want a professional writer for a summaries
RoRo
rickcryderloewenstein2007.pdf
767
! 2007 by JOURNAL OF CONSUMER RESEARCH, Inc. " Vol. 34 " April 2008 All rights reserved. 0093-5301/2008/3406-0003$10.00
Tightwads and Spendthrifts
SCOTT I. RICK CYNTHIA E. CRYDER GEORGE LOEWENSTEIN*
Consumers often behave differently than they would ideally like to behave. We propose that an anticipatory pain of paying drives “tightwads” to spend less than they would ideally like to spend. “Spendthrifts,” by contrast, experience too little pain of paying and typically spend more than they would ideally like to spend. This article introduces and validates the “spendthrift-tightwad” scale, a measure of in- dividual differences in the pain of paying. Spending differences between tightwads and spendthrifts are greatest in situations that amplify the pain of paying and smallest in situations that diminish the pain of paying.
They were so skewed and squint-eyed in their minds, their misering or extravagance mocked all reason. (Dante’s Inferno, “Canto VII: The Hoarders and the Wasters”)
E conomic models of decision making are consequen-tialist in nature. They assume that decision makers choose between alternative courses of action based on a cognitive evaluation of the desirability (i.e., “utility”) and likelihood of their consequences. This does not, however, imply that consequentialist decision makers are devoid of
*Scott I. Rick ([email protected]) is a visiting professor of operations and information management at the Wharton School, Uni- versity of Pennsylvania, Philadelphia, PA 19104. Cynthia E. Cryder ([email protected]) is a doctoral student, and George Loewen- stein ([email protected]) is the Herbert A. Simon Professor of Eco- nomics and Psychology, both at the Department of Social and Decision Sciences, Carnegie Mellon University, Pittsburgh, PA 15213. This article is based on the first author’s dissertation. For helpful comments, the authors thank the editor, the associate editor, three anonymous reviewers, Dan Ariely, Eloise Coupey, Robyn Dawes, Michael DeKay, J. Wesley Hutchinson, Eric Johnson, Uzma Khan, Jennifer Lerner, Julie Ozanne, Kathleen Vohs, Joachim Vosgerau, Roberto Weber, Christian Wheeler, Patti Williams, Gal Zauberman, and participants at the 2005 Society for Judgment and Decision Making conference in Toronto, the 2006 Judg- ment and Decision Making preconference at Society for Personality and Social Psychology in Palm Springs, the 2006 Behavioral Decision Re- search in Management conference in Los Angeles, the Second Annual Whitebox Advisors Graduate Student Conference at Yale, and the 2007 Society for Consumer Psychology conference in Las Vegas. They also thank NBC’s WCAU affiliate, the Globe and Mail, and John Tierney of the New York Times for their invaluable assistance in collecting data. This research was supported in part by grants from the Center for Be- havioral Decision Research at Carnegie Mellon and the Russell Sage Foundation, an NSF Graduate Research Fellowship to Rick, and a Mac- Arthur Foundation network grant to Loewenstein.
John Deighton served as editor and Baba Shiv served as associate editor for this article.
Electronically published October 19, 2007
emotion or immune to its influence. To see why, it is useful to draw a distinction between “expected” and “immediate” emotions (Loewenstein et al. 2001; Loewenstein and Lerner 2003; Rick and Loewenstein, forthcoming).
Expected emotions are those that are anticipated to occur as a result of the outcomes associated with different possible courses of action. For example, in deciding whether to pur- chase a candy bar, a consumer might imagine the pleasure she would feel while eating it and possibly the guilt she would feel after indulging. The key feature of expected emo- tions is that they are experienced when the outcomes of a decision materialize, but not at the moment of choice; at the moment of choice they are only cognitions about future emotions.
Immediate emotions, like expected emotions, can arise from thinking about the future consequences of one’s de- cision. However, unlike expected emotions, immediate emo- tions are experienced at the moment of choice. For instance, when deciding whether to purchase the candy bar, the con- sumer might immediately feel pangs of guilt at the thought of consuming all those calories.
A role for expected emotions in decision making is per- fectly consistent with the consequentialist perspective of eco- nomics. There is nothing in the notion of utility maximization that rules out the idea that the utility an individual associates with an outcome might arise from a prediction of emo- tions—for example, one might assign higher utility to an Italian restaurant dinner than a French restaurant dinner be- cause one anticipates being happier at the former. By contrast, consequentialist decision makers are assumed to be immune to the influence of immediate emotions; such emotions are presumably “epiphenomenal” by-products of, but not deter- minants of, decisions (Loewenstein et al. 2001; Loewenstein and Lerner 2003; Rick and Loewenstein, forthcoming).
An implication of this consequentialist perspective in the domain of consumer choice is that prices are assumed to
768 JOURNAL OF CONSUMER RESEARCH
deter spending only through thoughts of forgone pleasure. That is, according to the standard economic account of in- tertemporal choice (Fisher 1930), people choose to consume immediately if the anticipated benefits of doing so exceed the forgone (discounted) benefits of future consumption. The price of a good captures the amount of future pleasure that must be sacrificed to finance immediate consumption.
The descriptive validity of this perspective rests on peo- ple’s inclination (or ability) to think of prices in terms of opportunity costs. The relationship between price and opportunity cost is frequently assumed to be transparent (Becker, Ronen, and Sorter 1974; Okada and Hoch 2004). A recent study by Frederick et al. (2006), however, suggests that many people do not spontaneously consider opportunity costs when making purchasing decisions. In one experiment, Frederick et al. (2006) asked participants whether they would be willing to purchase a desirable video for $14.99. All that varied was simply whether the option of not buying was framed as “do not buy” or “keep the $14.99 for other purchases.” Drawing people’s attention to the pleasure that is forgone by consuming immediately significantly reduced their willingness to buy the video, suggesting that many people do not spontaneously perceive prices in a manner consistent with standard economic theory (see also Jones et al. 1998).
One reason why opportunity costs do not spontaneously come to mind may be that cognitive constraints simply make it too difficult to determine what exactly is forgone by consuming immediately. If people relied solely on cog- nitively nebulous representations of forgone consumption, most people would likely spend compulsively. Vague no- tions of forgone pleasures are unlikely to provide com- pelling motivation to control current spending (Loewen- stein and O’Donoghue 2006). One way consumers can solve this problem is the cultivation of negative emotions in re- sponse to the prospect of spending. Prelec and Loewenstein (1998; see also Zellermayer 1996) propose that consumers rely on an immediate “pain of paying” to control their spend- ing. Doing so would likely simplify decision making; instead of comparing the pleasure of consuming immediately to the anticipated pleasure of consuming later, consumers can in- stead compare immediate pleasure to immediate pain.
Knutson et al. (2007) examined whether an immediate pain of paying deters spending in an experiment in which partic- ipants chose whether or not to purchase a series of consumer goods while having their brains scanned with functional mag- netic resonance imaging (fMRI). In each trial, participants first saw the available product, then saw its price, and finally decided whether or not to purchase it. As soon as participants saw the price, activation in the insula—a region previously associated with experiencing a variety of painful stimuli such as disgusting odors (Wicker et al. 2003), unfair ultimatum game offers (Sanfey et al. 2003), and social exclusion (Ei- senberger, Lieberman, and Williams 2003)—was signifi- cantly greater for products that were ultimately not pur- chased than for products that were ultimately purchased. Thus, when opportunity costs are not explicitly represented,
people appear to rely on an anticipatory pain of paying to deter their spending.
Of course, given the massive amount of credit card debt accrued by many Americans (Bucks, Kennickell, and Moore 2006), one would not expect all people to feel the pain of paying intensely. Indeed, people likely differ in their ten- dency to experience the pain of paying, and these individual differences likely have important behavioral implications. At sufficiently high levels, the pain of paying may deter spending even more than would a deliberative (i.e., con- sequentialist) consideration of the pleasures that are forgone by consuming immediately (see Loewenstein, Rick, and Co- hen [2008] for a review of neuroeconomic research sug- gesting that affective and deliberative desires often conflict). Suppose, for example, that dining out at a nice restaurant tonight requires you to forgo dining out at an even nicer restaurant next month. People who experience an intense pain of paying may behave as if dining out tonight requires giving up several nicer dinners next month. That is, their affective reaction to spending may lead them to spend less than their more deliberative selves would prefer. We refer to such consumers as tightwads.
By contrast, at sufficiently low levels, the pain of paying may deter spending less than would a deliberative consid- eration of forgone pleasures. In the scenario above, people who experience minimal pain of paying may behave as if dining out tonight requires giving up nothing next month. That is, the failure to feel the pain of paying may lead these consumers to spend more than their consequentialist selves would prefer. We refer to such consumers as spendthrifts.
At intermediate levels, the pain of paying may produce behavior consistent with deliberative considerations of for- gone pleasures. That is, people who experience some mod- erate amount of pain of paying may behave as if dining out tonight requires giving up exactly one dinner at an even nicer restaurant next month. Such “unconflicted” consumers should therefore tend to spend about as much as their more deliberative selves would prefer.
Of course, note that considering the implications of in- tense pain of paying is only important if tightwads represent a substantial portion of the population. This might seem unlikely given the intense attention toward impulsive spend- ing in the media and the academic literature (Baumeister 2002; Faber and O’Guinn 1992; Hoch and Loewenstein 1991; O’Guinn and Faber 1989; Rook 1987; Rook and Fisher 1995; Stern 1962; Valence, d’Astous, and Fortier 1988; Vohs and Faber 2007; Weun, Jones, and Beatty 1998). However, people are not uniformly impulsive across situ- ations; for example, utility from savoring and dread moti- vates many to delay good outcomes and accelerate bad out- comes. Loewenstein (1987) found that, on average, people were willing to pay more to obtain a kiss from a movie star of their choice when that kiss was delayed by three days than when it was immediately obtainable.
More recent research has found that some people are often frustratingly unable to indulge themselves. Kivetz and Si- monson (2002), for example, note that some “hyperopic”
TIGHTWADS AND SPENDTHRIFTS 769
consumers who are excessively farsighted require commit- ment devices to indulge themselves. For instance, they found that most women who hypothetically chose a spa package valued at $80 over $85 in cash said they did so because they feared they would otherwise use the cash on more utilitarian expenditures such as rent or groceries.
Ameriks et al. (2003) also found evidence of (anticipated) underindulgence in a recent survey of TIAA-CREF clients. They asked respondents to imagine that they had been given 10 gift certificates that were each redeemable for a fancy dinner. The gift certificates expire in 2 years, and respon- dents were asked how many they would ideally like to use during the first year and how many they anticipate actually using during the first year. While many people thought they would actually use more than they would ideally like to use during the first year, Ameriks et al. (2003) found that even more people thought they would actually use less than they would ideally like to use during the first year. While the above studies did not address spending per se, they do sug- gest that some people find it painful to indulge.
This article introduces and validates a spendthrift-tight- wad (ST-TW) scale that measures individual differences in the tendency to experience a pain of paying.1 While it would be ideal to measure such differences directly (e.g., via brain imaging), such methods are currently too costly to be ef- ficient. Questionnaire measures of the pain of paying are less costly, although they face their own set of challenges. People are not always aware of their emotional processes; as LeDoux (1996) notes, conscious feelings are merely the tip of the emotional iceberg. Moreover, even if people had perfect access to their emotional processes, they may not always be completely forthcoming about them. Our scale, therefore, measures individual differences in the pain of paying somewhat indirectly. Rather than asking respondents to introspect regarding the emotions they experience while shopping, we ask them to indicate the extent to which their typical spending habits diverge from their desired spending habits. If individual differences in the pain of paying pro- duce divergence between typical and desired spending hab- its, then self-reports of that divergence should serve as an appropriate proxy for the pain of paying.
We begin by introducing the ST-TW scale and evaluating its reliability. We then evaluate its discriminant validity by assessing its relationship with 28 potentially related scales from the economics, psychology, and marketing literatures. The measure that is most closely related to ours is the fru- gality scale of Lastovicka et al. (1999, 88), who concep- tualize frugality as a “unidimensional consumer lifestyle trait
1The scale consists of four items from a large questionnaire developed by Prelec et al. (1997), who were also interested in whether individuals differ in their tendency to experience a pain of paying. For each item in their questionnaire, they examined whether the proportion of participants endorsing responses suggestive of “tightwaddism” was different than the proportion of participants endorsing responses suggestive of “spendthrift- iness.” We extend their analysis by demonstrating that four of the items from their questionnaire form a coherent scale. In this article we test the validity of this scale, demonstrate that it predicts economically important behaviors, and examine the moderating role of situational factors.
characterized by the degree to which consumers are both restrained in acquiring and in resourcefully using economic goods and services to achieve longer-term goals.” Tightwads and the highly frugal may therefore look similar in terms of spending, but frugality-scale items such as “making better use of my resources makes me feel good” (Lastovicka et al. 1999, 89) suggest that the psychological mechanism un- derlying individual differences in frugality is distinct from the psychological mechanism underlying individual differ- ences along the ST-TW scale. We later present evidence suggesting that the highly frugal spend conservatively be- cause they enjoy saving, not because the prospect of spend- ing pains them.
We then examine the relationship between the ST-TW scale and actual spending. While we find strong relationships between ST-TW scores and credit card debt and savings, we find little relationship between ST-TW scores and in- come. This suggests that the differences in credit card debt and savings between spendthrifts and tightwads are more likely attributable to differences in spending habits than to differences in income.
Of course, individual differences are not all-powerful de- terminants of behavior. Our theoretical framework does not predict that spendthrifts will spend more than tightwads across all domains. If tightwads are particularly prone to experience the pain of paying, they should spend less when situational factors intensify the pain of paying than when situational factors mitigate the pain of paying. If spendthrifts are not particularly prone to experience the pain of paying (and thus less sensitive than tightwads to such situational factors), spending differences between tightwads and spend- thrifts should be greatest when situational factors intensify the pain of paying. By contrast, such spending differences should be smallest when situational factors mitigate the pain of paying. We conclude by presenting two experiments that test this critical prediction.
DEVELOPMENT AND VALIDATION OF THE SPENDTHRIFT-TIGHTWAD SCALE
When considering items to include in our scale, we con- sulted a survey previously developed by Prelec, Loewen- stein, and Zellermayer (1997) and administered to two sam- ples of passengers waiting to board flights at airports. The survey included a variety of questions about respondents’ spending habits, including several that appeared to measure the divergence between one’s typical spending habits and one’s desired spending habits. We selected four items from this survey that, based on face validity, appeared to precisely capture this divergence. These four items comprise the ST- TW scale presented in the appendix.
We administered the ST-TW scale to 13,327 respondents over a 31-month period beginning in October 2004. Re- spondents were drawn from four populations. Nearly one- fifth of all respondents ( ) were students at Car-N p 2,649 negie Mellon or the University of Pittsburgh, parents of students, or staff members. Most respondents (N p
770 JOURNAL OF CONSUMER RESEARCH
TABLE 1
SPENDTHRIFT-TIGHTWAD DISTRIBUTIONS BY SAMPLE (%)
Globe and Mail ( )n p 154
New York Times ( )n p 10,331
Pittsburgh ( )n p 2,649
NBC ( )n p 193 Total
Tightwad 36 25 21 19 24 Unconflicted 57 60 61 52 60 Spendthrift 6 15 18 29 15
NOTE.—Some columns do not sum to 100% due to rounding errors.
TABLE 2
DEMOGRAPHIC CHARACTERISTICS BY SAMPLE
Globe and Mail New York Times NBC
Gender (female p 1) .28a .46b .77c Age 43.09a 39.75b 43.30a Education (1bachelor’s p 1) .47a .64b .27c
NOTE.—Means within a row that have different subscripts differ at the level.p ! .01
) were readers of the New York Times. On January10,331 16, 2007, New York Times columnist John Tierney (2007) wrote a piece for the Science Times section that discussed another article coauthored by two of the present authors (Knutson et al. 2007) and included a link that interested readers could click to take our survey. A small number of respondents ( ) were viewers of a nightly newsN p 193 broadcast in Philadelphia. On February 5, 2007, NBC’s WCAU affiliate ran a story on tightwads and spendthrifts and referred viewers to their Web site, which featured a link to our survey. The remaining respondents ( ) wereN p 154 readers of the Globe and Mail, one of Canada’s most widely circulated newspapers. On April 27, 2007, Globe and Mail columnist Carolyn Abraham (2007) wrote a piece for the Report on Business section that discussed Knutson et al. (2007) and included a link that interested readers could click to take our survey.
Exploratory factor analysis of the four items yielded one factor with an eigenvalue greater than one. Confirmatory factor analysis using SAS PROC CALIS subsequently sug- gested that a single-factor model fit the data well, with a goodness-of-fit index of .99, a Bentler’s Comparative Fit Index of .97, and a normed fit index of .97. The factor- loading estimates (item 1, .99; item 2a, .62; item 2b, .54; item 3, .47) were all significant, with all t-statistics exceed- ing 54 ( ).p ! .001
We therefore simply summed scale responses for each participant as a measure of their location on the ST-TW scale. Since scale sums can possibly take on 23 different values (sums range from 4 to 26), we divide the scores as closely as possible into three equally sized groups of sums. We classify tightwads as those with scale sums from 4 to 11, “unconflicted” consumers as those with scale sums from 12 to 18, and spendthrifts as those with scale sums from 19 to 26. We employ a trichotomized division instead of a dichotomized one (i.e., tightwad or spendthrift) because we
believe that many consumers experience minimal diver- gence between their actual and desired spending habits. We propose that this lack of conflict is reflected in ST-TW scores that minimally diverge from the midpoint of the scale (15).
This criterion results in a larger number of tightwads than spendthrifts; in fact, tightwads outnumber spendthrifts by a 3 : 2 ratio (3,248 tightwads vs. 2,046 spendthrifts). The mean ST-TW score of 14.38 was significantly less than the scale midpoint ( ; ), and thet(13,326) p 18.35 p ! .000001 distribution of ST-TW scale scores was significantly skewed ( ). This is a surprising finding given the intensep ! .001 attention to impulsive spending in the media and the aca- demic literature.
However, note that determining whether “tightwaddism” is more prevalent than “spendthriftiness” depends heavily on the populations from which participants are drawn. Table 1 shows the number of tightwads, unconflicted consumers, and spendthrifts in each sample. Note, for example, that tightwads outnumber spendthrifts by 30% in the Globe and Mail sample ( ; ), whereas spend-2x (1) p 40.80 p ! .0001 thrifts outnumber tightwads by 10% in the NBC sample ( ; ).2x (1) p 5.11 p ! .03
It is worth examining whether demographic measures can account for sample differences in ST-TW distributions. Al- though we collected little demographic information from the student-heavy Pittsburgh sample, we included several demographic measures in our other surveys. We find that three measures differ across these samples: gender, age, and education. Table 2 presents mean demographic values by sample.
Before examining whether these measures mediate the relationship between sample and ST-TW scores, we first examine their relationship with ST-TW scores. Across all samples, we collected gender data from 10,912 respondents (47.6% female). Females are no more likely to be tightwads than spendthrifts (20% (1,018/5,195) vs. 19% (1,012/5,195);
TIGHTWADS AND SPENDTHRIFTS 771
FIGURE 1
MEAN ST-TW SCORES BY AGE GROUP
NOTE.—Error bars represent standard errors of the mean.
TABLE 3
RELATIONSHIP BETWEEN ST-TW SCORES AND SAMPLE AND DEMOGRAPHICS
Model 1 Model 2 Model 3 Model 4 Model 5
NBC .05*** .03** .05*** .04*** .03** Globe and Mail !.04*** !.03** !.04*** !.04*** !.03** Gender . . . .17*** . . . . . . .16*** Age . . . . . . !.06*** . . . !.04*** Education . . . . . . . . . !.06*** !.04*** R 2 .004 .031 .008 .007 .034
NOTE.—Regression weights are standardized. ** .p ! .01 *** .p ! .001
), but males are more than two and a half times2x (1) p .02 more likely to be tightwads than spendthrifts (29% (1,673/ 5,717) vs. 11% (636/5,717); ; ).2x (1) p 583.6 p ! .0001
We collected age data from 10,760 respondents (range, 18–100; mean, 38.3). The zero-order correlation between age and ST-TW scores is small: !.07 (t(10,758) p !7.35; ). However, this correlation hides the non-p ! .0001 linear pattern shown in figure 1. Among the 187 respondents aged 71 and over, tightwads outnumber spendthrifts 49 to 9 ( ; ). Indeed, tightwads on average2x (1) p 32.7 p ! .0001 are over three years older than spendthrifts ( ,M p 38.8TW
; ; ). Of course, thisM p 35.5 t(4,271) p 7.57 p ! .000001ST does not necessarily mean that people move toward the tightwad end of the continuum as they age. The data were collected across people, rather than across time, and the pattern above may therefore reflect the effects of growing up in different generations. Longitudinal research should seek to understand how (and why) one’s location on the ST- TW scale changes over time.
We also find a modest relationship between ST-TW scores and education. We asked 9,596 respondents (in all samples except the student-heavy Pittsburgh sample) to report the highest level of education they had completed. About 64% had more than a bachelor’s degree (i.e., at least some grad- uate school). Tightwads are 9% more likely than spendthrifts to have more than a bachelor’s degree (66% (1,580/2,391) vs. 57% (815/1,422); ; ). Moreover,2x (1) p 29.3 p ! .0001 among the 9,117 respondents who went to or were currently attending college, we find that tightwads and spendthrifts are attracted to different types of majors. For each reported major, we computed the average ST-TW score among re- spondents who reported it as their sole major. The three majors with the lowest ST-TW means were engineering
(13.2; ), computer science (13.51; ), andn p 645 n p 371 natural science (13.92; ), whereas the three ma-n p 1,724 jors with the highest ST-TW means were humanities (14.87;
), communication (14.92; ), and socialn p 1,075 n p 216 work (16.46; ).n p 41
Finally, we examine whether sample differences along these demographic dimensions mediate the relationship be- tween sample and ST-TW scores. Table 3 presents a series of regressions predicting ST-TW scores in the NBC, Globe and Mail, and New York Times samples; New York Times is the omitted category. Note that the demographic measures only slightly reduce the significance of the sample dummies. Thus, although these measures differ across samples and correlate with ST-TW scores, none fully mediates the re- lationship between sample and ST-TW scores.
We will now examine the ST-TW scale’s validity. We begin by demonstrating that the scale is reliable, in terms of both internal consistency and test-retest reliability. We then examine whether the ST-TW scale is distinct from mea- sures of basic psychological constructs, marketing con- structs, patience, and socially desirable responding. Finally, we demonstrate the scale’s construct validity by demon- strating that it predicts self-reported savings and credit card debt.
Reliability The ST-TW scale is reliable, as reflected by a standardized
Cronbach’s alpha of .75 and average inter-item correlation of .42. To assess test-retest reliability, the scale was re- administered to 447 people 2–539 days after its first ad- ministration, with an average of 207 days between the two administrations. The correlation between ST-TW scores at time 1 and at time 2 was .83 ( ; )t(212) p 21.9 p ! .0001 when administrations were separated by 2–180 days (mean separation of 78 days). That this correlation is comparable to the 3-month test-retest reliability for the Big Five Inven- tory ( ; John and Srivastava 1999, 22), a measure ofr p .85 some of the most universally accepted traits in personality psychology, strongly suggests that our scale captures a stable construct. Moreover, the correlation between ST-TW scores at time 1 and at time 2 was .70 ( ; )t(134) p 11.5 p ! .0001 when administrations were separated by 181–360 days
772 JOURNAL OF CONSUMER RESEARCH
TABLE 4
SCALE CORRELATIONS AND RELIABILITY ESTIMATES
Scale N Alpha Correlation w/ST-TW
Basic psychological constructs: Affect Intensity Measure Short Form (Geuens and
De Pelsmacker 2002; Larsen and Diener 1987) 138 .78 .22** Big Five Inventory (BFI; John, Donahue, and Kentle 1991)
extraversion subscale 140 .85 .13 BFI agreeableness subscale 138 .75 .09 BFI conscientiousness subscale 139 .84 !.13 BFI neuroticism subscale 138 .76 .11 BFI openness subscale 137 .82 !.04 Maximization scale (Schwartz et al. 2002) 1,363 .67 !.07* Regret scale (Schwartz et al. 2002) 1,399 .82 !.08** Regulatory focus questionnaire (RFQ; Higgins et al. 2001)
promotion subscale 1,387 .69 .02 RFQ prevention subscale 1,412 .79 !.13** Self-control short form (Tangney et al. 2004) 198 .84 !.25** Sensation-seeking form V (Zuckerman 1994) 54 .87 .27* Test of Self Conscious Affect (TOSCA) 3-Guilt Subscale
(Tangney and Dearing 2002) 138 .75 !.27** TOSCA-3 Detachment Subscale 138 .74 !.01 TOSCA-3 Externalization Subscale 139 .70 !.03 TOSCA-3 Shame Subscale 139 .67 !.08
Marketing constructs: Compulsive buying (Faber and O’Guinn 1992) 58 .73 !.15 Frugality (Lastovicka et al. 1999) 1,955 .84 !.46** Materialism Nine-Item Short Form (Richins 2004) 257 .83 .26** Price consciousness (Lichtenstein et al. 1993) 136 .81 !.40** Sale proneness (Lichtenstein et al. 1993) 135 .87 .00 Value consciousness (Lichtenstein et al. 1990) 136 .89 !.33**
Patience: Barratt Impulsivity Scale Form 11 (Patton, Stanford, and
Barratt 1995) 56 .81 .08 Time preference 709 .62 .12** Zimbardo Time Perspective Inventory (ZTPI) short form
present subscale (Keough, Zimbardo, and Boyd 1999) 59 .64 .23+
ZTPI short form future subscale 58 .79 !.21 Socially desirable responding:
Balanced inventory of desirable responding (Paulhus 1984) 57 .81 !.04 Concern for appropriateness (Lennox and Wolfe 1984) 76 .89 .08
NOTE.—For each scale in this table, we investigated whether a quadratic model ( ) fit significantly better ( ) than a linear2scale p b + b (ST-TW) + b (ST-TW ) p ! .050 1 2 model ( )). The quadratic model fit better for three scales: sensation-seeking ( ), extraversion ( ), and openness ( ). Our measurescale p b + b (ST-TW b ! 0 b 1 0 b 1 00 1 2 2 2 of time preference consisted of two items: item 1: A p $100 immediately, B p $__ in one year; and item 2: A p $__ immediately, B p $400 in one year. Respondents were asked to fill in the blanks to make A and B equally attractive. Time preference was computed as follows: let x be the response, r be the rate of time preference, and t be the number of years between A and B. Assuming an exponential discount function (i.e., time consistency), item 1 implies 100 p
; i.e., . Similarly, item 2 implies . Higher values of r reflect greater impatience. We averaged the implied r from each item to!rtxe r p !ln(100/x) r p !ln(x/400) create our measure of time preference.
+ that .p ! .10 r ( 0 * .p ! .05 ** .p ! .01
(mean, 241 days), and .72 ( ; ) whent(95) p 10.2 p ! .0001 administrations were separated by over 360 days (mean, 443 days).
Discriminant Validity
To assess the ST-TW scale’s discriminant validity, we assessed its relationship with 28 potentially related scales from the economics, psychology, and marketing literatures. Table 4 presents these scales, their standardized alphas (Iacobucci and Duhachek 2003, 486), and their zero-order correlations with the ST-TW scale. We first examine the
relationship between the ST-TW scale and the scale with which it is most highly correlated, Lastovicka et al.’s (1999) measure of frugality ( ). We then examine the re-r p !.46 lationship between the ST-TW scale and measures of basic psychological constructs, marketing constructs, patience, and socially desirable responding.
Relationship with Frugality. In this section we first use confirmatory factor analysis to demonstrate that the ST- TW and frugality constructs are distinct. We then attempt to uncover how these constructs differ. We conclude that the evidence suggests that frugality is driven by a pleasure of
TIGHTWADS AND SPENDTHRIFTS 773
TABLE 5
RELATIONSHIP BETWEEN PAIN OF PAYING AND ST-TW AND FRUGALITY SCORES
Model 1 Model 2 Model 3
ST-TW !.42*** . . . !.42*** Frugality . . . .21*** .02 R 2 .18 .04 .18
NOTE.—Regression weights are standardized. *** .p ! .001
TABLE 6
RELATIONSHIP BETWEEN PLEASURE OF SAVING AND ST-TW AND FRUGALITY SCORES
Model 1 Model 2 Model 3
ST-TW !.18*** . . . !.01 Frugality . . . .45*** .45*** R 2 .03 .20 .20
NOTE.—Regression weights are standardized. *** .p ! .001
saving, as compared with tightwaddism, which is driven by a pain of paying.
As an initial, theory-free test of whether the ST-TW and frugality scales are distinct, we conducted a confirmatory factor analysis of the ST-TW and frugality items. Specif- ically, we tested whether a model in which two factors, ST-TW and frugality, were allowed to covary fit the data better than a unidimensional model that assumes perfect correlation between the two factors. The chi-square statis- tic associated with the former model is ,2x (53) p 602.68 whereas the chi-square statistic associated with the latter model is , a highly significant differ-2x (54) p 1,118.08 ence ( ; ; ), which suggests2Dx p 515.40 df p 1 p ! .0001 that the two constructs are distinct (Anderson and Gerbing 1988).
Of course, confirmatory factor analysis cannot shed light on how these constructs differ. An examination of frugality- scale items (e.g., “making better use of my resources makes me feel good,” Lastovicka et al. 1999) suggests that the highly frugal may spend conservatively because they enjoy saving, not because the prospect of spending pains them.
To examine this hypothesis, we first asked 966 respon- dents from the Pittsburgh sample to rate the extent to which they agree with the statement “spending money is painful for me” on a 1 (strongly disagree) to 5 (strongly agree) scale. Table 5 shows the results of a series of regressions predicting agreement with the statement. Models 1 and 2 suggest that both ST-TW and frugality scores predict the extent to which people find spending money painful. How- ever, the standardized coefficient for ST-TW is twice the magnitude of that for frugality, the R2 is more than four times as large, and when both scales are entered into a multiple regression (model 3), only ST-TW scores remain significantly related to a self-reported pain of paying.2 This
2Note that these results do not necessarily support the claim that tight- wads are particularly likely to experience an anticipatory pain of paying. Rather, it is possible that tightwads only find spending painful once they have spent money. That is, it may not be the case that tightwads are particularly likely to be deterred from spending by an anticipatory pain of paying, but rather that they are particularly likely to regret their purchases or feel guilty about them. To examine the extent to which the pain of paying is an emotion experienced at the moment of choice, we asked 652 Pittsburgh respondents, “When do you find spending money most painful?” Respondents could choose one of five options: minutes before making a purchase, seconds before making a purchase, the moment of purchase, seconds after making a purchase, or minutes after making a purchase. Consistent with the claim that the pain of paying is an anticipatory emotion
provides strong evidence that tightwaddism is more closely related to the pain of paying than is frugality.
Next, we examine whether frugality is more closely re- lated to the pleasure of saving than is tightwaddism. We asked 316 respondents from the Pittsburgh sample to rate the extent to which they agree with the statement “saving money is pleasurable for me” on a 1–5 scale. Note that these respondents were part of the larger set of 966 respondents who rated their agreement with the statement “spending money is painful for me,” which allows us to assess the relationship between the pleasure of saving and the pain of paying. Somewhat surprisingly, we find that the extent to which people report experiencing a pain of paying is vir- tually independent of the extent to which they report ex- periencing a pleasure of saving ( ). This stronglyr p .08 suggests that tightwads do not find spending painful because they find saving pleasurable.
Table 6 shows the results of a series of regressions pre- dicting agreement with the pleasure of saving statement. Models 1 and 2 suggest that both ST-TW and frugality scores predict the extent to which people find saving money plea- surable. However, in this case the standardized coefficient for frugality is more than twice the coefficient for ST-TW, the R2 is almost seven times the magnitude, and when both scales are entered into a multiple regression (model 3), only frugality scores remain significantly related to the pleasure of saving.
Thus, although both tightwads and the highly frugal may spend conservatively, they appear to do so for different rea- sons. Conservative spending by tightwads is likely driven by a pain of paying, whereas conservative spending by the highly frugal is likely driven by a pleasure of saving.
Relationship with Other Scales. The first section of table 4 presents the relationship between the ST-TW scale and measures of several psychological constructs. Such comparisons are important to make, as individual differences
among tightwads, we find that tightwads are significantly more likely to find spending most painful before the moment of purchase rather than after the moment of purchase (83/138 vs. 32/138; ; ).2x (1) p 38.77 p ! .0001 Tightwads are also significantly more likely than unconflicted consumers (183/400) and spendthrifts (36/114) to report that spending is most painful before the moment of purchase ( , , and2 2x (1) p 8.51 p ! .01 x (1) p
, , respectively). Spendthrifts, unlike tightwads, are signif-20.44 p ! .0001 icantly more likely to find spending most painful after the moment of purchase rather than before the moment of purchase (56/114 vs. 36/114;
; ).2x (1) p 7.29 p ! .01
774 JOURNAL OF CONSUMER RESEARCH
TABLE 7
CREDIT CARD DEBT BY ST-TW CLASSIFICATION (%)
Tightwad ( )n p 2,406
Unconflicted ( )n p 5,780
Spendthrift ( )n p 1,430
Total ( )n p 9,616
Do not use 8 6 7 7 Pay off balance 73 63 37 61 $1–$5,000 in debt 11 18 29 18 $5,001–$20,000 in debt 6 10 20 10 Over $20,000 in debt 2 3 8 3
NOTE.—Some columns do not sum to 100% due to rounding errors.
in spending behavior may be the consequence of individual differences in more basic personality traits (Lastovicka 1982). Note that we find a negative correlation between the ST-TW scale and the short form of Tangney, Baumeister, and Boone’s (2004) self-control scale ( ). The mod-r p !.25 esty of this correlation is not particularly surprising given Tangney et al.’s (2004, 314) distinction between self-control and self-regulation. Their view suggests that tightwads and spendthrifts do not necessarily differ in trait levels of self- control, but rather that they both have problems of self- regulation. That is, tightwads are unable to suspend (oth- erwise beneficial) self-control when doing so would be desirable, and spendthrifts have an inability to exert self- control when doing so would be desirable.
The second section of table 4 presents the relationship between the ST-TW scale and measures of several con- structs from the marketing literature. Although our scale, like many other scales in marketing, is intended to predict spending behavior, we believe our scale uniquely captures individual differences in the pain of paying. Note, for ex- ample, that there is only a modest correlation with Faber and O’Guinn’s (1992) measure of compulsive buying ( ) and Richins’s measure of materialism (r p !.15 r p
). The former correlation is negative because lower.26 scores on Faber and O’Guinn’s (1992, 468) measure are diagnostic of greater compulsive buying. The weakness of these correlations likely reflects the fact that spendthrifts do not overspend primarily because they are trying to al- leviate negative affect (a hallmark of compulsive buying) or because they are trying to impress others (a hallmark of materialism), but rather because they fail to experience sufficient pain of paying. ST-TW scores are more strongly related to value consciousness (Lichtenstein, Netemeyer, and Burton 1990; ) and price consciousnessr p !.33 (Lichtenstein, Ridgway, and Netemeyer 1993; ),r p !.40 but the size of these correlations suggests that neither is the defining characteristic of tightwaddism.
The third section of table 4 presents the relationship be- tween the ST-TW scale and measures of patience and im- pulsivity. We conduct such comparisons to examine whether our scale simply captures individual differences in valuation of future consumption. The generally low correlations sug- gest that it does not. Indeed, it would have been surprising if these correlations were high, as our scale is intended to capture individual differences in the immediate pain of pay-
ing, which people presumably rely on so that they do not have to think carefully about what is given up in the future.
The fourth section of table 4 presents the relationship between the ST-TW scale and two measures of socially desirable responding. Since “tightwad” and “spendthrift” have a somewhat negative connotation (see item 1 of the ST-TW scale) and since Mr. A and Mr. B are both described as doing something they do not want to do (see item 3), people who tend to respond in a socially desirable fashion may be motivated to categorize themselves as unconflicted consumers. However, we find no significant curvilinear (or linear) relationship between ST-TW scores and balanced inventory of desirable responding (Paulhus 1984) or concern for appropriateness (Lennox and Wolfe 1984) scores, sug- gesting that the ST-TW scale is not influenced by socially desirable responding.
Construct Validity
Having provided evidence of the reliability and discrim- inant validity of the ST-TW scale, we next test the basic hypothesis that spendthrifts should generally spend more than tightwads. We did so by asking respondents in the Globe and Mail, New York Times, and NBC samples to report their total savings and current level of credit card debt. We also asked respondents to report their annual in- come in order to determine whether any tightwad/spendthrift differences in savings or credit card debt could be attributed to differences in income rather than to differences in spend- ing habits.
We first examine the credit card responses ( ).N p 9,616 Respondents were asked, “approximately how much credit card debt do you have?” and could indicate one of nine intervals, ranging from $1–$500 to over $50,000. The in- tervals were shown in terms of both Canadian and U.S. dollars for Globe and Mail respondents. Respondents could alternatively indicate “I pay off my balance in full each month” or “I do not use credit cards.”
Table 7 presents the frequency of different credit card responses by consumer type. Notice that tightwads are about as likely as spendthrifts to abstain from using credit ( ; ). When we focus only on credit2x (1) p 1.35 p p .25 card users, we find that spendthrifts are three times more likely than tightwads to carry debt. Sixty percent (804/ 1,330) of spendthrift credit users carry debt, but only 20%
TIGHTWADS AND SPENDTHRIFTS 775
TABLE 8
AMOUNT SAVED BY ST-TW CLASSIFICATION (%)
Tightwad ( )n p 2,326
Unconflicted ( )n p 5,654
Spendthrift ( )n p 1,414
Total ( )n p 9,394
$0–$10,000 24 31 52 33 $10,001–$50,000 24 25 21 24 $50,001–$100,000 12 10 8 10 $100,001–$250,000 12 11 7 11 Over $250,000 28 22 12 22
NOTE.—Some columns do not sum to 100% due to rounding errors.
TABLE 9
INCOME BY ST-TW CLASSIFICATION (%)
Tightwad ( )n p 2,349
Unconflicted ( )n p 5,669
Spendthrift ( )n p 1,413
Total ( )n p 9,431
$0–$10,000 8 7 9 8 $10,001–$50,000 32 31 36 32 $50,001–$100,000 33 34 32 33 $100,001–$250,000 22 23 17 22 Over $250,000 5 5 6 5
(451/2,213) of tightwad credit users do ( ;2x (1) p 583.2 ). Spendthrifts are not only more likely to be inp ! .0001
debt, but they also carry more debt. When we focus only on credit card users in debt, we find that tightwads are more likely than spendthrifts to carry $1–$5,000 in debt (61% (273/451) vs. 51% (413/804); ;2x (1) p 9.79 p !
), whereas spendthrifts are more likely than tightwads.005 to carry over $20,000 in debt (13% (108/804) vs. 9% (42/ 451); ; ).2x (1) p 4.66 p ! .05
Next, we examine savings ( ) and incomeN p 9,394 ( ) responses. Respondents were asked, “approx-N p 9,431 imately how much money do you have in savings?” and “what is your annual income?” Respondents could select one of 12 intervals, ranging from $0–$10,000 to over $250,000. Intervals were shown in terms of both Canadian and U.S. dollars for Globe and Mail respondents.
Table 8 presents the frequency of different savings re- sponses by consumer type. The distribution of savings among tightwads differs from the distribution among spendthrifts ( ; ), but we observe2x (4) p 333.39 p ! .0001 particularly strong differences at the two extremes of amount saved. Spendthrifts are more than twice as like- ly as tightwads to have less than $10,000 in savings ( ; ), and tightwads are more than2x (1) p 293.48 p ! .0001 twice as likely as spendthrifts to have more than $250,000 in savings ( ; ).2x (1) p 129.06 p ! .0001
Thus, spendthrifts carry more debt and save less than tightwads. It is worth examining whether there are analogous tightwad/spendthrift differences in income. Table 9 presents the frequency of different income responses by consumer type. The distribution of income among spendthrifts differs from the distribution among tightwads ( ;2x (4) p 15.23
). This difference is driven by the fact that spend-p ! .005
thrifts are 4% more likely than tightwads to have incomes ranging from $10,001 to $50,000 ( ; ),2x (1) p 7.32 p ! .01 whereas tightwads are 5% more likely than spendthrifts to have incomes ranging from $100,001 to $250,000 ( ; ). Tightwad/spendthrift differences2x (1) p 10.51 p ! .005 in the three other income categories fail to reach significance (all ; all ). Thus, although the income dis-2x (1) X 1 p 1 .30 tribution among tightwads differs from the income distri- bution among spendthrifts, these small differences are un- likely to account for large differences in credit card debt and savings noted earlier.
Table 10, for example, focuses on the 3,751 tightwads and spendthrifts for whom we have both income and credit card data. Focusing only on credit card users, spendthrifts are significantly more likely than tightwads to carry credit card debt at each income level (all ; all2x (1) 1 19 p !
). The results suggest that tightwad/spendthrift differ-.0001 ences in credit card debt do not depend heavily on income.
TESTS OF OUR THEORETICAL FRAMEWORK
The evidence presented thus far suggests that the ST-TW scale is reliable, distinct from related constructs, and pre- dictive of credit card debt and savings. However, individual differences are not all-powerful determinants of behavior. Our theoretical framework does not predict that spendthrifts will spend more than tightwads across all domains. If tight- wads are particularly prone to experience the pain of paying, they should spend less when situational factors intensify the pain of paying than when situational factors mitigate the pain of paying. If spendthrifts are not particularly prone to experience the pain of paying (and thus less sensitive than
776 JOURNAL OF CONSUMER RESEARCH
FIGURE 2
PROPORTION WILLING TO PAY FEE (STUDY 1)
TABLE 10
CREDIT CARD DEBT BY ST-TW CLASSIFICATION AND INCOME (%)
Tightwad Spendthrift
Income Do not use Pay off balance Carry debt Do not use Pay off balance Carry debt
$0–$10,000 35 56 9 24 46 30 $10,001–$50,000 11 67 22 8 27 65 $50,001–$100,000 4 73 23 3 35 62 $100,001–$250,000 2 83 16 4 48 48 Over $250,000 2 92 6 2 64 33
NOTE.—Rows within a particular consumer type sum to 100%; some do not due to rounding errors.
tightwads to such situational factors), spending differences between tightwads and spendthrifts should be greatest when situational factors intensify the pain of paying. By contrast, such spending differences should be smallest when situa- tional factors mitigate the pain of paying. Studies 1 and 2 experimentally test this hypothesis.
Study 1
In study 1 we experimentally manipulate whether or not the (hypothetical) fee to have a product shipped overnight is framed as “small.” Presumably, framing the fee as small makes it seem less painful to pay, and thus spending dif- ferences between tightwads and spendthrifts are predicted to be smallest in the “small” fee condition.
Participants. Over a 5-month period beginning in March 2005, 538 Carnegie Mellon students responded to a series of online surveys that, among other items, included a question that asked participants if they would be willing to pay a hypothetical fee. The sample included 88 tightwads, 112 spendthrifts, and 338 unconflicted consumers.
Method. All surveys began with the ST-TW scale. Sur- vey respondents were then asked, “suppose that in exchange for completing a survey for Amazon.com, you could receive your choice of one of the four DVD box sets listed below. If you choose to complete the survey, the box set will be shipped to you for free within four weeks. Which box set would you most like to receive?” The list of DVD box sets included season 1 of the Sopranos, seasons 1 and 2 of Sein- feld, seasons 1 and 2 of Family Guy, and season 1 of Chap- pelle’s Show.
Respondents were then asked one of two questions, which differed by only one word. Respondents in the $5 fee (small $5 fee) condition were asked, “would you be willing to pay a (small) $5 fee to receive the box set by overnight delivery, rather than waiting four weeks?” There were 243 and 295 respondents in the “$5 fee” and “small $5 fee” conditions, respectively.
Results. Since our hypothesis focuses on tightwads and spendthrifts, we will initially focus exclusively on the be- havior of tightwads and spendthrifts. Figure 2 presents the proportion of tightwads and spendthrifts willing to pay the
fee in each condition. We analyzed the data with factorial logistic regression, which treated the binary decision as the dependent variable and type of framing and consumer type as the independent variables. We find no significant main effect of the type of framing ( ; ). Par-2x (1) p .06 p p .81 ticipants are not significantly more likely to pay the small $5 fee than the $5 fee (33% (37/113) vs. 25% (22/87)). However, there is a significant main effect of consumer type ( ; ). Spendthrifts are significantly more2x (1) p 8.93 p ! .01 likely than tightwads to pay the fee (38% (42/112) vs. 19% (17/88)). Most importantly, we find a significant interaction between the type of framing and consumer type ( 2x (1) p
; ). Spendthrifts are only 9% more likely than4.21 p ! .05 tightwads to pay the “small” $5 fee (37% (23/63) vs. 28% (14/50); ; ), but they are 31% more2x (1) p .92 p p .34 likely to pay the $5 fee (39% (19/49) vs. 8% (3/38);
; ). Viewed in ratio terms, spend-2x (1) p 10.8 p p .001 thrifts are nearly five times more likely than tightwads to
TIGHTWADS AND SPENDTHRIFTS 777
pay the $5 fee, but almost equally likely to pay the “small” $5 fee.
Finally, we examine the behavior of unconflicted con- sumers, about whom we did not have a hypothesis. Pooling across framing conditions, 27% of unconflicted consumers pay the fee, as compared to 19% of tightwads and 38% of spendthrifts. Like spendthrifts, unconflicted consumers are fairly insensitive to the framing manipulation: 28% (51/182) pay the small $5 fee, and 26% (40/156) pay the $5 fee ( ; ).2x (1) p .24 p p .62
Discussion. Consistent with our hypothesis, we find that spending differences between tightwads and spend- thrifts are smallest when situational factors mitigate the pain of paying. One limitation of this study, however, is that there was no manipulation check to verify that we were indeed manipulating the pain of paying. Accordingly, it is difficult to rule out alternative explanations. One alternative expla- nation, consistent with the education data presented earlier, could be that tightwads are more thoughtful. The source of the framing (the experimenter) is presumably credible, and thoughtful consumers may infer that they are being told that the $5 fee for overnight shipping is small because it is indeed small relative to typical overnight shipping fees (Grice 1975). In study 2 we therefore attempt to replicate the in- teraction observed here and more definitively attribute it to situational differences in the pain of paying.
Study 2
Like study 1, study 2 examines the hypothesis that spend- ing differences between tightwads and spendthrifts will be smallest when situational factors diminish the pain of pay- ing. To manipulate the pain of spending, we vary whether a (hypothetical) massage is framed as necessary to relieve back pain (utilitarian) or desired because it would be plea- surable (hedonic). If the pain of paying is buffered by the magnitude or length of anticipated benefits (Prelec and Loewenstein 1998), and the utilitarian massage provides more long-term benefits than the purely hedonic massage, then the utilitarian massage should be less painful to pay for than the hedonic massage. As a result, spending differ- ences between tightwads and spendthrifts should be greater in the hedonic condition than in the utilitarian condition.
Participants. This experiment was conducted with three samples and a total of 1,087 participants. One version of our Globe and Mail survey included this experiment and was taken by 36 tightwads, 6 spendthrifts, and 60 uncon- flicted consumers. One version of our New York Times sur- vey also included this experiment and was taken by 131 tightwads, 83 spendthrifts, and 331 unconflicted consumers. We also contacted tightwads and spendthrifts who had pre- viously taken either a New York Times or NBC survey that did not include the present experiment and asked them to take a new survey that only included this experiment. The amount of time between the two surveys varied from 1 to 3 months. Of the 1,019 tightwads contacted, 273 partici-
pated in the present experiment, and of the 696 spendthrifts contacted, 167 participated. Across all samples, 440 tight- wads, 256 spendthrifts, and 391 unconflicted consumers participated.
Method. In the Globe and Mail and New York Times samples, we employed a design. Specifically, we var-2 # 2 ied whether participants faced the massage scenario im- mediately before or immediately after completing the ST- TW scale and whether they were assigned to the utilitarian or hedonic scenario. In the New York Times/NBC sample, participants were classified as tightwads or spendthrifts based on their original survey responses. For these partic- ipants, we only varied whether they were assigned to the utilitarian or hedonic scenario; they did not complete the ST-TW scale a second time.
In the utilitarian condition, participants read the following scenario. Imagine that your back has been bothering you lately. You discuss the situation with a physician, who rec- ommends a therapeutic massage. You shop around and find an excellent clinic that offers a therapeutic massage for $100. Your insurance does not cover the cost. And in the hedonic condition, participants read the following scenario. Imagine that you find massages very pleasurable (no need to imagine if this is actually true). You shop around and find an excellent spa that offers a pleasurable massage for $100.
Immediately following each scenario, participants were asked, “would you get the massage?” They could indicate either yes or no. There were 548 respondents in the utili- tarian condition and 539 in the hedonic condition.
A manipulation check followed the decision (cf. Shiv and Fedorikhin 1999). Specifically, participants were asked, “how painful would it be to pay for the massage?” Responses were rated on a 1 (not at all painful) to 7 (very painful) scale.
Manipulation Check. Factorial ANOVA treating self- reported pain of paying as the dependent variable, and mas- sage and consumer type as the independent variables, reveals a significant main effect of massage type (F(1, 692) p
; ). As predicted, participants in the utilitarian48.23 p ! .0001 condition find paying for the massage significantly less painful than do participants in the hedonic condition ( , ). There is also a signifi-M p 3.92 M p 4.76utilitarian hedonic cant main effect of consumer type ( ;F(1, 692) p 34.35
); tightwads find paying for the massage signifi-p ! .0001 cantly more painful than do spendthrifts ( ,M p 4.60TW
). There is no significant interaction betweenM p 3.87ST consumer type and massage type ( ;F(1, 692) p 1.29 p p
)..26
Results. Since our hypothesis focuses on tightwads and spendthrifts, we will initially focus exclusively on the be- havior of tightwads and spendthrifts. We first examine whether the order (or presence) of the ST-TW scale influ- enced choices. Overall, willingness to pay for a utilitarian massage does not vary significantly with ST-TW scale placement: 68% buy when the scenario precedes the scale,
778 JOURNAL OF CONSUMER RESEARCH
FIGURE 3
PROPORTION WILLING TO BUY MASSAGE (STUDY 2)
67% buy when the scenario follows the scale, and 75% buy when the scale is not present ( ;2x (2) p 2.26 p p
). Likewise, willingness to pay for a hedonic massage.32 does not vary with ST-TW scale placement: 32% buy when the scenario precedes the scale, 30% buy when the scenario follows the scale, and 34% buy when the scale is not present ( ; ). Moreover, willingness to2x (2) p .44 p p .80 buy either massage is not significantly influenced by ST- TW scale placement among either consumer type (all
; all ). In our analyses below, we there-2x (2) ! 4.10 p 1 .10 fore collapse responses across this factor.
Figure 3 presents the proportion of tightwads and spend- thrifts willing to purchase each type of massage. We an- alyzed the data with factorial logistic regression, which treated the binary purchase decision as the dependent var- iable and massage and consumer type as the independent variables. We find a significant main effect of massage type ( ; ). Participants are signifi-2x (1) p 24.37 p ! .0001 cantly more willing to buy the utilitarian massage than the hedonic massage (72% (258/357) vs. 33% (111/339)). We also find a significant main effect of consumer type ( ; ). Tightwads are significantly2x (1) p 22.10 p ! .0001 less willing than spendthrifts to buy a massage (47% (206/ 440) vs. 64% (163/256)). Most importantly, we find a sig- nificant interaction between consumer type and massage type ( ; ). Spendthrifts are 9% more2x (1) p 4.15 p ! .05 likely than tightwads to buy the utilitarian massage (78% (100/128) vs. 69% (158/229); ; ) and2x (1) p 3.41 p 1 .05 26% more likely to buy the hedonic massage (49% (63/128) vs. 23% (48/211); ; ). Viewed in2x (1) p 25.35 p ! .0001 ratio terms, spendthrifts are more than twice as likely as tightwads to buy the hedonic massage, but almost equally likely to buy the utilitarian one.
Finally, we examine the behavior of unconflicted con- sumers, about whom we did not have a hypothesis. Like tightwads and spendthrifts, unconflicted consumers are largely unaffected by the placement of the ST-TW scale. In the utilitarian condition, 70% buy when the scenario pre- cedes the scale and 77% buy when the scenario follows the scale ( ; ). In the hedonic condition,2x (1) p 1.36 p p .24 39% buy when the scenario precedes the scale and 38% buy when the scenario follows the scale ( ;2x (1) p .02 p p
). Collapsing responses across the order factor, uncon-.89 flicted consumers are significantly more willing to buy the utilitarian massage than the hedonic massage (73% (140/ 191) vs. 38% (76/200); ; ). The2x (1) p 49.23 p ! .0001 35% difference among unconflicted consumers falls in be- tween the 46% difference among tightwads (69% vs. 23%) and the 29% difference among spendthrifts (78% vs. 49%).
Discussion. Taken together, the two studies presented here support our hypothesis that spending differences be- tween tightwads and spendthrifts will be smallest when sit- uational factors diminish the pain of paying. Nevertheless, it would be useful to test our hypothesis using other ma- nipulations to vary the pain of paying. For example, Prelec and Loewenstein (1998) propose that it is less painful to spend token currencies (e.g., casino chips, beads at Club
Med) than regular money. Our framework predicts that dif- ferences in spending between tightwads and spendthrifts will be greater when both are spending regular money than when both are spending token currencies.
GENERAL DISCUSSION
Consequentialist models of decision making assume that emotions experienced at the moment of choice are epiphe- nomenal, simply a by-product of the decision-making pro- cess. The only emotions assumed to influence decision mak- ing are those that are anticipated to occur if various courses of action are taken. However, the results presented here suggest that individual differences in tendencies to experi- ence an immediate emotion, the pain of paying, powerfully influence spending behavior.
That so many people in our sample of over 13,000 ex- perience the pain of paying intensely is counterintuitive given the incredibly low rates of saving in the United States. How can the two phenomena be reconciled? One possibility is that our samples are not representative of the population. Future research should examine the distribution of ST-TW scores in other samples. Another possibility is simply that it is difficult for many people to make ends meet. Income constraints coupled with uninsured health problems or other unpleasant surprises, or the need to pay for child care, dental work, transportation, and other routine expenses, can drive even tightwads into debt. It is also possible that, while many tightwads apparently experience more pain than they would like to experience, far fewer feel as much pain as they would need to feel to ensure sufficient savings.
Of course, individual differences are not all-powerful de- terminants of behavior, and we find that tightwads and spendthrifts behave similarly when situational factors di- minish the pain of paying. Indeed, an alternative explanation for the coexistence of widespread undersaving and tight- waddism could be that many retail environments provide
TIGHTWADS AND SPENDTHRIFTS 779
increasingly painless ways to pay. Paying by credit, for ex- ample, is becoming less and less painful. Credit transactions can now be executed with a single mouse click (e.g., Am- azon.com’s patented One-Click checkout) or a single tap of a key fob (e.g., MasterCard’s “contactless” PayPass credit card).
Directions for Future Research
Although our experiments suggest that tightwads are most sensitive to situational determinants of the pain of paying, spendthrifts may be the most distinctive of the three types of consumers identified by the ST-TW scale. That is, un- conflicted consumers often appear more similar to tightwads than to spendthrifts. For example, unconflicted credit users were 13% less likely than tightwad credit users, and 27% more likely than spendthrift credit users, to pay off their balance in full each month. Similarly, unconflicted consum- ers are only 7% more likely than tightwads, and 21% less likely than spendthrifts, to have $10,000 or less in savings. However, the behavior of unconflicted consumers is not al- ways more similar to that of tightwads than to that of spend- thrifts; for instance, tightwads were the only type of con- sumer who appeared sensitive to the framing manipulation in study 1. Future research should examine more explicitly whether unconflicted consumers are more distinct from tightwads or spendthrifts, in terms of both behavior and the pain of paying.
Moreover, while the present research focused on the be- havioral implications of individual differences in the pain of paying, these differences may have hedonic consequences as well. Prior research on compulsive spending and de- pression suggests that we should observe a linear relation- ship between happiness and ST-TW scores, whereby spend- thrifts are most unhappy (Black et al. 1998; Faber and Christenson 1996). However, if both tightwads and spend- thrifts consistently deviate from their desired spending hab- its, then we should observe a curvilinear relationship be- tween ST-TW scores and happiness, whereby unconflicted consumers are happiest. Indeed, this may be another domain in which tightwaddism differs from frugality. If the highly frugal derive pleasure from saving and consistently save,
then we should observe a linear relationship between fru- gality scores and happiness, whereby the most frugal are happiest.
Future research should also seek to establish a direct link between the actual experience of anticipatory pain and one’s location on the ST-TW scale. Although the present research is highly suggestive of such a link, we have yet to establish a link between ST-TW scores and physiological or brain- imaging data. In addition to examining whether a correlation between physical measures of pain and ST-TW scores exists, future work should also examine whether medications that reduce pain and anxiety (e.g., lorazepam; Paulus et al. 2005) have a particularly strong effect on the spending behavior of tightwads.
Finally, future research should devise ways to identify tightwads and spendthrifts “in the wild.” Although many businesses want to know which customers are tightwads and which are spendthrifts, using a scale to perform the diagnoses would be difficult. However, easily observable behaviors may be highly diagnostic of tightwaddism or spendthriftiness. For example, when shopping online, people can often search for products based on price or quality. When shopping for flights, for instance, consumers often reveal whether price or more hedonic concerns (e.g., number of stops, departure/arrival times, type of seat) are a priority. Such information could serve as a valuable proxy for one’s ST-TW score.
Conclusion
When we have presented this research at meetings, we sometimes take a show of hands of how many people can easily classify themselves as tightwads and of how many people are personally familiar with people they view as extreme tightwads. Both questions generally produce a large fraction of raised hands. The research reported here, there- fore, supports the commonplace intuition that people reliably differ in the extent to which they are tightwads or spend- thrifts, and it shows that this trait can be measured with a simple scale that is reliable, valid, and predictive of a wide range of important consumer behaviors.
APPENDIX
THE SPENDTHRIFT-TIGHTWAD SCALE
NOTE.—Items 2b and 3 are reverse-scored.
TIGHTWADS AND SPENDTHRIFTS 781
REFERENCES
Abraham, Carolyn (2007), “Your Brain on Gucci,” Globe and Mail, http://www.reportonbusiness.com/servlet/story/RTGAM.2007 0425.wrob-gucci-0427/BNStory/specialROBmagazine/home.
Ameriks, John, Andrew Caplin, John Leahy, and Tom Tyler (2003), “Measuring Self Control,” working paper, TIAA-CREF In- stitute, New York, NY, 10017.
Anderson, James C. and David W. Gerbing (1988), “Structural Equation Modeling in Practice,” Psychological Bulletin, 103 (3), 411–23.
Baumeister, Roy F. (2002), “Yielding to Temptation: Self-Control Failure, Impulsive Purchasing, and Consumer Behavior,” Journal of Consumer Research, 28 (March), 670–76.
Becker, Selwyn W., Joshua Ronen, and George H. Sorter (1974), “Opportunity Costs—an Experimental Approach,” Journal of Accounting Research, 12 (2), 317–29.
Black, Donald W., Susan Repertinger, Gary R. Gaffney, and Janelle Gabel (1998), “Family History of Psychiatric Comorbidity in Persons with Compulsive Buying: Preliminary Findings,” American Journal of Psychiatry, 155 (7), 960–63.
Bucks, Brian K., Arthus B. Kennickell, and Kevin B. Moore (2006), “Recent Changes in U.S. Family Finances: Evidence from the 2001 and 2004 Survey of Consumer Finances,” Fed- eral Reserve Bulletin, http://www.federalreserve.gov/Pubs/ Bulletin/2006/financesurvey.pdf.
Eisenberger, Naomi I., Matthew D. Lieberman, and Kipling D. Williams (2003), “Does Rejection Hurt: An fMRI Study of Social Exclusion,” Science, 302 (5643), 290–92.
Faber, Ronald J. and Gary A. Christenson (1996), “In the Mood to Buy: Differences in the Mood States Experienced by Com- pulsive Buyers and Other Consumers,” Psychology and Mar- keting, 13 (December), 803–19.
Faber, Ronald J. and Thomas C. O’Guinn (1992), “A Clinical Screener for Compulsive Buying,” Journal of Consumer Re- search, 19 (December), 459–69.
Fisher, Irving (1930), The Theory of Interest as Determined by Impatience to Spend Income and Opportunity to Invest It, New York: Macmillan.
Frederick, Shane, Nathan Novemsky, Jing Wang, Ravi Dhar, and Stephen Nowlis (2006), “Neglect of Opportunity Costs in Consumer Choice,” working paper, Marketing Department, Sloan School of Management, Massachusetts Institute of Technology.
Geuens, Maggie and Patrick De Pelsmacker (2002), “Developing a Short Affect Intensity Scale,” working paper, Vlerick Leu- ven Gent Management School, Ghent University.
Grice, H. Paul (1975), “Logic and Conversation,” in Syntax and Semantics, Vol. 3, Speech Acts, ed. Peter Cole and Jerry Mor- gan, New York: Academic Press, 41–58.
Higgins, E. Tory, Ronald S. Friedman, Robert E. Harlow, Lorraine C. Idson, Ozlem N. Ayduk, and Amy Taylor (2001), “Achievement Orientations from Subjective Histories of Suc- cess: Promotion Pride versus Prevention Pride,” European Journal of Social Psychology, 31, 3–23.
Hoch, Stephen J. and George Loewenstein (1991), “Time-Incon- sistent Preferences and Consumer Self-Control,” Journal of Consumer Research, 17 (March), 492–507.
Iacobucci, Dawn and Adam Duhachek (2003), “Advancing Alpha: Measuring Reliability with Confidence,” Journal of Consumer Psychology, 13 (4), 478–87.
John, Oliver P., Eileen M. Donahue, and Robert L. Kentle (1991), “The Big Five Inventory—Versions 4a and 54,” Institute for
Personality and Social Research, University of California, Berkeley.
John, Oliver P. and Sanjay Srivastava (1999), “The Big-Five Trait Taxonomy: History, Measurement, and Theoretical Perspec- tives,” in Handbook of Personality: Theory and Research, 2nd ed., ed. Lawrence A. Pervin and Oliver John, New York: Guilford, 102–39.
Jones, Steven K., Deborah Frisch, Tricia J. Yurak, and Eric Kim (1998), “Choices and Opportunities: Another Effect of Fram- ing on Decisions,” Journal of Behavioral Decision Making, 11 (3), 211–26.
Keough, Kelli A., Philip G. Zimbardo, and John N. Boyd (1999), “Who’s Smoking, Drinking, and Using Drugs? Time Per- spective as a Predictor of Substance Use,” Basic and Applied Social Psychology, 21 (2), 149–64.
Kivetz, Ran and Itamar Simonson (2002), “Self-Control for the Righteous: Toward a Theory of Precommitment to Indul- gence,” Journal of Consumer Research, 29 (September), 199–217.
Knutson, Brian, Scott Rick, G. Elliott Wimmer, Drazen Prelec, and George Loewenstein (2007), “Neural Predictors of Pur- chases,” Neuron, 53 (January), 147–56.
Larsen, Randy J. and Ed Diener (1987), “Affect Intensity as an Individual Difference Characteristic: A Review,” Journal of Research in Personality, 21 (1), 1–39.
Lastovicka, John L. (1982), “On the Validation of Lifestyle Traits: A Review and Illustration,” Journal of Marketing Research, 19 (February), 126–38.
Lastovicka, John L., Lance A. Bettencourt, Renée S. Hughner, and Ronald J. Kuntze (1999), “Lifestyle of the Tight and Frugal: Theory and Measurement,” Journal of Consumer Research, 26 (June), 85–98.
LeDoux, Joseph E. (1996), The Emotional Brain, New York: Simon & Schuster.
Lennox, Richard D. and Raymond N. Wolfe (1984), “Revision of the Self-Monitoring Scale,” Journal of Personality and Social Psychology, 46 (6), 1349–64.
Lichtenstein, Donald R., Richard G. Netemeyer, and Scot Burton (1990), “Distinguishing Coupon Proneness from Value-Con- ciousness,” Journal of Marketing, 54 (July), 54–67.
Lichtenstein, Donald R., Nancy M. Ridgway, and Richard G. Ne- temeyer (1993), “Price Perceptions and Consumer Shopping Behaviors: A Field Study,” Journal of Marketing Research, 30 (May), 234–45.
Loewenstein, George (1987), “Anticipation and the Valuation of Delayed Consumption,” Economic Journal, 97 (387), 666–84.
Loewenstein, George and Jennifer Lerner (2003), “The Role of Affect in Decision Making,” in Handbook of Affective Sci- ence, ed. Richard J. Dawson, Klaus R. Scherer, and H. Hill Goldsmith, Oxford: Oxford University Press, 619–42.
Loewenstein, George and Ted O’Donoghue (2006), “‘We Can Do This the Easy Way or the Hard Way’: Negative Emotions, Self-Regulation and the Law,” University of Chicago Law Review, 73 (Winter), 183–206.
Loewenstein, George, Scott Rick, and Jonathan Cohen (2008), “Neu- roeconomics,” Annual Review of Psychology, forthcoming.
Loewenstein, George, Elke U. Weber, Christopher K. Hsee, and Ned Welch (2001), “Risk as Feelings,” Psychological Bul- letin, 127 (2), 267–86.
O’Guinn, Thomas C. and Ronald J. Faber (1989), “Compulsive Buying: A Phenomenological Exploration,” Journal of Con- sumer Research, 16 (September), 147–57.
Okada, Erica M. and Stephen J. Hoch (2004), “Spending Time
782 JOURNAL OF CONSUMER RESEARCH
versus Spending Money,” Journal of Consumer Research, 31 (September), 313–23.
Patton, Jim H., Matthew S. Stanford, and Ernest S. Barratt (1995), “Factor Structure of the Barratt Impulsiveness Scale,” Journal of Clinical Psychology, 51 (6), 768–74.
Paulhus, Delroy L. (1984), “Two-Component Models of Socially Desirable Responding,” Journal of Personality and Social Psychology, 46 (3), 598–609.
Paulus, Martin P., Justin S. Feinstein, Gabriel Castillo, Alan N. Simmons, and Murray B. Stein (2005), “Dose-Dependent De- crease of Activation in Bilateral Amygdala and Insula by Lorazepam during Emotion Processing,” Archives of General Psychiatry, 62 (March), 282–88.
Prelec, Drazen and George Loewenstein (1998), “The Red and the Black: Mental Accounting of Savings and Debt,” Marketing Science, 17 (1), 4–28.
Prelec, Drazen, George Loewenstein, and Ofer Zellermayer (1997), “Closet Tightwads: Compulsive Reluctance to Spend,” paper presented at the Association for Consumer Research Confer- ence, Denver.
Richins, Marsha L. (2004), “The Material Values Scale: Measure- ment Properties and Development of a Short Form,” Journal of Consumer Research, 31 (June), 209–19.
Rick, Scott and George Loewenstein (forthcoming), “The Role of Emotion in Economic Behavior,” in The Handbook of Emo- tion, 3rd ed., ed. Michael Lewis, Jeannette M. Haviland-Jones, and Lisa Feldman-Barrett, New York: Guilford.
Rook, Dennis W. (1987), “The Buying Impulse,” Journal of Con- sumer Research, 14 (September), 189–99.
Rook, Dennis W. and Robert J. Fisher (1995), “Normative Influ- ences on Impulsive Buying Behavior,” Journal of Consumer Research, 22 (December), 305–13.
Sanfey, Alan G., James K. Rilling, Jessica A. Aronson, Leigh E. Nystrom, and Jonathan D. Cohen (2003), “The Neural Basis of Economic Decision-Making in the Ultimatum Game,” Sci- ence, 300 (5626), 1755–58.
Schwartz, Barry, Andrew Ward, John Monterosso, Sonja Lyubom-
irsky, Katherine White, and Darrin R. Lehman (2002), “Max- imizing versus Satisficing: Happiness Is a Matter of Choice,” Journal of Personality and Social Psychology, 83 (5), 1178–97.
Shiv, Baba and Alexander Fedorikhin (1999), “Heart and Mind in Conflict: The Interplay of Affect and Cognition in Consumer Decision Making,” Journal of Consumer Research, 26 (De- cember), 278–92.
Stern, Hawkins (1962), “The Significance of Impulsive Buying Today,” Journal of Marketing, 26 (April), 59–62.
Tangney, June P., Roy F. Baumeister, and Angie L. Boone (2004), “High Self-Control Predicts Good Adjustment, Less Pathol- ogy, Better Grades, and Interpersonal Success,” Journal of Personality, 72 (2), 271–324.
Tangney, June P. and Ronda L. Dearing (2002), Shame and Guilt, New York: Guilford.
Tierney, John (2007), “The Voices in My Head Say ‘Buy It!’ Why Argue?” New York Times, January 16, D1.
Valence, Gilles, Alain d’Astous, and Louis Fortier (1988), “Com- pulsive Buying: Concept and Measurement,” Journal of Con- sumer Policy, 11 (September), 419–33.
Vohs, Kathleen D. and Ronald J. Faber (2007), “Spent Resources: Self-Regulatory Resource Availability Affects Impulse Buy- ing,! Journal of Consumer Research, 33 (March), 537–47.
Weun, Seungoog, Michael A. Jones, and Sharon E. Beatty (1998), “Development and Validation of the Impulse Buying Ten- dency Scale,” Psychological Reports, 82 (3), 1123–33.
Wicker, Bruno, Christian Keysers, Jane Plailly, Jean-Pierre Royet, Vittorio Gallese, and Giacomo Rizzolatti (2003), “Both of Us Disgusted in My Insula: The Common Neural Basis of Seeing and Feeling Disgust,” Neuron, 40 (3), 655–64.
Zellermayer, Ofer (1996), “The Pain of Paying,” unpublished dis- sertation, Department of Social and Decision Sciences, Car- negie Mellon University, Pittsburgh, PA.
Zuckerman, Marvin (1994), Behavioral Expressions and Biosocial Bases of Sensation Seeking, Cambridge: Cambridge Univer- sity Press.
