. When browsing in stores, it is important for me to handle all kinds of products. (A)
6. If I can’t touch a product in the store, I am reluctant to purchase the product. (I)
7. I like to touch products even if I have no intention of buying them. (A)
8. I feel more confident making a purchase after touching a product. (I)
9. When browsing in stores, I like to touch lots of products. (A)
10. The only way to make sure a product is worth buying is to actually touch it. (I)
11. There are many products that I would only buy if I could handle them before purchase. (I)
12. I find myself touching all kinds of products in stores. (A)
NOTE.—Scale descriptors ranged from ⫺3 (strongly disagree) to +3 (strongly agree). A p autotelic scale
item; Ip instrumental scale item.

Next, to assess content validity, a group of 12 under- administered additional scales in order to assess the scales’
graduate students were given construct descriptions and validity. Questionnaires were administered via the Internet
asked to classify the 50 items as assessing either the in- with 555 completed for a response rate of 28%.
strumental aspects of purchase-related touch or the autotelic The theoretical model specifying two latent factors un-
aspects of touch. Ten of the items were ambiguous and were derlying the NFT construct was tested by performing a con-
eliminated. Twenty-two of the items were classified as in- firmatory factor analysis on the variance covariance matrix
strumental and 18 items as autotelic by at least 10 of the using LISREL 8 (Jöreskog and Sörbom 1993). The overall
12 evaluators. fit indexes were acceptable, but examination of the item
reliabilities (Fornell and Larcker 1981), significant stan-
dardized residuals, and patterns in the modification indexes
Purification: Studies 1 and 2
indicated the need for further scale purification. This infor-
The 40-item scale was administered to a sample of 135 mation, along with an examination of the content of the
undergraduate students to provide preliminary estimates of items, suggested that one each of the autotelic and instru-
reliability and scale structure. The 18-item autotelic scale mental items could be dropped (see exhibit 1 for a list of
had a coefficient a of .93 while the 22-item instrumental the 12 scale items).
scale had an a of .88. Individual scale items were then To replicate the derived 12-item scale, a fourth study con-
assessed by examining item-to-total correlations and results taining NFT and scales used in assessing validity was con-
of a principal axis exploratory factor analysis with oblimin ducted with a new sample of university students (n p
rotation leading to additional purification to 20 items. The 418). To assess scale structure, a two-factor model using the
correlation between the instrumental and the autotelic di- variance covariance matrix was estimated using LISREL 8
mensions was .66, p ! .05. (Jöreskog and Sörbom 1993). Competing measurement
In the second purification study, the 20-item NFT scale models were examined for a one-factor model (x 2 p
was included in a survey that was mailed to a randomly 367.4, df p 54, p ! .01) versus a two-factor model with no
selected sample of 746 staff members of a university. Of correlation between the two factors (x 2 p 128.5, df p 54,
these, 267 were completed for a response rate of 36%. The p ! .01) versus a correlated two-factor model (x 2 p 88.2,
coefficient a was .95 for the autotelic dimension and .90 df p 53, p ! .01). A two-factor model with correlated di-
for the instrumental dimension. An exploratory factor anal- mensions (f p .64) provides a better fit relative to these
ysis using principal axis factoring yielded a two-factor so- competing models.
lution with an average instrumental loading of .74 and an Further examination of the solution revealed that all es-
average autotelic loading of .85. The correlation between timates were of the proper sign and were statistically sig-
the instrumental and the autotelic dimensions was .64, nificant. The reliabilities and fit statistics were all acceptable.
p ! .05. Additional examination of the item-level data in As noted above, the chi-square test was significant, but as
terms of factor loadings and cross loadings, communalities, Jöreskog and Sörbom (1993) note, the chi square should be
and item-to-total correlations resulted in further reduction regarded more as a measure of fit than as a strict test statistic,
to 14 items (autotelic p 7 items and instrumental p 7 because it may not be realistic to assume that the hypoth-
items). esized model holds exactly in the population. The NNFI fit
index was .94 and the CFI was .96, which exceed the .90
Dimensionality and Reliability: Studies 3 and 4 recommendation (Bentler 1992). The RMSEA of .07 rep-
resents reasonable errors of approximation in the population
In study 3, the 14-item NFT scale was administered to a and indicates a good fit (Byrne 1998), and RMR p .08.
random sample of 2,000 staff members at a university. We Reliability for the autotelic dimension was .89 and .87 for

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 433

TABLE 1

CONSTRUCT VALIDITY TESTS FOR NEED FOR TOUCH (STANDARDIZED STRUCTURAL EQUATION COEFFICIENTS)

Instrumental—NFT Autotelic—NFT Overall NFT

Type of asssessment Study 3 Study 4 Study 3 Study 4 Study 4

Response bias tests: Social desirability ⫺.05 .00 ⫺.02 ⫺.01 ⫺.02
Convergent validity: Need for Tactile Input NA .88* NA .59* .75*
Discriminant validity:
Need for Cognition NA .00 NA .03 .10
Need to Evaluate NA ⫺.01 NA .04 .05
Nomological validity:
Experiential shopping ⫺.04 .01 .31* .26* .27*
Impulse buying .09 .04 .24* .23* .28*
Catalog purchasing ⫺.43* ⫺.33* .08 .06 ⫺.26*
Telephone purchasing ⫺.37* ⫺.35* ⫺.13 .02 ⫺.22*
Internet purchasing ⫺.35* ⫺.36* ⫺.08 ⫺.03 ⫺.30*
NOTE.—NA p not available.
*Statistically significant at p ! .05.

the instrumental dimension (Fornell and Larcker 1981). The Tests of Response Bias. The potential confounding of
AVE for the autotelic dimension was .74 and .71 for the responses to the NFT scale by social desirability response
instrumental dimension, which exceeded the criterion of .50 bias was assessed, as has been advocated in consumer re-
(Fornell and Larcker 1981). As tests of discriminant validity, search (Mick 1996; Tian et al. 2001). This assessment was
the squared correlation between the two dimensions conducted by including the Crowne and Marlow (1964)
(f 2 p .41) was less than the AVEs and the confidence in- scale in studies 3 (a p .73) and 4 (a p .78). Across both
terval around f(.64 Ⳳ .22) did not contain 1.0. Across the studies and dimensions NFT was not related as predicted
two studies, the correlated two-factor model represents a with socially desirable responding (table 1). Further tests
reasonable fit of the data and reflects the theoretical structure also show that social desirability did not moderate nor sup-
of the items. press the subsequent relationships used in assessing the con-
struct validity of the NFT scale (Mick 1996).
Construct Validity
Convergent Validity. Recently, Citrin, Stem, Spangen-
As indicated previously, a number of measures were in- berg, and Clark (forthcoming) developed a six-item scale
cluded in studies 3 and 4 for validity purposes. To test the to measure consumer Need for Tactile Input (NTI). Although
predicted relationships, the coefficients were estimated using not formally defined as such, our understanding of the au-
LISREL 8 (Jöreskog and Sörbom 1993). The relationship thors’ characterization of the NTI input is that the scale is
between each measure and the overall NFT scale as well as oriented toward assessing brand/product evaluations (Citrin
the individual paths to the autotelic and instrumental di- et al., forthcoming). This suggests that the NTI domain over-
mensions were examined and are summarized in table 1.2 laps with the instrumental dimension of the NFT scale. For
example, items relate to the need to touch to evaluate product
2
When a scale is made up of multiple dimensions, differences exist in quality or to evaluate the physical characteristics of a prod-
both past practices for scale assessment as well as the theoretical origin
of these practices. For example, both Richins and Dawson (1992) and Tian,
uct. This characterization is consistent with their finding that
Bearden, and Hunter (2001) validated scales with multiple correlated di- the NTI scale was negatively related to the prior purchase
mensions at the overall composite level, whereas Netemeyer, Burton, and of products over the Internet. Thus, we expect that the NTI
Lichtenstein (1995) conducted a dimension-based analysis. Space does not scale will have a higher positive relationship with the in-
permit a detailed discussion of the arguments reflecting a composite vs. strumental dimension of NFT versus the autotelic dimen-
dimensional analysis, and the reader is referred to Carver (1989) and Hull,
Lehn, and Tedlie (1991) for additional details. We believe that the decision sion. The NTI scale (a p .96) was related positively to
to pursue one research strategy over another should be based on the level instrumental NFT (b p .88, p ! .05, table 1) and autotelic
of abstraction of the underlying theory and the goals of the researcher. For NFT (b p .59, p ! .05), with this difference statistically
instance, one researcher may be interested in and have theoretical predic- significant as predicted (chi-square equivalence test:
tions as to how separate dimensions of job satisfaction differentially relate
to a battery of antecedents or consequences, whereas another researcher
Dx 2 p 108.7, p ! .05). This supports the convergent valid-
may be interested in making predictions about the overall welfare of em- ity of the NFT scale, while at the same time underscoring
ployees and therefore focuses on composite job satisfaction (Richard Ba- its distinction from the NTI scale in terms of our addition
gozzi, personal communication with authors, June 10, 2001). For our pur- of individual differences in autotelic NFT.
poses, to avoid any potential interpretational confounding and to
demonstrate the merits of developing a multiple-dimensional NFT scale,
we analyze our results at the dimension level and selectively at the com- Discriminant Validity. To assess discriminant validity
posite level. Future researchers might choose to either employ the com- we included in study 4 the NFC scale (Cacioppo and Petty
posite NFT scale and/or its dimensions in their investigations. 1982; 18 items, a p .87) and the NTE scale (Jarvis and

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
434 JOURNAL OF CONSUMER RESEARCH

Petty 1996; 16 items, a p .79). Although NFC, like NFT the “experience” and would be driven more by the desire
overlap in the domain of information-acquisition behavior, for fun rather than to acquire information about, evaluate,
NFC taps this domain at a more macro level that is also not or purchase a product. Hence, we expected that experiential
specific to the consumption context. In contrast, the NFT shopping would relate positively with autotelic NFT and not
scale is more molecular in terms of its sense-specific focus be significantly related to instrumental NFT. Both studies 3
and narrower in tapping consumption behavior. Thus, we and 4 included the experiential shopping motivation scale
predict the scales will not be related. Similarly, NTE rep- (Dawson et al. 1990; a’s p .86 and .87). For the NFT sub-
resents information acquisition, but in the context of chronic scales, the autotelic dimension was, as predicted, signifi-
evaluation. This chronic form of evaluation across such ob- cantly related to the experiential motivation for shopping
jects as social issues, groups, and future behaviors is in (b’s p .31 and .26, p ! .05; table 1), but not related to
contrast to the product-specific nature of NFT. Evaluation instrumental NFT (b’s p ⫺.04 and .01, p 1 .05).
represents only one aspect of the NFT, particularly when The final assessment of nomological validity examined
instrumental is contrasted with autotelic touch. Thus, we the relationship between buying impulsiveness and NFT.
predict that NTE and NFT should not be related. These Buying impulsiveness (Rook and Fisher 1995) is defined as
predictions were supported (see table 1) as neither dimension a consumer’s tendency to buy spontaneously, unreflectively,
of NFT was related to NFC or NTE. Both findings support immediately, and kinetically. “Highly impulsive buyers are
the discriminant validity of the NFT scale. more likely to experience spontaneous buying; their shop-
ping lists are more ‘open’ and receptive to sudden, unex-
Nomological Validity. As evidence of nomological va- pected buying ideas” (Rook and Fisher 1995, p. 306). In
lidity, we considered various direct-marketing media such this sense, the impulse purchase trait is characterized by the
as shopping via catalog, over the telephone, and on the lack of a salient purchase goal, at least at the start of the
Internet. Evidence for a “visual preview model” has been shopping experience. Because autotelic NFT is more spon-
found that states that vision provides a quick glance that taneous and also characterized by a nonsalient purchase
results in broad but coarse information about the haptic goal, it is expected that buying impulsiveness will be pos-
properties of an object, information that is useful in directing itively related to autotelic NFT. However, because instru-
further processing (Klatzky, Lederman, and Matula 1993). mental NFT is more reflective and concerns a salient pur-
Thus, viewing a catalog or a Web page may reveal that more chase goal, it is expected that buying impulsiveness will be
detailed information about a haptic property is available, yet unrelated to instrumental NFT. Included in the question-
not readily attainable. A consumer who values this haptic naires for studies 3 and 4 was the nine-item buying impul-
information would be expected to be less likely to purchase siveness trait scale (Rook and Fisher 1995; a’s p .90, .88).
products via marketing channels where direct product touch The relationship between buying impulsiveness and auto-
is prohibited. This is consistent with the findings reported telic NFT was significant and positive (b’s p .24 and .23,
by Citrin et al. (forthcoming), who found consumer need p ! .05, table 1), as predicted. Additionally, the relationship
for more instrumentally oriented tactile information to be between buying impulsiveness and instrumental NFT was
negatively related to Internet purchase behavior. For these insignificant (b’s p .09 and .04, p 1 .05), again confirming
reasons, a negative relationship is expected between pur- our predictions.
chasing via these nontouch media and instrumental NFT.
Additionally, when consumers purchase via these direct me-
dia, it is more likely that they are engaged in shopping Summary and Discussion of Construct Validity
behavior with a salient purchase goal rather than for fun. Results
However, autotelic touch is concerned with touch without
a salient purchase goal. For this reason, we expect there to The first set of studies detailed the development and initial
be no relationship between purchasing via these media and assessment of the NFT individual difference scale. The NFT
autotelic NFT. scale was purified and a measurement model was specified
Both study 3 and study 4 included questions that asked based on several considerations. These analyses suggest that
participants whether they purchased via catalog, over the the NFT scale has two dimensions with desirable levels of
telephone, or on the Internet (e.g., “I order products using reliability, while not being confounded by social desirability
the Internet” using a five-point scale with endpoints “never” bias.
and “very often”). As predicted, across both studies instru- Convergent validity was supported by the relationship
mental NFT was negatively related to all three forms of between instrumental NFT and the newly developed NTI
direct-purchase behavior, while, also as predicted, autotelic scale (Citrin et al., forthcoming). The finding that the NTI
NFT was not related to the propensity to purchase products scale differentially correlated with the autotelic versus in-
via catalog, telephone, or the Internet (table 1). strumental dimension of NFT underscores the importance
To further assess the nomological validity of NFT, the of incorporating both dimensions within the NFT scale. Dis-
experiential motivation for shopping (Dawson, Bloch, and criminant validity of NFT was also supported through the
Ridgway 1990) was examined. Experiential shopping relates absence of a relationship between the dimensions of NFT
to social or recreational motivations rather than to purchase and both NFC and NTE.
products. In other words, this is shopping for the sake of Nomological validity of the scale was supported through

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 435

several predicted relationships. Purchasing via direct media would reflect more of a prepurchase task orientation, sub-
was negatively related to instrumental NFT but not related jects were asked to evaluate products while concurrently
to autotelic NFT. Buying impulsiveness was positively re- verbalizing their thoughts. Consistent with past research is
lated to the autotelic dimension of NFT, but not the instru- the prediction that if higher NFT individuals act as chronic
mental dimension. Additionally, experiential shopping processors of haptic information, then they should attend to
(Dawson et al. 1990) was positively related to autotelic NFT and verbalize haptic attributes earlier in their protocol than
and unrelated to instrumental NFT. those lower in NFT (Higgins et al. 1982). We also predict
The first set of studies provided strong support for the that as an individual difference advantage in accessibility,
psychometric properties of the NFT scale. One important results should replicate across the two products.
issue not addressed in these studies concerns why individ-
uals might differ in their preference for haptic information. Stimuli A sweater and tennis racket were chosen
Underlying our premise for individual differences in haptic through a pretest as products possessing salient haptic in-
information preferences is a greater accessibility to haptic formation. Forty-eight undergraduate students rated 15 prod-
information for those higher versus lower in NFT. To em- ucts in terms of whether touch played an important role in
pirically examine this premise, the following two studies their decision process on a seven-point scale (1 p touch is
were conducted. not important at all to 7 p touch is extremely important).
The mean importance of touch in the decision to purchase
was rated as equal for the two products (sweater: M p
NFT AND CHRONIC ACCESSIBILITY 6.4 vs. tennis racket: M p 6.2, p 1 .05).
Understanding the Nature of NFT Sample and Procedure. One hundred and seventy five
As defined previously, NFT is based on a preference for undergraduate students recruited from marketing classes in-
the extraction and utilization of information obtained dividually evaluated the products sequentially while their
through the haptic system. An underlying issue is the basis concurrent verbal protocols were audiotaped. A verbaliza-
for this preference for haptic information, which we predict tion warm-up procedure was used for a pen. Following this,
stems from a differential accessibility to haptic information either the sweater or the tennis racket was positioned on a
for higher versus lower NFT. We propose that this differ- raised table (order was counterbalanced). Following the
ential accessibility is based on the chronic accessibility of product examinations, participants filled out the NFT scale.
stored haptic information by those varying in NFT. In this After completion, they were asked whether they knew the
sense, chronic or long-term accessibility refers to an acti- purpose of the study. No participant realized that the focus
vation readiness potential of stored information and reflects of the study was to assess their haptic processing.
long-term processing influences on activation (Higgins Independent Variables. The 12-item NFT scale was
1996). In the person perception literature, evidence has used to assess individual differences in haptic information
shown a relationship between individual differences in the processing (reliability overall NFT p .92; instrumental
chronic accessibility of particular constructs and differences a p .90; autotelic a p .95). The entire range of the scale
in responding to stimuli (e.g., Higgins and Brendl 1995; from ⫺36 to 36 was represented in the sample. Higher and
Higgins, King, and Mavin 1982). lower NFT was determined by a median split with 84 sub-
We conjecture that persons higher in NFT are more likely jects scoring greater (lower) than the median of 11.5 cate-
to have haptic information accessible and to seek haptic gorized as higher (lower) in NFT. The correlation between
information and to use it as they form judgments. Consistent the autotelic and instrumental dimensions of NFT was .63,
with this view is that chronics on a particular construct are p ! .05.
individuals who list in a free recall task construct relevant
information sooner than do nonchronics (Higgins et al. Dependent Measures: Accessibility of Haptic Infor-
1982). In the attitude literature, salient beliefs are accessible mation. Ninety-six pages of verbal protocol were tran-
beliefs and are related to attitude toward an object. These scribed for the 175 subjects. Each statement was classified
beliefs are also commonly elicited as the first two or three according to whether or not it related to an assessment of the
beliefs in a free recall exercise (Fishbein and Ajzen 1975). product’s haptic properties (DeLong, Bye, and Larntz 1991).
Consequently, if haptic information is more chronic in its Examples of haptic comments would be statements related
accessibility for those higher in NFT, we would expect that to the feel or softness of the fabric or the weight of the tennis
persons higher in NFT will access touch related attributes racket. To evaluate the coding system, an experimenter and
sooner and faster than those lower in NFT. an independent judge coded the first 10 subjects (250 separate
statements), and agreement was 98%. Consistent with Higgins
Chronic Accessibility and Thought Verbalizations: (1996), our measure of accessibility was based on the ordering
Study 5 of haptic versus nonhaptic thought verbalizations. Specifi-
cally, the measure allows for a differential number of thoughts
Overview. This experiment consisted of two factors and weights the order of thoughts by the reverse of the number
based on a median split of the NFT scale and two products of thoughts. For instance, if a person verbalized 5 total
containing salient haptic attributes. To induce a goal that thoughts and the first and third were haptic related, the nu-

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
436 JOURNAL OF CONSUMER RESEARCH

merator would be 8 (5 + 3) and the denominator would be instrumental touch versus the spontaneous-enjoyment-ori-
15 (5 + 4 + 3 + 2 + 1) for a weighted haptic accessibility mea- ented autotelic form of touch. Under conditions where con-
sure of .67.3 The index is scored so such that higher values sumers do not possess a purchase-oriented processing goal,
correspond to haptic thoughts occurring earlier in the eval- we would expect more of a bottom-up or stimulus-driven
uation process. form of touch that should lead to a chronic accessibility
advantage for individuals higher in autotelic versus instru-
Results: Haptic Thoughts and NFT. Consistent with mental touch.
our chronic accessibility explanation, individuals higher in Support for an autotelic NFT advantage when a salient
NFT utilized haptic information earlier in their product eval- purchase goal does not exist comes from both the devel-
uations than their lower NFT counterparts. For the sweater, opmental literature and the conceptual nature of these two
individuals higher in NFT verbalized haptic attribute infor- forms of haptic processing. McClelland et al. (1989) noted
mation sooner than those lower in NFT (M p .46 vs. that implicit motives are acquired earlier in life on the basis
M p .30, t(82) p 3.6, p ! .05). This pattern of results also of prelinguistic affective experiences versus self-attributed
occurred for the tennis racket when comparing higher versus instrumental motives that are based on explicit acquisition
lower NFT (M p .33 vs. M p .22, t(86) p 2.8, p ! .05). of values or goals important for later childhood and adult
This same pattern of results also occurred at the dimension achievement. Similarly in their review, Bushnell and Boud-
level. reau (1991) conclude that infants perceive the material prop-
erty of texture as early as five to nine months of age. By
Discussion of Study 5 this age, infants exhibit preferential behavior for autotelic
touch by stroking textured objects that provide pleasurable
Across both products, results indicate that haptic infor- tactile feedback. In contrast, more functional haptic touch
mation is more chronically accessible to subjects who scored develops later (eight to nine months). Thus, there is a pri-
higher on the NFT scale versus those scoring lower on the macy in the development of more autotelic versus more
scale. People higher in NFT mentioned haptic attribute in- functional interactions with haptic object properties (Bush-
formation earlier in their evaluations than those lower in nell and Boudreau 1991). Additionally, our conceptualiza-
NFT, supporting greater accessibility of haptic information tion of autotelic touch is also consistent with McClelland
for the former. In study 5, subjects were instructed to eval- et al.’s (1989) discussion of the compulsive nature of an
uate the two products, thereby inducing a more purchase- implicit motive. As a spontaneous behavior without a salient
versus shopping-oriented type of processing goal (Babin et purchase goal, autotelic touch, like an implicit motive, is a
al. 1994). However under more experiential-oriented shop- less effortful automatic process that is not consciously con-
ping where a prespecified end purchase goal is not induced trolled and is more driven by the presence of a triggering
(Holbrook and Hirschman 1982), we would expect a more stimulus (Bargh 1984). These characteristics are also re-
spontaneous autotelic form of touch to be more prevalent. flective of impulsive behavior (Rook and Fisher 1995) and,
For example, an individual high in autotelic NFT, though as our previous studies show, autotelic but not instrumental
not intending to personally make a purchase, may accom- touch is positively related to impulse-buying behavior. Thus,
pany a friend who is looking for clothes. The individual we predict that absent a purchase goal, individuals higher
while browsing through a clothing store may spontaneously (lower) in autotelic touch will access haptic information
reach out a hand and suddenly notice a sweater’s unexpected faster (slower) than individuals higher versus lower in in-
softness. This example of a more spontaneous compulsive strumental touch. The next experiment directly addresses
form of haptic processing is one of our key distinctions this issue while providing an additional test of our chronic
between autotelic and instrumental touch. The next study accessibility explanation for individual differences in NFT.
extends study 5 results by examining differences in spon-
taneous chronic accessibility across sets of haptic and non- Overview. To induce a task that lacked a purchase-ori-
haptic stimuli between those higher versus lower in autotelic ented goal, we had subjects perform a lexical decision task
versus instrumental NFT. (Rubenstein, Lewis, and Rubenstein 1971) using reaction-
time methodology. Study 6 was a two (levels of NFT: be-
Chronic Accessibility and Spontaneous Haptic tween) # three (text stimulus: within) # 10 (trials: within)
mixed factorial design. It was expected that under the stim-
Processing: Study 6 ulus-driven nature of this task, for individuals higher versus
Accessibility and Autotelic versus Instrumental lower in autotelic NFT, their chronic accessibility of haptic
Touch. As our results in the previous study suggest, in- information would lead to shorter reaction times for haptic
creased accessibility of haptic attributes when evaluating words, but not for nonhaptic words or nonwords. However,
products is based on an individual preference for haptic we condition this effect by an additional interaction with
information. However, this product-evaluation task does not the trials manipulation. By its nature, chronic accessibility
distinguish between the more goal-driven utilitarian form of represents an individual’s activation readiness of stored in-
formation (Higgins 1996). Thus, higher compared to lower
3
We thank one of the reviewers for suggesting this index of haptic thought autotelic NFT should demonstrate a reaction-time advantage
accessibility. for haptic words initially, but this advantage should dissipate

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 437

in later trials. Once those lower in autotelic NFT have ac- Each of the 10 trials consisted of responses to the 30 text
cessed the haptic information, they should also possess an stimuli. At the end of each trial, a page was inserted that
activation readiness on subsequent trials and should dem- instructed subjects to “press any key” when they were ready
onstrate equal accessibility with higher autotelic NFT. We to begin. This allowed subjects a short rest between trials
also predict no interaction with trials for nonhaptic words before proceeding at their own pace through the 10 trials.
or nonwords. Given that the task does not induce a salient Need for Touch was measured with the 12-item scale
purchase-oriented goal, we predict no differences in this (overall reliability p .93; instrumental a p .90; autotelic
pattern of effects for individuals higher versus lower in in- a p .92). Lower and higher autotelic (instrumental) NFT
strumental NFT. were determined by median splits, with 53 (56) subjects
scoring above the median of 3 (4) classified as higher along
Procedure One hundred university students partici- each NFT dimension. The correlation between the autotelic
pated in the study for extra credit from a psychology de- and instrumental dimensions was .72, p ! .05.
partment subject pool. On entering the research lab, subjects
signed a consent form and completed a questionnaire that Dependent Measures. In the lexical decision task, sub-
contained the NFT scale. Subjects then received instructions jects were instructed to press a computer key based on
for the lexical decision task. They were informed that text whether the text was a word or a nonword. The keys (“z”
in the form of English language words and nonwords would or “/”) were counterbalanced for the correct answer. The
be presented one at a time on a computer screen and to response time in milliseconds served as the dependent var-
identify if a stimulus was a word by pressing the appropriate iable. In total, subjects provided 300 responses across the
key. They were instructed to proceed through the screens 10 trials, averaging 95% correct responses; differences
as fast as possible, but to be accurate as well. Subjects first across treatment groups were of no consequence (correla-
completed a practice trial using different test stimuli and tions with treatments ranged from .04 to .09, p 1 .05). Given
then proceeded to the first of 10 additional trials adminis- the sample size and nature of reaction times the square root
tered on the computer. Following completion, subjects were of reaction time was used as the dependent variable as rec-
thanked and debriefed. ommended by Hair, Anderson, Tatham, and Black (1998).
Actual reaction times are used when discussing the results.
Independent Variables. The stimuli consisted of a
combination of 30 words and nonwords. Ten nonhaptic Overview of Results. There are a number of basic ef-
words, 10 pronounceable nonwords, and 10 haptically re- fects that are significant and consistent with the nature of
lated words were selected from an initial pretest. In the the study. We provide an overview of these results first,
pretest, 10 subjects assessed 106 words comprising a mix- before addressing the NFT-related results. There are signif-
ture of nonhaptic, haptic, and pronounceable nonwords, the icant differences between the text stimuli (F(2, 97) p
latter drawn from studies conducted by Rubenstein et al. 51.7, p ! .05) with longer reaction times for nonwords
(1971). Subjects were to indicate whether they associated (M p 573 msec) than either nonhaptic (M p 518 msec) or
the words/nonwords with any combination of their five haptic words (M p 513 msec). There is also a main effect
senses by placing an X under the options designated as sight, for trials (F(9, 90) p 31.1, p ! .05). As would be expected,
smell, touch, hearing, taste, or “none of these.” From these there is a learning effect with longer reaction times at trial
responses, 10 nonwords were selected using the criterion 1 (M p 780 msec) and a steady decline in subsequent trials
that at least nine of the 10 subjects had to indicate that it (e.g., trial 10; M p 411 msec). There is no overall signif-
was not related to the five senses. Similarly, at least nine icant difference between the NFT groups (lower NFT:
of 10 subjects had to rate a word to be associated with touch M p 536 msec vs. higher NFT: M p 533 msec,
to generate the 10 haptically related words. Nonhaptic words F(1, 98) p .001, p 1 .05). This applies to the NFT dimen-
were selected such that the words were not to be haptically sions as well (higher autotelic: M p 527 msec vs. lower
related, with majority category ratings ranging from taste autotelic: M p 543 msec, F(1,98) p .6, p 1 .05; higher in-
(three words), smell (one word), sight (two words), and strumental: M p 535 msec vs. lower instrumental: M p
“none of these” (four words). This resulted in the selection 534 msec, F(1, 98) p .01, p 1 .05).
of the following: 10 haptic words (“rough,” “grasp,”
“scratchy,” “smooth,” “firmness,” “hold,” “coarse,” Accessibility of Haptic Information. It was expected
“squeeze,” “soft,” and “grip”), 10 nonhaptic words (“sugar,” that for individuals higher versus lower in autotelic NFT,
“task,” “debt,” “tray,” “wine,” “lock,” “grass,” “salad,” their chronic accessibility of haptic information would lead
“hour,” and “fate”), and 10 pronounceable nonwords to shorter reaction times for haptic words, but not for non-
(“hosk,” “pronk,” “staim,” “rolt,” “groot,” “blesp,” “trene,” haptic or nonwords. There is a marginally significant three-
“sneap,” “tors,” and “slint”). The stimuli were displayed way interaction (F(18, 81) p 1.7, p p .06) between auto-
individually on the computer screen in 72-point bold Hel- telic NFT, text stimuli, and trials, but not for instrumental
vetica font. Following each text stimulus was a filled oval NFT (F(18, 81) p .7, p 1 .05). However, more important is
that masked the presentation area for 1.5 sec. The stimuli the pattern of this interaction. One primary comparison is
were presented in four randomly selected orders that were the expected significant difference (F(1, 98) p 2.1, p !
counterbalanced across treatments. .05) between higher (727 msec) versus lower autotelic NFT

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
438 JOURNAL OF CONSUMER RESEARCH

(841 msec) for haptic words for the first trial. Thus, there direct product experiences than after exposure to advertising.
is a 114 msec advantage in the speed of access for those However, Wright and Lynch (1995) maintain that confidence
higher in autotelic NFT for haptic words. In contrast, as in judgment may not necessarily be greater after direct ex-
expected, no significant differences exist between higher and perience but depends on the type of attribute information
lower autotelic NFT for nonhaptic words (M’s p 701 msec communicated. Their “media congruence” hypothesis dis-
and 752 msec, respectively, F(1, 98) p 1.0 , p 1 .05) or non- tinguishes between search and experience attributes. Briefly,
words (M’s p 792 msec and 874 msec, respectively, consumers can acquire search-attribute information (e.g., the
F(1, 98) p 1.4, p 1 .05). The advantage for higher versus brand name or the color of a sweater) from secondhand
lower autotelic NFT in accessibility for haptic words dis- sources such as product descriptions or advertising without
appeared on the second trial (higher: 606 msec vs. lower: directly trying (or touching) a product. In contrast, expe-
586 msec, F(1, 98) p .7, p 1 .05) and persisted for the other rience attributes (e.g., the feel of a sweater) can be ascer-
text stimuli (nonhaptic words, higher: 576 msec vs. lower: tained only by use (or possibly touch) because they are
589 msec, F(1, 98) p .5, p 1 .05; and nonwords, higher: dependent on subjective experience. Wright and Lynch
665 msec vs. lower: 669 msec, F(1, 98) p .1, p 1 .05). This (1995) found that direct product experience is best suited
equal accessibility for all text stimuli continued through the to experience attributes, whereas printed material (such as
tenth trial (haptic words, higher: 372 msec vs. lower: 408 advertising) is best suited to search attributes.
msec, F(1, 98) p 1.2, p 1 .05; nonhaptic words, higher: 398 In certain retail scenarios, such as shopping on the In-
msec vs. lower: 424 msec, F(1, 98) p 1.0, p 1 .05; and non- ternet, a consumer has no direct opportunity to experience
words, higher: 410 msec vs. lower: 459 msec, F(1, 98) p a product through touch before purchasing via these media.
1.4, p 1 .05). Identical tests for instrumental NFT did not In addition, in-store obstacles such as retail display cases
reveal any significant differences in reaction times by type preclude or diminish the opportunity to directly experience
of text across the trials. There was no significant difference a product through touch. Using a media congruence (Wright
between higher (769 msec) vs. lower instrumental NFT (794 and Lynch 1995) argument, a sweater, which we argue pos-
msec, F(1, 98) p .4, p ! .05) for haptic words for the first sesses experience attributes, should best be evaluated by
trial. There were also no significant differences across higher touch. However, we propose that NFT will moderate this
and lower instrumental NFT for nonhaptic words (M’s p relationship.
727 msec and 723 msec, respectively, F(1, 98) p .1, p 1 Results from the prior two studies support the premise
.05) or nonwords (M p 830 msec and 832 msec, respec- that individuals who are more internally motivated (higher
tively, F(1, 98) p .03, p 1 .05). This same pattern of effects NFT) to examine the haptic attributes of a product have
persisted for all subsequent trials. greater memory accessibility to haptic information. Past re-
Results support our chronic accessibility explanation by search has demonstrated that as information accessibility
demonstrating an initial spontaneous reaction-time advan- increases, the likelihood that this information will be used
tage for higher versus lower autotelic NFT for haptic words, as input for judgment and choice also increases (e.g., Hig-
but a lack of advantage in subsequent accessibility. Our gins and Brendl 1995; Lingle and Ostrom 1979). This would
results also show that under a non-purchase-oriented goal suggest that for a product salient in haptic attributes (such
this reaction-time advantage does not occur for individuals as a sweater), higher NFT individuals would be more con-
higher in instrumental NFT. On the one hand, results are fident in their judgment when they can, versus cannot, di-
consistent with study 5 in supporting our chronic accessi- rectly touch the product. In contrast, lower NFT individuals’
bility explanation for preferences in NFT. However, in con- confidence in their judgments should be less affected by the
trast to the previous study, by varying the task, we dem- lack of opportunity to actually touch a product high in ex-
onstrate in the current study a more automatic form of perience attributes, and these individuals will likely be con-
chronic accessibility for autotelic touch. We return to a dis- tent just to visually examine the product (Klatzky et al.
cussion of this issue in the conclusions section. Before that, 1993).
we examine whether NFT will moderate judgments under
conditions where consumers are able to directly experience Overview of Study. The purpose of study 7 was to
products or because of distribution channel or merchandis- experimentally assess the moderating effects of NFT on
ing reasons (e.g., displays or packaging) touch is not direct experience and product evaluation. Study 7 was a 2
permitted. (levels of NFT) # 2 (direct product experience) between-
subjects factorial design.
NFT AND DIRECT PRODUCT Independent Variables. Direct experience with the
EXPERIENCE product was operationalized as either a barrier or no barrier
Direct Experience and NFT: Study 7 to touch. In the no-barrier condition, the sweater was folded
on a table with the subject able to touch the product if they
Direct experience with an object has been reported to desired. In the barrier-to-touch condition, the sweater was
increase confidence in judgment (Smith and Swinyard similarly displayed but placed under Plexiglas simulating a
1983). For example, Smith and Swinyard (1988) found that retail display case.
product-belief strength and confidence were greater after Need for Touch was measured with the 12-item scale

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 439

(overall reliability p .95; instrumental a p .89; autotelic Discussion of Study 7

a p .94). Lower versus higher NFT were determined by a
median split with the 29 subjects scoring above the median Study 7 tested predictions concerning the interaction be-
(a score of one) classified as high NFT. The correlation tween NFT and the opportunity to obtain haptic information
between the autotelic and instrumental dimensions was .74, through direct experience. When individuals could directly
p ! .05. experience a product by touch during evaluation, confidence
in their product judgment was greater for higher, but not for
Procedure and Stimulus. Sixty subjects were informed lower NFT individuals. Since haptic information is preferred
they were needed for a study on product evaluation. In the by higher NFT individuals, there is no substitute for directly
first phase, subjects completed a multipart questionnaire that experiencing this haptic information. While lower NFT sub-
included the NFT scale and then signed up for a second jects’ confidence judgments did not change across barrier
phase that occurred at least five weeks later. This time period conditions, higher NFT subjects were more confident when
was used to disassociate the two parts of the study, which they could touch to evaluate the product. Need for Touch
was confirmed in interviews with the participants following was found to moderate the media-congruence relationship.
the study. Two of the 60 subjects failed to return for the If a product contains salient haptic information, higher NFT
second phase, resulting in a sample size of 58. Individually individuals have a need to experience the product directly.
tested subjects were asked to evaluate the product with either In contrast confidence for those lower in NFT is attained
no opportunity to touch (displayed under Plexiglas) or full through a visual examination of the product. In essence, the
opportunity to touch (sitting on a table). After evaluating same attribute could be considered an experience attribute
the sweater, subjects filled out a measure of confidence in for those higher in NFT, while a search attribute for those
their judgment. lower in NFT.
The sweater used in study 5 was also used in this ex-
periment. The sweater had a fictitious but neutral brand
name (Baxter), which was also determined by a pretest. A CONCLUSIONS AND IMPLICATIONS FOR
written description, adapted from a Land’s End catalog de- FUTURE RESEARCH
scribed the sweater’s construction, fabric, and available
sizes. Synthesis of Results across Studies
Dependent Measures. Confidence in judgment was Evidence has been provided for an individual difference
measured by using two seven-point scales (“not very con- in preference for haptic information. A 12-item NFT scale
fident” to “very confident” and “not very sure” to “very was developed and its psychometric properties empirically
sure”). An overall confidence measure was calculated by assessed in a set of studies (studies 1–4). The scale possessed
taking the mean of the two items (r p .97). the hypothesized two-factor structure, demonstrated high re-
Results. There was the predicted significant interaction liability, and was found to systematically adhere to a battery
between NFT and direct experience (F(1, 54) p 5.02, p ! of theoretically grounded relationships. Additional results
.05). Consistent with expectations, individuals higher, but showed that for those higher in NFT, the lack of direct
not lower, in NFT had more confidence in their judgment experience through a barrier to touch resulted in less con-
of the product when they actually touched the sweater. fidence in their judgment. Elevated confidence through di-
Higher NFT subjects who directly experienced the product rect experience did not occur for those lower in NFT (study
had more confidence in their judgment than those higher in 7). Haptic information was also found to be more chronically
NFT in the barrier-to-touch (retail case) condition (M’s p accessible for those higher in NFT as measured by both
6.25 vs. 5.15, F(1, 54) p 21.05, p ! .05). For individuals order of thoughts elicited during product evaluation and in
lower in NFT, confidence in judgment was not dependent performing a haptic related reaction-time task (studies 5 and
on whether they were able to actually touch the sweater 6). Results vary, however, across the autotelic and instru-
(M’s p 5.71 and 5.67, F(1, 54) p .40, p 1 .05).4 mental dimensions, an issue we turn to next.
4
At the dimensional level, higher autotelic NFT individuals who directly
experienced the product had more confidence in their judgment than those Accessibility for Autotelic and Instrumental NFT
in the barrier-to-touch condition (M p 6.2 vs. 5.2, F(1, 54) p 7.8, p !
.05). For those lower in autotelic NFT, confidence in judgment was not Our research demonstrates (studies 5 and 6) that individ-
dependent on touching the sweater (M’s p 5.8 and 5.7, F(1, 54) p .30, uals higher in NFT are more likely to chronically access
p 1 .05). This pattern of results also occurred for individuals who differed haptic information. Using thought verbalizations, those
in instrumental NFT (higher instrumental touch: M p 6.2 vs. lower in-
strumental no touch: M p 4.8, F(1, 54) p 18.2, p ! .05; and lower instru-
higher in both autotelic and instrumental NFT mentioned
mental touch: M p 5.8 vs. lower instrumental no touch: M p 5.9, haptic information sooner than individuals lower on these
F(1, 54) p .15, p 1 .05). In conducting the latter analyses, one concern was dimensions. In study 6, we found that those higher in au-
whether several of the instrumental items that contained confidence in their totelic NFT demonstrated faster access to haptic informa-
wording perhaps artificially inflated the relationship with attitude confi- tion, whereas this did not occur for instrumental NFT. One
dence. Additional analyses were conducted by partitioning the dimension
into parcels for items that did vs. did not share this wording. Results were key difference between these two studies was the nature of
identical across the parcels supporting the robustness of the findings as the task. In study 5 individuals evaluated products, which
reported. elicited a more purchase-oriented form of processing. In

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
440 JOURNAL OF CONSUMER RESEARCH

contrast, in study 6 the task was to identify words where Haptics and Product Design
no evaluation or purchase motivation was present. Results
from both studies support our premise of chronic accessi- Haptics also has implications for the area of product de-
bility for NFT, but qualitative differences exist in the nature sign and development. Consideration of haptics is being
of autotelic versus instrumental haptic processing. Our used to revolutionize the design of automobiles with tactile
premise is that autotelic processing is more automatic and feedback components emerging. For example, BMW’s new
spontaneous, whereas instrumental processing is a more con- flagship 7 Series includes an iDrive user interface, designed
trolled and a more conscious process. Although our results by Immersion Corporation, that consolidates control of over
reflect this distinction, more research is needed to directly 700 functions in one knob. iDrive “provides complex sen-
examine these differences in the nature of chronic acces- sations that let practiced users tunnel through multiple
sibility of haptic information. For instance, Bargh (1984) menus without looking at the display. For example, when
has described a set of criteria for differentiating automatic scrolling down a phone list the knob will emit a tactile
from controlled processes, including the manipulation of ‘bump’ whenever a name is passed to give the operator a
processing capacity, the salience of stimuli, and the consis- sense of speed and location” (Whitfield 2002, p. 37). The
tency of information. For the latter, it has been reported that primary goal of iDrive is to reduce driver distraction through
consistent personality traits for those possessing chronic ac- the transfer of some information load from the visual to the
cessibility to related categories were processed automati- tactile system. This may be differentially suitable for high
cally, whereas inconsistent traits were processed in a more versus low NFT individuals. According to resource-match-
controlled conscious fashion (Bargh, Thein, and Friedman ing theory (Anand and Sternthal 1987), optimum processing
1983). Additional research of this nature would be useful is attainable when the resources allocated to a task match
in furthering our understanding of individual differences in those required for the task. High NFT consumers have haptic
the processing of autotelic and instrumental information. information accessible while using less of their cognitive
capacity processing the haptic information and may more
easily adapt to the iDrive. However, low NFT may initially
have to expend additional resources focusing on the tactile
NFT: Attention and Memory information and be even more distracted from visual infor-
mation. It is possible that designs with tactile feedback may
As our research demonstrates, haptic information is more be differentially preferred by, and differentially effective for,
chronically accessible to touch-oriented individuals. Thus, high versus low NFT consumers.
higher NFT individuals are able to more readily retrieve this A way to address this may be to encourage high and low
type of information from memory. Related to this is the NFT consumers to have input into the product-design pro-
interplay between memory and attention and how this re- cess. The use of visual imagery in the product-design pro-
lationship may also play a role in differentiating between cess has been found to influence the customer appeal of a
individuals who vary in NFT. For instance, research in- design of a new product (Dahl, Chattopadhyay, and Gorn
volving an individual’s visual versus verbal style of pro- 1999). Perhaps high and low NFT individuals differentially
cessing (Heckler et al. 1993) reported a person’s preference image haptic-based product information. How haptic-based
to engage in a type of processing has attentional as well as mental imagery affects the accessibility of different types
working memory implications. This may indicate that per- of product representations may provide insights into the
sons higher in NFT are also more likely to attend to haptic product-design process.
information than are their lower NFT counterparts. In ad-
dition, the differential haptic accessibility for higher com- Compensation for Haptic Information
pared to lower NFT individuals may indicate that higher
NFT individuals are more likely to form richer mental prod- With the growth of online shopping, the consequences of
uct representations, which include haptic properties in mem- an inability to touch on product evaluations are important
ory. Research needs to be conducted on the nature of the to understand. Brand names, low prices, or other nonhaptic
memory system for haptic information and the mechanisms compensation mechanisms (Kirmani and Rao 2000) may
for its encoding and retrieval. Included in the latter are the signal both high and low NFT shoppers to forgo product
role that cognitive elaboration strategies play in affecting touch before purchase. However, barriers to touch inhibit
the salience of haptic information and whether this facilitates the use of haptic information and consequently decrease
or inhibits the utilization of information available through confidence in product evaluations for high NFT, but not low
other sensory forms of input. It is not clear from past re- NFT, individuals. This raises the question of how to com-
search (Schiff and Oldak 1990; Welch and Warren 1980) pensate for haptic information when touch is unavailable.
whether haptic forms of information might differentially Concrete haptic written descriptions and visual depictions
affect (either facilitate or inhibit) the use of other perceptual of products can partially enhance acquisition of certain types
sense-based forms of information (e.g., smell or sounds), of touch information (Peck and Childers 2003). In addition,
particularly for individuals who are less reliant on (lower devices to provide haptic information have been developed
NFT consumers) versus consumers that are more attuned to (e.g., Logitech iFeel mouse) to attempt to provide haptic
haptic information. information when it is unavailable (Burdea 1996). However,

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
INDIVIDUAL DIFFERENCES IN TOUCH 441

are these devices more effective at compensation for low Cognition,” Journal of Personality and Social Psychology,
versus high NFT individuals? This is an important area for 42 (1), 116–131.
consumer research. Additionally, the haptic interfaces de- Carver, Charles (1989), “How Should Multifaceted Personality
veloped are still crude relative to direct haptic exploration, Constructs Be Tested? Issues Illustrated by Self-Monitoring,
and the sense of touch is thought to be the most complicated Attributional Style, and Hardiness,” Journal of Personality
and Social Psychology, 56 (4), 577–585.
sense to replicate (Moneyline 2000). Until then, mechanisms
Churchill, Gilbert A., Jr. (1979), “A Paradigm for Developing Bet-
for compensating for an inability to touch products are im- ter Measures of Marketing Constructs,” Journal of Marketing
portant to investigate. Research, 16 (February) 64–73.
As this article asserts, it is clear that much more research Citrin, Alka, Donald Stem, Jr., Eric Spangenberg, and Michael
is needed in the domain of product touch. Not only is there Clark (forthcoming), “Consumer Need for Tactile Input: An
an individual difference in the preference for haptic infor- Internet Retailing Challenge,” Journal of Business Research.
mation, but also different situations may motivate shoppers Crowne, Douglas R. and David Marlowe (1964), The Approval
to want to touch before purchase and for different types of Motive, New York: John Wiley.
products. How these factors impact the study of consumer Dahl, Darren W., Amitava Chattopadhyay, and Gerald J. Gorn
touch can now be assessed by the incorporation in future (1999), “The Use of Visual Mental Imagery in New Product
research of individual differences in the motivation to ac- Design,” Journal of Marketing Research, 26 (1)18–28.
quire and utilize haptic forms of product information. Dawson, Scott, Peter H. Bloch, and Nancy M. Ridgway (1990),
“Shopping Motives, Emotional States, and Retail Outcomes,”
Journal of Retailing, 66 (Winter), 408–427.
[David Glen Mick served as editor and William O. DeLong, Marilyn Revell, Elizabeth Kersch Bye, and Kinley Larntz
Bearden served as associate editor for this article.] (1991), “Perceptions of Successful Apparel Products,” Home
Economics Research Journal, 19 (3), 245–251.
Fiore, Ann Marie, Josephine Maria Moreno, and Patricia A. Kimle
REFERENCES (1996), “Aesthetics: A Comparison of the State of the Art
Outside and Inside the Filed of Textiles and Clothing. III.
Anand, Punam and Brian Sternthal (1987), “Strategies for De- Appreciation Process, Appreciator and Summary Compari-
signing Persuasive Messages: Deductions from the Resource sons,” Clothing and Textiles Research Journal, 14 (3),
Matching Hypothesis,” in Cognitive and Affective Responses 169–184.
to Advertising, ed. Patricia Cafferata and Alice Tybout, Lex- Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention
ington, MA: Lexingon Books, 135–159. and Behavior: An Introduction to Theory and Research, Read-
Anderson, James C. and David W. Gerbing (1988), “An Updated ing, MA: Addison-Wesley.
Paradigm for Scale Development Incorporating Unidimen- Fornell, Claes and David F. Larcker (1981), “Evaluating Structural
sionality and Its Assessment,” Journal of Marketing Research, Equation Models with Unobservable Variables and Measure-
25 (2), 186–192. ment Error,” Journal of Marketing Research, 18 (1), 39–50.
Babin, Barry J., William R. Darden, and Mitch Griffin, (1994), Gibson, James J. (1966), The Senses Considered as Perceptual
“Work and/or Fun: Measuring Hedonic and Utilitarian Shop- Systems, Boston: Houghton Mifflin.
ping Value,” Journal of Consumer Research, 20 (4), 644–656. Hair, Joseph Jr., Rolph Anderson, Ronald T. Tatham, and William
Bargh, John A. (1984), “Automatic and Conscious Processing of Black (1998), Multivariate Data Analysis, 5th ed., Upper Sad-
Social Information,” in Handbook of Social Cognition, Vol. dle River, NJ: Prentice-Hall.
3, ed. Robert Wyer, Jr. and T. Srull, Hillsdale, NJ: Erlbaum,
Heckler, Susan E., Terry L. Childers, and Michael J. Houston
1–44.
(1993), “On the Construct Validity of the SOP Scale,” Journal
Bargh, John A., Roman Thein, and Richard S. Friedman (1983),
of Mental Imagery, 17 (3–4), 119–132.
“Category Accessibility and Perceptual Hypotheses: Conse-
Higgins, E. Tory (1996), “Knowledge Activation: Accessibility,
quences for the Attentional Selection, Recall, and Use of Cat-
egory-Consistent and Inconsistent Person Information,” work- Applicability, and Salience,” in Social Psychology: Handbook
ing paper, Department of Psychology, New York University, of Basic Principles, ed. E. Tory Higgins and A. W. Kruglanski,
10003. 133–168, New York: Guilford.
Bentler, Peter M. (1992), “On the Fit of Models to Covariances Higgins, E. Tory and C. Miguel Brendl (1995), “Accessibility and
and Methodology,” Psychological Bulletin, 112 (3), 400–404. Applicability: Some ‘Activation Rules’ Influencing Judg-
Burdea, Gigore C. (1996), Force and Touch Feedback for Virtual ments,” Journal of Experimental Social Psychology, 31 (3),
Reality, New York: Wiley. 218–243.
Burke, Raymond R. (1997), “Do You See What I See? The Future Higgins, E. Tory, Gillian A. King, and Gregory H. Mavin (1982),
of Virtual Shopping,” Academy of Marketing Science, 25 (4), “Individual Construct Accessibility and Subjective Impres-
352–360. sions and Recall,” Journal of Personality and Social Psy-
Bushnell, Emily W. and J. Paul Boudreau (1991), “The Devel- chology, 43 (1), 35–47.
opment of Haptic Perception During Infancy,” in The Psy- Holbrook, Morris B. and Elizabeth C. Hirschman (1982), “The
chology of Touch, ed. Morton A. Heller and William Schiff, Experiential Aspects of Consumption: Consumer Fantasies,
Hillsdale, NJ: Erlbaum, 139–161. Feelings, and Fun,” Journal of Consumer Research, 9 (2),
Byrne, Barbara M. (1998), Structural Equation Modeling with LIS- 132–140.
REL, PRELIS, and SIMPLIS: Basic Concepts, Applications, Hornik, Jacob (1992), “Tactile Stimulation and Consumer Re-
and Programming, Mahwah, NJ: Erlbaum. sponse,” Journal of Consumer Research, 19 (3), 449–458.
Cacioppo, John T. and Richard E. Petty (1982), “The Need for Hull, Jay, Daniel Lehn, and Judith Tedlie (1991), “A General Ap-

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms
442 JOURNAL OF CONSUMER RESEARCH

proach to Testing Multifaceted Personality Constructs,” Jour- nenberg (1997), “Exploring the Implications of the Internet
nal of Personality and Social Psychology, 61 (6), 932–945. for Consumer Marketing,” Journal of the Academy of Mar-
Jarvis, W. Blair G. and Richard Petty (1996), “The Need to Eval- keting Science, 25 (4), 329–346.
uate,” Journal of Personality and Social Psychology, 70 (1), Richins, Marsha and Scott Dawson (1992), “A Consumer Values
172–194. Orientation for Materialism and Its Measurement: Scale De-
Johansson, Roland S. (1978), “Tactile Sensibility in the Human velopment and Validation,” Journal of Consumer Research,
Hand: Receptive Field Characteristics of Mechanoreceptive 19 (3), 303–316.
Units in the Glabrous Skin Area,” Journal of Physiology, 281 Rook, Dennis W. and Robert J. Fisher (1995), “Normative Influ-
(2), 101–123. ences on Impulsive Buying Behavior,” Journal of Consumer
Jöreskog, Karl G. and Dan Sörbom (1993), LISREL 8: Structural Research, 22 (3), 305–313.
Equation Modeling with the SIMPLIS Command Language, Rubenstein, Herbert, Spafford Lewis, and Mollie Rubenstein
Chicago: Scientific Software International. (1971), “Evidence for Phonemic Recoding in Visual Word
Kirmani, Amna and Akshay Rao (2000), “No Pain, No Gain: A Recognition,” Journal of Verbal Learning and Verbal Be-
Critical Review of the Literature on Signaling Unobservable havior, 10 (6), 645–657.
Quality,” Journal of Marketing, 64 (2), 1–27. Schiff, William and Rivka Oldak (1990), “Accuracy of Judging
Klatzky, Roberta L. and Susan J. Lederman (1992), “Stages of Time to Arrival: Effects of Modality, Trajectory, and Gender,”
Manual Exploration in Haptic Object Identification,” Percep- Journal of Experimental Psychology: Human Perception and
tion & Psychophysics, 52 (6), 661–670. Performance, 16 (2), 303–316.
——— (1993), “Toward a Computational Model of Constraint- Sherry, John F., Jr. (1990), “A Sociocultural Analysis of a Mid-
Driven Exploration and Haptic Object Identification,” Per- western Flea Market,” Journal of Consumer Research, 17 (1),
ception, 22 (5), 597–621. 13–30.
Klatzky, Roberta L., Susan J. Lederman, and Dana E. Matula Sherry, John F., Jr., Mary Ann McGrath, and Sidney L. Levy
(1993), “Haptic Exploration in the Presence of Vision,” Jour- (1993), “The Dark Side of the Gift,” Journal of Business
nal of Experimental Psychology: Human Perception and Per- Research, 28 (3), 225–244.
formance, 19 (4), 726–743. Siegel, Rudolph E. (1970), Galen on Sense Perception, Basel:
Lederman, Susan J. and Roberta L. Klatzky (1987), “Hand Move- Karger.
ments: A Window into Haptic Object Recognition,” Cognitive Smith, Robert E. and William R. Swinyard (1983), “Attitude-Be-
Psychology, 19 (3), 342–368.
havior Consistency: The Impact of Product Trial versus Ad-
Lingle, John H. and Thomas M. Ostrom (1979), “Retrieval Selec-
vertising,” Journal of Marketing Research, 20 (3), 257–267.
tivity in Memory-Based Impression Judgments,” Journal of
——— (1988), “Cognitive Response to Advertising and Trial: Be-
Personality and Social Psychology, 37 (2), 180–194.
lief Strength, Belief Confidence and Product Curiosity,” Jour-
McClelland, David, Richard Koestner, and Joel Weinberger (1989),
Psychological Review, 96 (4), 690–702. nal of Advertising, 17 (3), 3–14.
Mick, David Glen (1996), “Are Studies of Dark Side Variables Spreen, Otfried and Esther Strauss (1991), “Tactile, Tactile and
Confounded by Socially Desirable Responding? The Case of Visual, and Tactile-Motor Tests,” A Compendium of Neurop-
Materialism,” Journal of Consumer Research, 23 (2), sychological Tests: Administration, Norms, and Commentary,
106–119. New York: Oxford University Press, 334–385.
Moneyline (2000), recording of January 19 CNN broadcast, In- Tian, Kelly, William O. Bearden, and Gary L. Hunter (2001),
tegrated Media Systems Center Engineering Research Center, “Consumers’ Need for Uniqueness: Scale Development and
University of Southern California, Los Angeles, video- Validation,” Journal of Consumer Research, 28 (1), 50–66.
cassette. Welch, Robert B. and David H. Warren (1980), “Immediate Per-
Netemeyer, Richard, Scot Burton, and Donald Lichtenstein (1995), ceptual Response to Intersensory Discrepancy, ” Psycholog-
“Trait Aspect of Vanity: Measurement and Relevance to Con- ical Bulletin, 88 (3), 638–667.
sumer Behavior,” Journal of Consumer Behavior, 21 (4), Whitfield, Kermit (2002), “Touch and Go,” Automotive Design &
612–626. Production, 114 (6), 36–38.
Peck, Joann and Terry L. Childers (2003), “To Have and To Hold: Wright, Alice A. and John G. Lynch, Jr. (1995), “Communication
The Influence of Haptic Information on Product Judgments,” Effects of Advertising versus Direct Experience When Both
Journal of Marketing, April 67 (2), 35–48. Search and Experience Attributes Are Present,” Journal of
Peterson, Robert A., Sridhar Balasubramanian, and Bart J. Bron- Consumer Research, 21 (4), 708–718.

This content downloaded from

190.119.238.15 on Fri, 05 Apr 2019 15:01:55 UTC
All use subject to https://about.jstor.org/terms