Académique Documents
Professionnel Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
The University of Chicago Press is collaborating with JSTOR to digitize, preserve and
extend access to The Journal of Business
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
B. P. S. Murthi
University of Texas at Dallas
Kannan Srinivasan
Carnegie Mellon University
ate states of evaluation can also exist, and some of them are explored
in this study.
Consistent with Stigler’s (1961) seminal paper on information acqui-
sition, we suggest that a consumer may find the evaluation of all rele-
vant marketing mix information on the brands on every occasion to
be a time-consuming and tedious activity. Similarly, Shugan (1980)
develops the idea of ‘‘cost of thinking’’ associated with purchase evalu-
ation and derives the costs associated with different evaluation strate-
gies. When a consumer does not search, brand choice may not be opti-
mal. Therefore, the consumer makes a trade-off between search costs
and the risk of a suboptimal choice. This trade-off leads the consumer to
engage in evaluation of prices and promotions only on some purchase
occasions.
Based on price search theory, we anticipate that the decision to eval-
uate on a given occasion may be influenced by economic returns from
search (Stigler 1961; Goldman and Johansson 1978), search costs or
time constraints (Ratchford 1982), knowledge (Alba and Hutchinson
1987), and individual household characteristics (Doti and Sharir 1981;
Carlson and Gieseke 1983). We develop surrogate measures to test the
influence of these factors on search propensity.
Further, consistent with Bass (1974) and Goldman and Johansson
(1978), we believe that observed explanatory variables can only par-
tially explain the variation across consumers in their evaluation tenden-
cies. An interesting issue is whether there is an unobserved intrinsic
propensity for evaluation among households. For instance, Marmor-
stein, Grewal, and Fishe (1992) suggest that some consumers may en-
joy price-comparison shopping, while Doti and Sharir (1981) suggest
that consumers dislike shopping. Hence, we model such variation
across consumers in their intrinsic search propensity, unexplained by
demographic and other observed factors.
Moreover, unmeasured factors such as the type of shopping trip and
in-store conditions may also affect the extent of evaluation on a given
occasion. For example, an individual may visit a store to do the weekly
grocery shopping trip or may make an unplanned trip. Also, the pur-
chase may be for a special occasion such as a party or for routine con-
sumption. Failure to control adequately for such a source of unobserved
heterogeneity across purchase occasions may erroneously indicate vari-
ation in intrinsic search propensity across consumers. Hence, we ex-
plicitly control for unobserved variation across purchase occasions.
A final motivation of our analysis is to recover the correct effect of
the marketing inputs on brand choice. Choice models yield an estimate
of the mean effect over purchase occasions, irrespective of whether
consumers attend to marketing inputs. Consequently, they may under-
estimate the actual consumer response to marketing mix elements for
some segments.
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
Brand Choice 231
冱 exp(V
i⫽1
ikn )
冱 exp(U
i⫽1
ikn )
stimuli in their choice. Bucklin and Lattin (1991) report that deal-prone
consumers are opportunistic. Further, the deal-prone consumers are
price sensitive and may be considered to be relative experts on prices
in the category. We expect them, as price experts (Alba and Hutchinson
1987), to exhibit a greater tendency to evaluate.
In summary, based on extant research, we identified nine factors that
might affect search propensity. Four factors vary across purchase occa-
sions within a household (Z skn ): weekday (DOW), category display
(CDISP), category feature (CFEAT), and store loyalty (STLOY). The
remaining five factors vary across households but do not change across
purchase occasions (Zhn ): income (INC), education (EDU), frequency
of purchases (FREQ), time availability (TIME), and deal-proneness
(DLP). We defer details on these measures until the next section. Thus,
the probability of being in an extended evaluation state is
冢 冣
S H
冢 冣
S H
1 ⫹ exp γ 0 ⫹ 冱γ Z
s⫽1
s skn ⫹ 冱γ Z
h⫽1
h hn
where
S
冱γ Zs⫽1
s skn ⫽ γ 1 DOWkn ⫹ γ 2 CDISPkn ⫹ γ 3 CFEATkn ⫹ γ 4 STLOYkn
and
H
冱γ Z
h⫽1
h hn ⫽ γ 5 INC n ⫹ γ 6 EDU n ⫹ γ 7 FREQ n ⫹ γ 8 DLP n ⫹ γ 9 TIME n .
兿 冮 冦兿 冮 冦冤
∞ ∞
L⫽ ( p kn ) prob( j| EE, ξ nk, δ n)
n⫽1 ⫺∞ k⫽1 ⫺∞
冥
⫹ (1 ⫺ p kn )prob( j| HE, ξ nk, δ n) φ(ξ nk )dξ nk φ(δ n )dδ n 冧 冧
N K
⫻ exp ⫺
冢 冣 冧 δ 2n
2σ 2δ
dδ n .
Since we estimate the mean of the intrinsic search propensity (γ̄ 0 ), the
mean of the normal distributions of δ n and ξ nk other than zero is not
identified. The parametric method of controlling for heterogeneity has
been employed by Chintagunta, Jain, and Vilcassim (1991), Gonul and
Srinivasan (1993), and Rossi, McCulloch, and Allenby (1996).
Heckman and Singer (1984) suggest that parameter estimates may
be sensitive to the assumed functional form of the random distribution
and propose a nonparametric specification to capture the unobserved
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
240 Journal of Business
兿 冱 冦兿 冱
R X
A. Data
We estimate the proposed brand choice models using scanner data from
A.C. Nielsen for the ketchup category. Three national brands account
for more than 90% of the market. The dominant brand is Heinz, and
the most popular size is 32 ounces. As proposed by Guadagni and Little
(1983), we treat each brand-size combination as a separate choice. We
consider six brand-size combinations of the national brands. We treat
the 32 ounce size of the remaining brands as the seventh brand-size
combination. Henceforth, we will refer to brand-size combination as a
brand. The shares of each brand are given in table 1. We consider a
data set with 225 households with at least 10 purchases of ketchup
during a 2-year period. Of these households, we randomly select 150
households for estimation and keep the remaining 75 households as a
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
242 Journal of Business
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
Gamma (γ) ⋅⋅⋅ ⋅⋅⋅ ⫺.966 ⫺11.08 ⫺.681 ⫺2.90 ⫺.464 ⫺3.38
BL2 (β 9 ) ⋅⋅⋅ ⋅⋅⋅ .212 2.11 .934 2.17 .869 2.16
BLP2 (β 10 ) ⋅⋅⋅ ⋅⋅⋅ 3.225 19.36 3.059 11.45 3.126 14.33
SL2 (β 11 ) ⋅⋅⋅ ⋅⋅⋅ .092 .83 .821 2.61 .934 2.48
Brand Choice
SLP2 (β 12 ) ⋅⋅⋅ ⋅⋅⋅ 1.167 6.28 .061 .27 .003 .02
σδ ⋅⋅⋅ ⋅⋅⋅ 1.825 10.08 ⋅⋅⋅ ⋅⋅⋅
A1 ⋅⋅⋅ ⋅⋅⋅ ⫺7.871 ⫺10.28
A2 ⋅⋅⋅ ⋅⋅⋅ 3.349 9.90
D1 ⋅⋅⋅ ⋅⋅⋅ ⫺.691 ⫺6.80
D2 ⋅⋅⋅ ⋅⋅⋅ ⫺.239 ⫺9.58
Log-likelihood ⫺1,629.80 ⫺1,601.90 ⫺1,554.90 ⫺1,550.40
U 2 null model .443 .452 .468 .470
U 2 equally likely model .555 .563 .575 .577
AIC 3,285.60 3,239.80 3,147.80 3,144.80
BIC 3,357.63 3,339.53 3,253.07 3,266.67
Log-likelihood on holdout sample ⫺799.55 ⫺785.67 ⫺779.03 ⫺777.79
Note.—See text for explanation of parameters. AIC ⫽ Akaike information criterion; BIC ⫽ Bayesian information criterion.
245
Purchase
Household Occasion
Heterogeneity Heterogeneity
A1 ⋅⋅⋅ ⫺6.021 3.79 ⫺.387 ⫺2.78
A2 ⋅⋅⋅ 3.234 3.24 .433 3.22
D1 ⋅⋅⋅ ⫺.624 ⫺2.58 1.078 2.49
D2 ⋅⋅⋅ ⫺.359 ⫺2.48 .262 1.98
Log-likelihood ⫺1,519.23 ⫺1,514.36
U 2 null model .481 .482
U 2 equally likely .585 .587
model
AIC 3,096.46 3,098.72
BIC 3,257.14 3,292.64
Log-likelihood on
holdout sample ⫺765.74 ⫺765.37
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
248 Journal of Business
proclivity for limited evaluation. Our findings on store loyalty and deal-
proneness are consistent with the evidence presented in Bucklin and
Lattin (1989). The empirical evidence suggests no significant effect of
frequency of purchase on evaluation propensity.
4. Demographics and restricted evaluation. High-income consum-
ers, as expected, are less likely to evaluate (γ 5 ⬍ 0) because of higher
valuation of their time. In contrast, consumers with higher education
exhibit a greater evaluation propensity (γ 6 ⬎ 0), perhaps due to better
ability to process information. As noted earlier, while the correlation
between education and income is positive, it is not high enough to cause
concern about collinearity. Time availability (TIME) increases the av-
erage propensity to evaluate extensively the brands (γ 9 ⬎ 0).
5. Impact of marketing mix variables on brand choice. All parame-
ter estimates of the marketing activity variables are significant and are
in the expected direction in all the models. In addition, we find that
the effect of the marketing mix variables in the extended evaluation
state is 20%–60% higher than that of the baseline logit model.
6. Evaluation propensity and brand choice probability. In the ap-
pendix, we derive the effect of evaluation propensity on the choice
probability. We see that the rate of change of choice probability with
evaluation is a function of the difference in the conditional probabili-
ties, that is, of [prob kn ( j| EE) ⫺ prob kn ( j |HE)]. This implies that, if
a firm believes the probability of choice for its brand increases with
evaluation (say, it has a more attractive price), then it pays for the
firm to encourage evaluation of the offering. But if there is a negative
difference between the conditional probabilities, the firm is better off
by discouraging consumer evaluation.
7. Performance of the models on the holdout sample. An important
disaggregate measure of performance of the models is to test if the
likelihood improvements of the models with the evaluation structure
carries through in the holdout sample. In tables 2 and 3, we also report
the log-likelihood values computed on holdout sample for all models
based on the parameters obtained from the calibration sample. We find
that the likelihood values mirror the pattern on the estimation sample.
The predictions of choices on holdout samples also favor the proposed
model.
B. Evaluation Propensity: A Household Trait?
In spite of the strong effect of the factors that constrain time or measure
a household’s valuation of time on extent of evaluation, the variance
explained by unobserved factors is more than the observed factors.
Hence, we conclude that there is a strong idiosyncratic component to
the extent of evaluation. The nonparametric distribution yields distinct
extended-evaluation and habitual-evaluation segments. These segments
bear semblance to the notion of individuals who like or dislike shop-
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
250 Journal of Business
V. Conclusion
In this article, we raise the possibility that a time-constrained consumer
may not evaluate all the marketing inputs on each purchase occasion for
the brands in a category. Several demographic and purchase-occasion
characteristics that constrain a household’s time or reflect opportunity
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
Brand Choice 253
冢 冣
S H
exp γ 0 ⫹ 冱 γ s Z skn ⫹ 冱γ Z h hn
冢 冣
S H
1 ⫹ exp γ 0 ⫹ 冱γ Z
s⫽1
s skn ⫹ 冱γ
h⫽1
h Z hn
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
254 Journal of Business
and
冢 冣
S H
exp γ 0 ⫹ 冱 γ s Z skn ⫹ 冱γ Z h hn
⫽ s⫽1 h⫽1
(A4)
冤1 ⫹ exp冢γ ⫹ 冱 γ Z ⫹ 冱 γ Z 冣冥
S H 2
0 s skn h hn
s⫽1 h⫽1
References
Alba, J. W., and Hutchinson, J. W. 1987. Dimensions of consumer expertise. Journal of
Consumer Research 13 (March): 411–54.
Bass, F. M. 1974. The theory of stochastic preference and brand switching. Journal of
Marketing Research 11 (February): 1–20.
Beatty, S. E., and Smith, S. M. 1987. External search effort: An investigation across several
product categories. Journal of Consumer Research 14 (June): 83–95.
Belk, R. W. 1975. Situational variables and consumer behavior. Journal of Consumer Re-
search 2 (December): 157–64.
Bucklin, R. E., and Lattin, J. M. 1991. A two state model of purchase incidence and brand
choice. Marketing Science 10 (Winter): 24–39.
Butler, J. S., and Moffitt, R. 1982. A computationally efficient quadrature procedure for
the one factor multinomial probit model. Econometrica 50:761–64.
Carlson, J. A., and Gieseke, R. J. 1983. Price search in a product market. Journal of Con-
sumer Research 9:357–65.
Chintagunta, P. K.; Jain, D. C.; and Vilcassim, N. J. 1991. Investigating heterogeneity in
brand preferences in logit models for panel data. Journal of Marketing Research 28
(November): 417–28.
Dickson, P. R., and Sawyer, A. G. 1990. The price knowledge and search of supermarket
shoppers. Journal of Marketing 54 (July): 42–53.
Doti, J. L., and Sharir, S. 1981. Households’ grocery shopping behavior in the short run:
Theory and evidence. Economic Inquiry 19 (April): 196–208.
Dunn, R.; Reader, S.; and Wrigley, N. 1983. An investigation of the assumptions of the
NBD model as applied to purchasing at individual stores. Applied Statistics 3:249–59.
Engel, J. F.; Kollat, D. T.; and Blackwell, R. D. 1973. Consumer Behavior. 2d ed. New
York: Holt, Rinehart, & Winston.
Fader, P. S., and McAlister, L. 1990. An elimination by aspects model of consumer re-
sponse to promotion calibrated on UPC scanner data. Journal of Marketing Research
27 (August): 322–32.
Farley, J. U. 1964. Brand loyalty and the economics of information. Journal of Business
37:370–81.
Frank, R. E.; Green, P. E.; and Seiber, H. F., Jr. 1967. Household correlates of purchase
price for grocery products. Journal of Marketing Research 4:54–58.
Gensch, D. H. 1987. A two-stage disaggregate attribute choice model. Marketing Science
6 (Summer): 223–39.
Goldman, A., and Johansson, J. K. 1978. Determinants of search for lower prices: An
empirical assessment of the economics of information theory. Journal of Consumer Re-
search 5 (December): 176–86.
Gonul, F., and Srinivasan, K. 1993. Modeling unobserved heterogeneity in multinomial
logit models: Methodological and managerial implications. Marketing Science 12:213–
229.
Grover, R., and Srinivasan, V. 1987. A simultaneous approach to market segmentation
and market structuring. Journal of Marketing Research 24 (May): 139–53.
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
Brand Choice 255
Guadagni, P. M., and Little, J. D. C. 1983. A logit model of brand choice. Marketing
Science 2 (Summer): 203–38.
Hauser, J. R. 1978. Testing the accuracy, usefulness, and significance of probabilistic
choice models: An information theoretic approach. Operations Research 6 (May–June):
406–21.
Hauser, J. R., and Wernerfelt, B. 1989. The competitive implications of relevant-set/re-
sponse analysis. Journal of Marketing Research 26(4):391–405.
Hauser, J. R., and Wernerfelt, B. 1990. An evaluation cost model of consideration sets.
Journal of Consumer Research 16 (March): 393–408.
Heckman, J. J. 1981. Statistical models for discrete panel data. In C. F. Manski and
D. McFadden (eds.), Structural Analysis of Discrete Data with Econometric Applica-
tions. Boston: MIT Press.
Heckman, J. J., and Singer, B. 1984. A method of minimizing the impact of distributional
assumptions in econometric models for duration data. Econometrica 52(2):271–320.
Howard, J. A. 1974. The structure of buyer behavior. In J. V. Farley, J. A. Howard, and
L. W. Ring (eds.), Consumer Behavior: Theory and Application. Boston: Allyn & Bacon.
Hoyer, W. D. 1984. An examination of consumer decision making for a common repeat
purchase product. Journal of Consumer Research 11 (December): 822–29.
Kahn, B. E., and Morrison, D. G. 1989. A note on ‘‘random’’ purchasing: Additional
insights from Dunn, Reader and Wrigley. Applied Statistics 38(1):111–14.
Kamakura, W., and Russell, G. 1989. A probabilistic choice model for market segmentation
and elasticity structure. Journal of Marketing Research 26 (November): 379–90.
Katona, G., and Mueller, E. 1955. A study of purchase decisions. In L. H. Clark (ed.),
Consumer Behavior: The Dynamics of Consumer Reaction. New York: New York Uni-
versity Press.
Lattin, J. 1987. A model of balanced choice behavior. Marketing Science 6 (Winter): 48–
65.
Marmorstein, H.; Grewal, D.; and Fishe, R. P. H. 1992. The value of time spent in price-
comparison shopping: Survey and experimental evidence. Journal of Consumer Re-
search 19:52–61.
Murthi, B. P. S., and Rao, R. C. 1997. Rational expectations and consumer brand choice.
Working paper. Dallas: University of Texas at Dallas.
Nelson, P. 1970. Information and consumer behavior. Journal of Political Economy
78:311–29.
Newman, J. W., and Staelin, R. 1972. Prepurchase information seeking for new cars and
major household appliances. Journal of Marketing Research 9 (August): 249–57.
Olshavsky, D. H., and Granbois, R. W. 1979. Consumer decision making—fact or fiction?
Journal of Consumer Research 6 (September): 93–100.
Park, C. W.; Iyer, E. S.; and Smith, D. C. 1989. The effects of situational factors on in-
store grocery shopping behavior: The role of store environment and time available for
shopping. Journal of Consumer Research 15 (March): 422–33.
Petty, R. E., and Cacioppo, J. T. 1986. Communication and Persuasion: Central and Pe-
ripheral Routes to Attitude Change. New York: Springer-Verlag.
Ratchford, B. 1982. Cost-benefit models for explaining consumer choice and information
seeking behavior. Management Science 28 (February): 197–212.
Roberts, J. H. 1989. A grounded model of consideration set size and composition. Advances
in Consumer Research 16:749–57.
Roberts, J. H., and Lattin, J. M. 1991. Development and testing of a model of consideration
set composition. Journal of Marketing Research 28(4):429–40.
Rossi, P. E.; McCulloch, R. E.; and Allenby, G. M. 1996. The value of purchase history
data in target marketing. Marketing Science 15(4):321–40.
Shugan, S. 1980. The cost of thinking. Journal of Consumer Research 7 (September): 99–
111.
Stigler, G. J. 1961. The economics of information. Journal of Political Economy 60 (June):
213–25.
Stroud, A., and Secrest, S. 1966. Gaussian Quadrature Formulas. Englewood Cliffs, N.J.:
Prentice Hall.
Sujan, M. 1985. Consumer knowledge: Effects on evaluation strategies mediating con-
sumer judgments. Journal of Consumer Research 12 (June): 31–46.
Udell, J. G. 1966. Prepurchase behavior of buyers of small electrical appliances. Journal
of Marketing 30 (October): 50–52.
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms
256 Journal of Business
Vilcassim, N. J., and Jain, D. C. 1991. Modeling purchase-timing and brand switching
behavior incorporating explanatory variables and unobserved heterogeneity. Journal of
Marketing Research 28 (February): 29–41.
Wilde, L. L. 1980a. The economics of consumer information acquisition. Journal of Busi-
ness 53(3, suppl.): S143–S165.
Wilde, L. L. 1980b. On the formal theory of inspection and evaluation in product markets.
Econometrica 48(5):1265–79.
This content downloaded from 115.164.221.22 on Sun, 13 Aug 2017 13:38:36 UTC
All use subject to http://about.jstor.org/terms