Content uploaded by Joann Peck
Author content
All content in this area was uploaded by Joann Peck on Oct 12, 2017
Content may be subject to copyright.
UNCORRECTED PROOF
1
2If I touch it I have to have it: Individual and environmental influences
3on impulse purchasing
4Joann Peck
a,
⁎, Terry L. Childers
b
5a
University of Wisconsin-Madison, 3114 Grainger Hall, 975 University Avenue, Madison, WI 53706-1323, United States
6b
University of Kentucky, United States
7Received 1 April 2005; received in revised form 1 November 2005; accepted 1 January 2006
8Abstract
9This research examines the influence of touch on impulse-purchasing behavior. We first replicate the Rook and Fisher [Rook DW, Fisher RJ.
10 Normative influences on impulsive buying behavior. J Consum Res 1995;22:305–13.] studies about the moderating effect of the normative
11 evaluation of impulse purchase on impulse-purchasing behavior. Extending the impulse-purchasing literature, we examine individual differences
12 in touch and how they affect impulsive-buying behavior. Results from a field experiment suggest that both individual and environmental touch-
13 related factors increase impulse purchasing.
14 © 2006 Published by Elsevier Inc.
15
16 Keywords: Touch; Need for touch; Impulse purchase behavior
17
1819 Almost all unplanned buying is a result of touching,
20 hearing, smelling or tasting something on the premises of
21 the store (Underhill, 1999, p. 158).
22
23 This paper focuses on how elements of touch can affect
24 impulse purchasing. Buying impulsiveness is defined as a
25 consumer's tendency to buy spontaneously, unreflectively,
26 immediately, and kinetically. “Highly impulsive buyers are
27 more likely to experience spontaneous buying; their shopping
28 lists are more ‘open’and receptive to sudden, unexpected
29 buying ideas”(Rook and Fisher, 1995, p. 306). This research
30 has two primary purposes. First, the research is designed to
31 replicate the Rook and Fisher (1995) findings concerning the
32 moderating effect of the normative evaluation of impulse
33 purchase on an impulse-purchase trait and impulse-purchase
34 behavior. This research also extends previous research by
35 examining how the element of touch might affect impulse-
36 purchase behavior. Specifically, individual differences in
37 preferences for touch information are expected to relate to
38 impulse purchasing through their common link to hedonic
39 purchase motivations. In addition, encouragement to touch at
40 point-of-purchase is expected to influence impulse purchasing.
411. Theoretical background
421.1. Impulse purchasing and touch
43Limited evidence indicates that touch can influence
44behavior. In studies of the interpersonal touch domain (people
45touching people), restaurant servers who briefly touched
46customers received larger tips than servers who did not
47touch (Crusco and Wetzel, 1984; Hornik, 1992; Stephen and
48Zweigenhaft, 1986). Individuals who were asked to sign a
49petition were found to be more compliant if they were briefly
50touched (Willis and Hamm, 1980), and shoppers who were
51touched were more willing to participate in mall intercept
52interviews (Hornik and Ellis, 1988). While interpersonal touch
53seems to influence behavior, particularly compliance behavior,
54whether individual differences in touch will be related to
55impulse purchase behavior is not clear. Indirect evidence,
56however, suggests that product touch may influence impulse
57purchases, at least for some people. Kacen and Lee (2002)
58report that individuals who are more independent engage in
59greater impulse-purchase behavior than those who are
60interdependent in self-concept. Recent research by Rama-
61nathan and Menon (2002) also provides insight into the
62influence that touch may have on impulse purchasing. These
63researchers posit and find that individuals prone to impulsive
Journal of Business Research xx (2006) xxx –xxx
+ MODEL
JBR-06172; No of Pages 5
⁎Corresponding author. Tel.: +1 608 262 3603; fax: +1 608 262 0394.
E-mail address: jpeck@bus.wisc.edu (J. Peck).
0148-2963/$ - see front matter © 2006 Published by Elsevier Inc.
doi:10.1016/j.jbusres.2006.01.014
ARTICLE IN PRESS
UNCORRECTED PROOF
64 behavior are driven by hedonic gratification. What's more,
65 impulsive individuals are more inclined to pick up or touch a
66 hedonic target (in this case, a cookie) than are non-impulsives.
67 For instance, in study 2, the researchers report that 58 percent
68 of impulsives picked up the cookie, while only 29 percent of
69 non-impulsives touched the cookie. These results suggest that
70 individual differences in touch are potentially important as we
71 further our understanding of the antecedents of impulse-
72 purchase behavior.
73 1.2. Impulse purchase and autotelic NFT
74 “It is people, not products, who experience consuming
75 impulses”(Rook and Hoch, 1985, p. 23). The impulse-purchase
76 trait is characterized by the lack of a salient purchase goal, at
77 least at the start of the shopping experience. Researchers appear
78 to agree that impulse buying involves a hedonic component
79 (Cobb and Hoyer, 1986; Hausman, 2000; Rook, 1987; Rook
80 and Fisher, 1995; Thompson et al., 1990; Ramanathan and
81 Menon, 2002). Consumers report that when they purchase
82 impulsively they feel uplifted (Cobb and Hoyer, 1986; Rook,
83 1987), and that they experience their needs for fun and novelty
84 being fulfilled (Hausman, 2000). These studies offer conceptual
85 support for a link between hedonic shopping motives and
86 impulse-buying behavior.
87 Peck and Childers (2003) have reported individual differ-
88 ences in consumers' “need for touch”(NFT); i.e., their
89 preferences and motivations for gleaning information through
90 touch. While two components of NFT exist, the autotelic
91 component of NFT relates to touch as a hedonic-oriented
92 response seeking fun, arousal, sensory stimulation, and
93 enjoyment (Holbrook and Hirschman, 1982). In the absence
94 of a salient purchase goal, this autotelic component of touch
95 corresponds to a more sensory form of processing. Results from
96 two experiments indicate that individuals who report a
97 preference for autotelic touch chronically access hedonic
98 information from memory (Peck and Childers, 2003). Similarly,
99 Ramanathan and Menon (2002) argue that hedonic gratification
100 underlies most impulse behavior, and that for impulsives,
101 hedonic motives are more chronically accessible. Additionally,
102 a positive and significant correlation is reported between
103 autotelic NFT and an individual trait scale measuring buying
104 impulsiveness (Peck and Childers, 2003). By extension,
105 autotelic NFT would also be positively related to actual
106 impulse-purchase behavior, which leads to Hypothesis 1.
107 Hypothesis 1. Individuals higher in autotelic NFT will
108 purchase more impulsively than individuals lower in autotelic
109 NFT.
110 1.3. Impulse purchase and environmental salience of haptic
111 information
112 “Planning is a relative term; consumers' plans are sometimes
113 contingent and altered by environmental circumstances”(Rook,
114 1987, p. 191). Not only may individual characteristics increase
115 impulse purchasing, but also characteristics of the environment
116may affect impulse purchasing through increasing the salience
117of touch. The characteristics of the situation (Bloch and Richins,
1181983; Houston and Rothschild, 1978) may increase interest in
119differentiated aspects of the environment and thus capture the
120consumer's attention. As Underhill (1999) notes, many
121consumers are influenced or that they make their decisions in-
122store versus outside of the store.
123Unique aspects of the in-store environment, such as music,
124lighting, layout, and signage, may affect a consumer's decision
125process (Underhill, 1999). In particular, a point-of-purchase
126sign encouraging touch exploration may increase the salience of
127touch information motivating individuals to touch and impul-
128sively purchase the displayed product. Support for this comes
129from Ramanathan and Menon (2002) who report that impulsive
130behavior occurred for both impulsives and non-impulsives
131when a hedonic goal was primed. The chronic accessibility of
132hedonic gratification combined with the primed hedonic goal
133elevated impulsive behavior for impulsives, while also
134stimulating hedonic gratification for non-impulsives as well.
135Thus, we expect that increasing the environmental salience of
136touch will stimulate increased impulse purchasing for both
137higher and lower autotelic NFT individuals. This leads to
138Hypothesis 2.
139Hypothesis 2. Increasing the environmental salience of touch
140will increase impulse purchasing for higher and lower autotelic
141NFT individuals.
1422. Overview of study
143This study was designed to investigate the link between
144impulse purchasing and both an environmental encouragement
145to touch and an individual preference for autotelic touch. This
146study was also designed to replicate the Rook and Fisher
147(1995) findings concerning the relationship between the
148impulse-buying trait and impulse-buying behavior. The design
149was a 2 (high versus low autotelic NFT) × 2 (point-of-
150purchase sign “feel the freshness”, or no sign) between-
151subjects design.
1522.1. Procedure
153This study took place in two parts. Part one consisted of a
154field experiment conducted in a Midwestern-city supermarket
155where shoppers were observed while they purchased peaches or
156nectarines. Shoppers who purchased at least one peach or
157nectarine were intercepted and asked to fill out a short half-page
158survey. (Only two shoppers approached the display but did not
159purchase the fruit.) The first part of the survey measured
160shoppers' level of impulse purchase. This survey also included a
161manipulation check to determine whether shoppers noticed the
162point-of-purchase sign we had displayed. Finally, shoppers
163were asked their name and address for a two-page follow-up
164survey. As an incentive to return the second part of the survey,
165shoppers were entered into a drawing to win a $100 (U.S.) gift
166certificate toward supermarket purchases. This part of the study
167was completed in 3 weeks. The follow-up survey, which was
2J. Peck, T.L. Childers / Journal of Business Research xx (2006) xxx–xxx
ARTICLE IN PRESS
UNCORRECTED PROOF
168 mailed, included the autotelic NFT scale, the buying-impul-
169 siveness trait scale (Rook and Fisher, 1995), the normative
170 evaluation of impulse purchase of peaches/nectarines, and
171 demographic measures.
172 2.2. Sample
173 Two hundred and thirty-nine shoppers participated in part 1
174 of the study. After 2 weeks, 173 surveys were returned, with
175 three unnamed, resulting in a usable sample size of 170
176 shoppers, for a response rate of 71 percent. The median age
177 category for the respondents was 35–44 years. The median
178 education level was a bachelor's degree (22 percent). The
179 annual household income of the shoppers ranged from under
180 $10,000 (U.S.) per year to over $100,000 per year. The median
181 annual household income was $60,000 to $69,999 (U.S.).
182 Thirty-eight members of the sample were male (22 percent).
183 2.3. Independent variables
184 2.3.1. Environmental touch salience
185 Environmental touch salience was manipulated by either
186 posting a sign encouraging shoppers to “feel the freshness”,or
187 by posting no sign over the fruit display. The sign followed the
188 normal sizing for this supermarket so as not to be conspicuous,
189 and measured only 9 × 6 in.
190 2.3.2. Autotelic NFT
191 Autotelic NFT was measured using the six-item autotelic
192 NFT scale (Peck and Childers, 2003) with sample items:
193 “Touching products can be fun”, and “I find myself touching all
194 kinds of products in stores”(α= .94). Scale item descriptors
195 ranged from −3 (strongly disagree) to +3 (strongly agree) with
196 the entire range represented in the sample. Higher and lower
197 autotelic NFT were divided by a median split (eighty-seven
198 individuals below the median were classified as lower in
199 autotelic NFT vs. eighty-three classified as higher in autotelic
200 NFT).
201 2.3.3. Buying-impulsiveness trait
202 The buying-impulsiveness trait was measured by the nine-
203 item buying-impulsiveness scale (α= .74) developed by Rook
204 and Fisher (1995).
205 2.3.4. Normative evaluation of impulse purchase
206 The normative evaluation measure assumes that consumers
207 may assess the appropriateness of buying something on impulse
208 along a continuum that ranges from relative neutrality to strong
209 disapproval or encouragement. The measure, also adapted from
210 Rook and Fisher (1995), contained the following question:
211 “You are planning to buy one type of fruit and you end up
212 buying four types of fruit. How would this make you feel?”The
213 shopper is given ten seven-point semantic differential scales
214 (α= .88) with endpoints good–bad, rational–crazy, wasteful–
215 productive, attractive–unattractive, smart–stupid, acceptable–
216 unacceptable, generous–selfish, sober–silly, mature–childish,
217 right–wrong.
2182.4. Dependent variables
2192.4.1. Actual impulse purchase behavior
220In-store buying impulsiveness was measured using three
221items adapted from Rook and Fisher (1995). The first item
222stated, “My decision to buy some type of fruit today was ––.”
223The second item stated, “My decision to buy peaches/nectarines
224today was ––.”And the third item stated, “My decision to buy
225the exact number of peaches/nectarines that I ended up
226purchasing was ––.”All three items had a scale ranging
227from zero to 4, with zero being “completely planned”to 4 being
228“completely unplanned.”The three items were summed
229(α= .72) for a measure of buying impulsiveness.
2303. Results
2313.1. Replication of Rook and Fisher (1995)
232One purpose of this study was to replicate Rook and Fisher
233(1995) concerning the relationship between the impulse-buying
234trait and consumers' buying behaviors. Rook and Fisher (1995)
235found that consumers' normative evaluations moderated the
236degree or strength of the relationship between the buying
237impulsiveness trait and impulse-buying behavior. This study
238replicated the analyses used by Rook and Fisher (1995). The
239mean normative evaluation of purchasing four types of fruit in
240this study was 24.7, a finding that was slightly lower but
241comparable to those results obtained by Rook and Fisher (1995)
242in which study 1 found a mean of 30.4, and study 2 found means
243of 28.1 for the sweater and 28.7 for the CD. The reliability of the
244normative evaluation scale in this grocery store study was also
245comparable to the reliability obtained in the Rook and Fisher
246(1995) paper: for this study α= .88, while for the Rook and
247Fisher findings, in study 1 α= .91, and in study 2 α= .90. A
248median split on shoppers' normative evaluations divided the
249sample into favorable (normative evaluation greater than or
250equal to 26, n= 79) and unfavorable (norm ative evaluation less
251than 26, n= 90) subsets.
252Next, we compared product moment correlations across
253normative subgroups. In the normatively favorable group, the
254correlation between the buying-impulsiveness trait and actual
255impulse-buying behavior was significant (r= .29, pb.05); yet,
256in the normatively unfavorable group, the correlation was
257insignificant (r=.06, pN.05). A Fisher's z-transformation
258revealed that the two correlations differed significantly
259(z= 2.36, pb.05). This finding replicates Rook and Fisher
260(1995) in both their study 1 and their study 2 (see Table 1).
261Additionally, Rook and Fisher (1995) used a different basis
262for defining normative groups to examine the robustness of their
263findings. They divided the sample into three groups and again
264computed the within-group correlations for both their study 1
265and their study 2. Our observational study, conducted in a
266grocery store setting, again replicated their findings that
267consumers' normative evaluations moderated the degree or
268strength of the relationship between the buying-impulsiveness
269trait and impulse-buying behavior. Our results are illustrated in
270Table 1.
3J. Peck, T.L. Childers / Journal of Business Research xx (2006) xxx–xxx
ARTICLE IN PRESS
UNCORRECTED PROOF
271 3.2. Impulse purchase and autotelic NFT
272 The first hypothesis predicted that individuals higher in
273 autotelic NFT would purchase more impulsively than indivi-
274 duals lower in autotelic NFT. Hypothesis 1 was supported with
275 a significant main effect for autotelic NFT on impulse purchase
276 (M= 4.5 and M= 5.5 for lower and higher autotelic NFT,
277 respectively, F[1,166] = 4.9, pb.05). The two-way interaction
278 was not significant (pN.05). In both the no-sign and the “feel-
279 the-freshness”conditions, individuals higher in autotelic NFT
280 purchased more impulsively than individuals lower in autotelic
281 NFT. (In the no-sign condition, Mvalues = 4.6 and 3.8 for high
282 and low NFT, respectively, F[1,166] = 3.2, pb.05; in the “feel-
283 the-freshness”condition, Mvalues = 6.4 and 5.4, F[1,166] = 3.2,
284 pb.05; Fig. 1.)
285 3.3. Impulse purchase and environmental salience of touch
286 information
287 We predicted that when a point-of-purchase sign (“feel the
288 freshness”) encouraged shoppers externally to touch, both high
289 and low autotelic NFT shoppers would purchase more
290 impulsively. This was supported with a main effect of
291 environmental salience (F[1,166] = 10.9, pb.05). Individuals
292 purchased more impulsively in the “feel-the-freshness”versus
293 the “no-sign”conditions (M= 5.9 vs. 4.1). We expected that
294 higher and lower autotelic individuals would be influenced by
295 the point-of-purchase sign. As expected, both higher and lower
296 autotelic NFT individuals purchased significantly more impul-
297sively in the “feel-the-freshness”versus the no-sign condition
298(for higher autotelic NFT, Mvalues= 4.6 and 6.4 for no sign and
299“feel the freshness”, respectively, F[1,166] = 6.0, pb.05; for
300lower autotelic NFT, Mvalues= 3 .8 and 5.4, F[1,166] = 4.9, Fig.
3011). These results support Hypothesis 2 and indicate that both
302higher and lower autotelic NFT individuals were influenced by
303the presence of the sign increasing the environmental salience of
304touch information.
3054. General discussion
3064.1. Summary of findings
307This study examines the relationship between impulse
308purchase and the individual difference in autotelic NFT, as
309well as an environmental encouragement to touch. In addition,
310the study replicated the research of Rook and Fisher (1995) with
311the correlation between the impulse-purchase trait and impulse-
312purchase behavior moderated by the normative evaluation of the
313impulse-purchase behavior. Results are consistent with expecta-
314tions. Overall, individuals higher in autotelic NFT purchased
315more impulsively than their lower autotelic NFT counterparts.
316In addition, for both higher and lower autotelic individuals, the
317environmental salience of touch information induced by the
318“feel-the-freshness”point-of-purchase sign increased impulse-
319purchasing behavior.
3204.2. Theoretical and managerial implications
321While all individuals were influenced by increasing the
322environmental salience of touch information, some individuals
323(those higher in autotelic NFT) had a higher impulse-purchase
324baseline; that is, they were more likely to make impulse
325purchases overall. Puri (1996) describes impulsiveness as a
326result of the relative accessibility of the costs and benefits of
327impulsiveness. Perhaps individuals higher in NFT have the
328benefits of touch more accessible in memory than those lower in
329NFT. Evidence shows that touch information in general is more
330accessible to those who are higher versus lower in their NFT
331(Peck and Childers, 2003). We could argue that buying peaches
332or nectarines impulsively has minimal costs for individuals.
333Because of this, the accessibility of the fun and benefits of touch
334may drive impulse purchase. An accessibility explanation
335supports the finding that higher NFT individuals purchase more
6.4
4.6
5.4
3.8
0
1
2
3
4
5
6
7
NO SIGN "FEEL THE FRESHNESS"
POINT OF PURCHASE SIGN
BUYING IMPULSIVENESS
HIGH AUTOTELIC NFT LOW AUTOTELIC NFT
Fig. 1. Buying impulsiveness by autotelic NFT and point of purchase sign.
Table 1t1:1
Correlations between impulse buying trait and actual impulse purchaset1:2
t1:3Normatively unfavorable,
median split
Normatively favorable,
median split
Normatively
unfavorable
Normatively
neutral
Normatively
favorable
t1:4Rook and Fisher, 1995: Study 1 r=−.002 r= .33 r= .08 r= .10 r= .36
t1:5pN.10 pb.01 pN.10 pN.10 pb.01
t1:6n= 102 n=110 n=74 n=69 n=69
t1:7Rook and Fisher, 1995: Study 2 r=−.02 r= .36 r= .07 r= .03 r= .58
t1:8pN.10 pb.01 pN.10 pN.10 pb.001
t1:9n=48 n=52 n=35 n=33 n=33
t1:10 This study r= .06 r= .29 r= .05 r= .17 r= .33
t1:11 pN.10 pb.01 pN.10 pN.10 pb.01
t1:12 n=90 n=79 n=64 n=59 n=46
4J. Peck, T.L. Childers / Journal of Business Research xx (2006) xxx–xxx
ARTICLE IN PRESS
UNCORRECTED PROOF
336 impulsively than lower NFT individuals, and the result that
337 increasing the environmental salience of touch information
338 increases impulse purchasing. An interesting extension of this
339 finding would be to repeat this study using a purchase in which
340 the costs would be determined to be greater, perhaps by
341 manipulating different types of products. In this case, the
342 difference in impulse purchasing between higher and lower
343 autotelic NFT individuals may be even more pronounced.
344 Another possible explanation involves the relative influence
345 of affect and cognition. Shiv and Fedorikhin (1999) found that
346 in a decision-making task, if processing resources are limited,
347 affective reactions that are evoked spontaneously have a greater
348 impact on the decision than do cognitions. A grocery store
349 environment could be argued to be a cognitively demanding
350 environment where resources are limited. Shoppers high in
351 autotelic NFT may experience stronger affective reactions
352 relating to the touch experience than those lower in NFT, which
353 may in turn drive the increased level of buying impulsiveness.
354 Examining physiological measures of higher and lower
355 autotelic NFT individuals when they touch products that
356 provide pleasant sensory feedback would be a method to
357 investigate this.
358 For managers, the link between touch and impulse purchase
359 has important implications. Touch in general was found to
360 increase impulse purchasing. Because of this, point-of-purchase
361 signs, displays, and packaging encouraging product touch may
362 increase impulse purchasing for both low and high NFT
363 shoppers. A note of caution is necessary. This research only
364 investigated the link between impulse purchase and product
365 touch for a product high in salience-of-touch attributes. Whether
366 this would translate for a product moderately high or low in
367 touch-attribute salience is not clear. However, increasing the
368 opportunities for consumers to touch products through both in-
369 store displays and store layout may increase impulse purchase.
370 This research replicated Rook and Fisher (1995) and
371 extended research on impulse purchasing by looking at the
372 role of touch and its relationship to impulse purchase. Both an
373 individual touch variable (autotelic NFT) and an environmental
374 touch variable (point-of-purchase sign encouraging touch)
375 increased impulse purchasing. Additional research is required
376 to examine further the mechanism by which touch leads to
377 impulse purchase.
378References
379Bloch P, Richins ML. A theoretical model for the study of product importance
380perceptions. J Mark 1983;47:69–81.
381Cobb CJ, Hoyer WD. Planned versus impulse purchase behavior. J Retail
3821986;62:67–81 [Winter].
383Crusco AH, Wetzel CG. The Midas Touch: the effects of interpersonal touch on
384restaurant tipping. Pers Soc Psychol Bull 1984;10(4):512–7.
385Hausman A. A multi-method investigation of consumer motivations in impulse
386buying behavior. J Consum Mark 2000;17(5):403–19.
387Holbrook MB, Hirschman EC. The experiential aspects of consumption:
388consumer fantasies, feelings, and fun. J Consum Res 1982;9(2):132–40.
389Hornik J. Tactile stimulation and consumer response. J Consum Res 1992;
39019:449–58.
391Hornik J, Ellis E. Strategies to secure compliance for a mall intercept interview.
392Public Opin Q 1988;52(4):539–51.
393Houston MJ, Rothschild ML. Conceptual and methodological perspectives on
394involvement. In: Jain S, editor. 1978 Educator's proceedings. Chicago (IL):
395American Marketing Association; 1978. p. 184–7.
396Kacen JJ, Lee JA. The influence of culture on consumer impulsive buying
397behavior. J Consum Psychol 2002;12(2):163–76.
398Peck J, Childers TL. Individual differences in haptic information processing: the
399“need for touch”scale. J Consum Res 2003;30(3):430–42.
400Puri R. Measuring and modifying consumer impulsiveness: a cost–benefit
401accessibility framework. J Consum Psychol 1996;5(2):87-113.
402Ramanathan S, Menon G. Don't know why, but I had this craving: goal-
403dependent automaticity in impulsive decisions. Working paper. New York
404University; 2002.
405Rook DW. The buying impulse. J Consum Res 1987;14:189–99 [September].
406Rook DW, Fisher RJ. Normative influences on impulsive buying behavior.
407J Consum Res 1995;22:305–13.
408Rook DW, Hoch SJ. Consuming impulses. In: Holbrook MB, Hirschman EC,
409editors. Advances in Consumer Research, vol. 12. Provo (UT): Association
410for Consumer Research; 1985. p. 23–7.
411Shiv B, Fedorikhin A. Heart and mind in conflict: the interplay of affect and
412cognition in consumer decision making. J Consum Res 1999:278–92
413[December].
414Stephen R, Zweigenhaft RL. The effect on tipping of a waitress touching male
415and female customers. J Soc Psychol 1986;126:141–2.
416Thompson CJ, Locander WB, Pollio HR. The lived meaning of free choice: an
417existential–phenomenological description of everyday consumer experi-
418ences of contemporary married women. J Consum Res 1990;17:346–61
419[December].
420Underhill P. Why we buy: the science of shopping. New York (NY): Simon and
421Schuster; 1999.
422Willis FN, Hamm HK. The value of interpersonal touch on securing compliance.
423J Nonverbal Behav 1980;5(1):49–55.
424
425
5J. Peck, T.L. Childers / Journal of Business Research xx (2006) xxx–xxx
ARTICLE IN PRESS