ideology and brand consumption - london business...
TRANSCRIPT
1
Ideology and Brand Consumption
Romana Khan, Kanishka Misra, Vishal Singh1
July 13, 2012
Abstract:
Do mundane daily choices, such as what brands to buy in a supermarket, reflect aspects of
our values and ideologies? This article presents a large scale field study to test whether traits
associated with a conservative ideology, as measured by voting behavior and religiosity,
manifest in routine, seemingly inconsequential product choices that consumers make. Across
a variety of frequently purchased products, we show that both measures of conservatism are
associated with a systematic preference for established national brands (as opposed to their
generic substitutes), and a lower propensity to try newly launched products. These findings
correspond with the psychological traits associated with a conservative ideology such as
preference for tradition and status quo, ambiguity/uncertainty avoidance, and skepticism
towards new experiences.
1 Romana Khan is Assistant Professor of Marketing at Ozyegin University, Istanbul, Turkey, Kanishka Misra is
Assistant Professor of Marketing at London Business School, Vishal Singh is Associate Professor of Marketing
at the Stern School of Business, New York University. Correspondence can be sent to: [email protected]
(Singh).
2
Ideology and Brand Consumption
Commentary in the popular press as well as opinion polls reveal polarization in the US
population on a range of issues such as the role of government and taxes, affirmative action
and aid to the disadvantaged, gun control, the death penalty, stem cell research, abortion, and
the sanctity of marriage. Across a large set of empirical studies, a self-reported left-right or a
conservative-liberal scale has been found to provide a useful and parsimonious
approximation to opinion on these wide ranging issues [1,2]. Recent research in social and
political psychology provides a theoretical framework for the liberal-conservative divide by
linking ideological proclivities to dispositional (or situational) differences in psychological
needs, cognitive styles, and personality traits [3,4,5,6]. For example, individuals who
gravitate towards conservative ideology are found to score low on integrative complexity,
openness to new experiences, and tolerance for uncertainty/ambiguity; and relatively high on
measures of conscientiousness, dogmatism, and need for order, structure, and closure [5,7,8].
In addition to explaining enduring differences in public opinions and attitudes, the
psychological approach to the study of ideology has provided a rich account for differences in
many theoretical domains such as the basis of moral foundations [9] and system justification
[10].
While the role of ideological differences in the socio-political domain is intuitive, can
psychological traits associated with broad ideologies also get reflected in mundane,
seemingly inconsequential daily choices? Consider religious affiliation for instance.
Religiosity has been shown to influence important life decisions such as marriage, school
attendance, participation in crime, and extra-marital affair [11]. Similarly, traits associated
with religiosity, for example risk aversion, manifest not only in individual attitudes towards
risks such as fastening of seat belt, maintaining a financial cushion, and purchase of medical
and auto insurance, but have also been shown to manifest at an organizational level, whereby
firms located in counties with higher levels of religiosity display lower degrees of financial
risk exposure [12]. Will similar phenomenon also get reflected in trivial choices such as
preference for established national brands as opposed to “riskier” generic alternatives? If
conservative values are associated with a preference for the status quo and skepticism
towards new experiences, will this translate to a lower propensity to try new products and
services? Will these traits manifest in even low involvement decisions such as trying a new
flavor of yogurt on the supermarket shelf? This article uses an extensive field data on product
purchases and measures of conservativeness to investigate these questions.
Products and brands can of course serve as a source of self-expression, and firms devote
substantial advertising resources to develop brand images and elicit specific feelings,
attitudes, and thoughts for their products. Consumers often develop deep rooted associations
with products to the point of associating human characteristics with brands [13], and often
choose products as a means of signaling aspects of their personality. For example, a market
research study on buyers of Toyota Prius shows that the top reason for buying the car was
that it “makes a statement about me” (“fuel efficiency” was ranked fifth most important)
[NY Times, July 4 2007]. Systematic differences in lifestyles and attitudes as often reported
in commercial surveys for say hybrid versus hummer drivers, or Mac versus PC users, is not
3
particularly surprising given the well defined positioning of such products. However, the
majority of such image products tend to be socially visible and are often more expensive (e.g.
hi-tech gadgets, fashion, and automobiles), so they can serve as a signal of social status or
make a statement about individual personality. The focus in this article instead is on low
involvement, frequently purchased products sold at supermarkets that are primarily for
private consumption. Extensive evidence exists that our judgments and behaviors are often
guided by implicit cognition that is spontaneous, effortless, and unconscious [14,15],
suggesting that even mundane daily choices such as what brand of detergent to buy or
whether to try a new brand of breakfast cereal, can reflect aspects of our deep-rooted
ideologies, values, and personality traits.
Our empirical strategy relies on a comprehensive scanner data from a variety of frequently
purchased products sold at supermarkets. We use these data to create county level measures
of “brand consumption” and relate those to conservativeness. To operationalise conservatism,
we create measures of political voting and religiosity, both of which have a high degree of
correspondence with conservative values. For example, both political voting and religiosity
associate positively with values that preserve order and tradition, protect against uncertainty
and threat, and negatively with measures of openness to new experiences and change [5, 16].
Research in cognitive neuroscience suggests that both political orientation and religiosity are
marked with similar reduced reactivity in the anterior cingulated cortex (ACC), a cortical
system that is involved in cognition and emotion [17, 18].
Our analysis exploits two aspects of the consumer packaged (CPG) industry to create
measures of brand consumption. First, most CPG categories in the US comprise of several
established national brands as well as generics2. Industry reports and academic research in
marketing suggests that generics are perceived by consumers as riskier and of lower quality
than national brands [19,20]. Since a major function of branding is to lower uncertainty and
simplify decision making [13], we might expect that, controlling for price and other socio-
economic factors, aspects of conservative values such as preference for tradition and
convention, and lower tolerance for ambiguity and complexity may get reflected in higher
reliance on national brands as opposed to generics. A second characteristic of the CPG
industry is frequent introductions of new products. In our data we observe over 4,000 new
products ranging from new brands to minor modifications of existing products (e.g. a new
flavor). Personality traits such as skepticism to new experiences linked with conservative
ideology might get reflected in lower acceptance and penetration of new products. Thus,
unlike previous research that primarily focuses on explicit and consciously accessible self-
reported measures of attitude and opinions [5], we hypothesize that aspects of ideological
differences may exist not just in a reasoned and explicit form [21], but may be reflected in
our routine daily behavior.
METHOD
2 We use the terms generics, store brands, and private label interchangeably in the paper.
4
The cornerstone of our empirical strategy is a comprehensive scanner database provided by
IRI [22]. The data provides weekly store sales for thousands of products organized into 26
product categories. These categories include both edible (e.g., frozen pizza, canned soup) and
non-edible (e.g., paper towels, laundry detergent) products. The data are obtained from 1,860
stores belonging to 135 supermarket chains, spanning a period of 6 years from 2001 to 2006.
There are 416 counties represented in the data, which cover 47% of the total US population.
As noted above, each product category includes several national brands and a store brand
(generic) alternative. In addition, we observe 4,151 new product launches across our 26
categories during the course of the data. Using these data, we create two measures of brand
consumption at the county level: (1) market share of generics in each category, and (2)
market share of new product introductions in the year after launch. Summary statistics on
both measures are reported in Table 1. The average market share of generics, across all
stores, categories and years, is 16%. However, there is considerable variation in market
shares across categories; and across markets within a category. The average new product
share is 1.3%; with a large degree of heterogeneity across counties and categories.
Conservative Ideology
Merriam-Webster defines conservatism as "a disposition in politics to preserve what is
established", or "the tendency to prefer an existing or traditional situation to change". We
operationalise conservatism using measures of political voting and religiosity. To provide
empirical support for using these measures, we analyzed the General Social Survey [GSS;
National Opinion Research Center, 2011] from 1972 to 2010, and the American National
Election Studies (ANES) survey data from 1948 to 2008. Both surveys contain self-reported
measures of ideology (1-extremely liberal to 7-extremely conservative) and political party
affiliation (1-strong Democrat to 7-strong Republican). For religiosity, we use the frequency
with which respondents attend religious services in the GSS data (measured on 9 point scale
1-Never to 9-More than once a week), and a binary indicator (Is religion important to the
respondent?) in the ANES survey. Table 2 provides the raw correlations between these
measures from the two surveys. The correlations suggest that both political affiliation and
religiosity capture aspects of conservative values, independent of each other.
County Level Measures of Religiosity & Political Voting
Religiosity is a complex, multi-dimensional construct. It encompasses the various levels at
which religion impacts people, including their cognitive values and beliefs, affective feelings
of spirituality and commitment, and behaviors such are prayer and church attendance. Since
our product purchase data are available at the aggregate county level, we cannot rely on
survey based self-reported measures of religiosity. Instead we use data provided by the
Association of Religion Data Archives (ARDA) on religious activity at the county level.
ARDA collects information on membership, adherence, and number of congregations for
major religions, and then provides further detail by breaking it down to the denomination
level. Our measures of religiosity are constructed from data contained in the 2000 report. The
primary measure of religiosity used in the empirical application is Adherence. The definition
of adherents is quite stringent, counting only full members of religious denominations and
5
non-members who attend services regularly. As such, it serves as a relatively accurate
measure of the population that is truly engaged in religious activity, rather than including
those who might identify with a particular denomination but not actively practice or engage
through participation.
To assess the political voting, we use county level votes in presidential elections between
1980 and 2008 [http://uselectionatlas.org]. Rather than rely on any particular election cycle,
our main measure of political conservatism uses the average percentage of Republican votes
in all elections between 1980 and 2008 (8 presidential elections). We also incorporate an
extensive set of demographic variables to control for other factors that might impact brand
consumption. These data are obtained from the US Census. The variables included are (a)
Income - median income, (b) Elderly - percent of the population over 65, (c) Unemployed -
percent unemployed, (d) Education - average number of years of education, (e) HHsize -
average household size, (f) Black - % African American population. Finally, we control for
store size using All Commodity Volume (ACV), which is a standard measure used in the retail
industry. Descriptive statistics for these measures are shown in Table 3.
RESULTS
Given the data above, our empirical approach involves conducting a series of regressions of
brand consumption on measures of conservatism. Results from the regression model are
presented in Table 4. The dependent variables in these regressions are the market shares of
the generics (left panel) and the market share of the new products in the year after launch
(right panel) 3
. The regressions control for the marketing mix variables (price, promotion) and
other socio-economic characteristics. Furthermore, while the quality of national brand (e.g.
Tide or Coca Cola) is constant across markets, the quality of non-branded generics may vary
across product types and retail chains. Hence, all regressions include a set of category and
chain fixed effects that control for any quality differences4.
The first row in Table 4 shows the results estimated by pooling our data from all categories,
and the subsequent rows show the regression estimates for each product category separately.
Looking first at the pooled estimates for the generics (left panel, top row), we find that the
coefficients associated with both religiosity and Republican are negative and statistically
significant, indicating that market shares for generics are significantly lower in conservative
counties. It is important to note that these estimates are identified based on the variation in
market shares between stores that belong to the same retail chain serving different counties.
Thus, they capture the net impact of conservativeness after controlling for a variety of socio-
economic characteristics, marketing mix variables, and a set of fixed effects that absorb any
differences in product quality across retail chains. The subsequent rows in the left panel show
the results from the category level analysis, where the dependent variable is the generic brand
3 We transform the share to log(share/1-share). This is a monotonic transformation that ensures full support on
the real line. 4 We do not report estimates for the control variables to conserve space. Full regression results are reported in an
online appendix available from the authors.
6
share in each category. The category level results are quite telling. In 19 out of the 26
categories, the effect of religiosity on generic brand penetration is negative and statistically
significant. For five categories the effect is insignificant, and there is just a single category
where religious counties are associated with higher market shares for generics. With minor
exceptions, the pattern repeats for the republican voting estimates as well5. Overall, 38 of the
52 (73%) of the coefficients are negative and statistically significant; 11 coefficients are
insignificant, and only 3 of the 52 coefficients are positive and significant. These results
provide strong evidence that conservative markets are associated with lower market shares
for generics and a higher reliance on established national brands.
The right panel in Table 4 provides estimates for the new product penetration. Recall that we
observe over 4,000 new products in our data across the 26 product categories. The first row
shows the pooled (across all categories) estimates, where we find that the market share of
new products is significantly lower in counties with higher levels of religiosity and
Republican vote. Looking at the category level estimates, we again find either negative (63%
of the estimates) or insignificant (37% of the estimates) coefficients for both religiosity and
voting. In none of the categories do we find higher penetration of new products in
conservative counties. In figure 1, we plot the marginal impact of one standard deviation
increase in conservativeness (religiosity and Republican party vote) on market shares of
generics and new products. The vertical line shows the overall impact pooled across all
products, while the bars show the marginal impact on individual product categories. With
minor exceptions, all estimates are negative and reasonably large. Taken together, these
results provide strong evidence that more conservative markets are associated with a higher
reliance on established national brands, and a lower penetration of new products.
Robustness Checks
Although not displayed to conserve space, we conducted a variety of robustness checks using
alternate measures of religiosity and political voting. First, the ARDA data provides measures
of Adherence at the denomination level. We repeated the entire analysis using the number of
adherents in each county who identify as Evangelical Protestant, Mainline Protestant,
Catholic, Jewish and Islamic. Second, we repeat the entire analysis using the 2000 County
Business Patterns published by the U.S. Census Bureau. The Census Bureau collects
information on the number of religious establishments defined as "those establishments
primarily engaged in operating religious organizations, such as churches, religious temples,
and monasteries and/or establishments primarily engaged in administering an organized
religion or promoting religious activities." Finally, for political voting, we also consider just
2004 presidential election as this coincides most closely with our sales data.
In addition to alternative measures of conservativeness, we also tested for differences in store
size, product availability, and breadth of generic brand offerings (# of UPC’s available) and
found no systematic differences based on conservative measures. Finally, we replicated our
5 Estimates for two categories (Toothpaste and Milk) are positive. Note that these are somewhat unusual
categories. Toothpaste has extremely low market share for generics (0.5%) and Milk has very large market share
(over 75%). See Table 1.
7
results for both generic brand share and new products using flexible Quantile regressions as
opposed to OLS6.
DISCUSSION
A large body of recent research has examined the social, cognitive, and motivational
underpinnings of political orientation [10], and the consequences of ideological differences
on attitudes and evaluations of socio-political issues, preference for political parties and
candidates, and voting behavior [23]. A relatively smaller literature has linked ideological
differences beyond the realm of politics, such as attitudes and preferences for various
institutions, and activities such as travel, films, and television [24, 25]. We add to this line of
research by examining whether values and tendencies underlying conservative ideology get
manifested in routine purchase decisions. Our empirical results, based on an extensive field
data, provide strong evidence that conservative ideology is associated with higher reliance on
established national brands (as opposed to generics) and a slower uptake of new products.
These results are consistent with traits typically associated with conservatism such as higher
risk aversion, skepticism towards new experiences, and a general preference for tradition,
convention, and status quo.
It is noteworthy that the results reported in the paper manifest in routine, low-involvement
product purchases that rarely exceed a few dollars. The inherent risks of product trial or post-
purchase dissonance are likely to be minimal for these products. Research in other domains
may reveal similar manifestation of psychological differences on purchase or consumption
behavior. For example, a large literature in psychology reports systematic differences in
cognitive processes between subjects from individualistic (Western) and collectivist (Eastern)
cultures [e.g., 26, 27, 28], which translates into differences in choice and decision making
[29, 30]. It would be interesting to examine whether these cultural differences permeate
aspects of daily activities and purchase behaviors. Our work is also related to the growing
body of research examining the impact of transmitted norms, beliefs, and behavior based on
ethnic, social class, or regional differences that may translate into differences in preferences
for products and practices [31,32,33,34].
There are of course several caveats to our analysis that merit caution. First, our analysis is
restricted to a specific set of frequently purchased functional product categories. Consumer
behavior for ostentatious or conspicuous products may be quite different [31]. Second, our
analysis is conducted at an aggregate county rather than individual level. An ideal data to
examine the relationship between ideology and brand consumption would be a consumer
panel with accurate measures of political and religious attitudes and purchase behavior over a
variety of products (including durables and fashion). However, such field data are rarely
available. Notwithstanding these limitations, this paper provides, to our best knowledge, the
first evidence of a relationship between ideology and brand consumption using an extensive
field data. The consistency of our results across a large set of product categories suggests that
aspects of ideological differences may indeed be reflected in our daily behavior at an
unconscious level or implicit manner. 6 Results of all robustness checks are reported in an online appendix available from authors.
8
References
1. J. Bafumi, R.Y. Shapiro, A New Partisan Voter The J. of Politics 71, 1-24 (2009).
2. J.T. Jost, The end of the end of ideology, Am. Psychologist 61, 651-670 (2006).
3. J. Z. Muller, Conservatism: Historical aspects. In N. J. Smelser & P. B. Baltes (Eds.),
International encyclopedia of the social and behavioral sciences (2624–2628).
Amsterdam: Elsevier (2001)
4. J. Duckitt, Adv. in Experimental Social Psychology 33, 41–113 (2001).
5. J.T. Jost, J. Glaser, A.W. Kruglanski, F. Sulloway, Political conservatism as
motivated social cognition, Psychological Bulletin 129, 339-375 (2003).
6. J.T. Jost, J. Glaser, A.W. Kruglanski, F. Sulloway, Exceptions that prove the rule:
Using a theory of motivated social cognition to account for ideological incongruities
and political anomalies, Psychological Bulletin 129, 383-393 (2003).
7. B. Altemeyer, The authoritarian specter, Harvard Univ. Press (1996).
8. R.R. McCrae, Social consequences of experiential openness, Psychological Bulletin
120, 323-337 (1996).
9. J. Haidt, J. Graham, When Morality Opposes Justice: Conservatives Have Moral
Intuitions that Liberals may not Recognize, Social Justice Research 20, 98-116
(2007).
10. J.T. Jost, A.C. Kay, H. Thorisdottir, Social and psychological bases of ideology and
system justification, Oxford Univ. Press (2009).
11. Michael E. McCullough, Brian L. B. Willoughby, Religion, self-regulation, and self-
control: Associations, explanations, and implications, Psychological Bulletin, 135(1),
69-93 (2009).
12. G. Hilary, K.W. Hui, Does religion matter in corporate decision making in America?,
Journal of Financial Economics 93, 455–473 (2009).
13. D.A. Aaker, Managing Brand Equity, Free Press (1991).
14. J.A. Bargh, The Four Horsemen of automaticity: Awareness, efficiency, intention, and
control in social cognition, in Handbook of social cognition 2nd ed., R.S. Wyer, Jr.,
T.K. Srull Eds. (Psychology Press), 1-40 (1994).
15. A.G. Greenwald, M. R. Banaji, Implicit social cognition: Attitudes, self-esteem, and
stereotypes, Psychological Review 102, 4-27 (1995).
16. S.H. Schwartz, S. Huismans, Value Priorities and Religiosity in Four Western
Religions, Social Psychology Quarterly 58, 88-107 (1995).
17. D.M., Amodio, J.T. Jost, S.L. Master, C.M. Yee, Neurocognitive correlates of
liberalism and conservatism, Nature Neuroscience 10, 1246-1247 (2007).
18. M. Inzlicht, I. McGregor, J.B. Hirsh, K. Nash, Neural markers of religious conviction.
Psychological Science 20, 385-392 (2009).
19. T. Erdem, Y. Zhao, A. Valenzuela: Performance of Store Brands, Journal of
Marketing Research, 41(1), (2004).
20. ADWEEK: Why risk prevents switching to Private Labels, July 10, 2009
21. J Jost, M. R. Banaji, B. A. Nosek, A decade of system justification theory. Political
Psychology, (2004).
22. B.J. Bronnenberg, M.W. Kruger, C.F. Mela, The IRI Marketing Data Set, Marketing
Science 27, 745-748 (2008).
23. R. S. Erikson, K. L. Tedin. American Public Opinion: Its Origins, Content, and
Impact, Boston: Allyn and Bacon, Seventh edition (2007).
9
24. D. Carney, J. T. Jost, S.D. Gosling, J. Potter, The Secret Lives of Liberals and
Conservatives: Personality Profiles, Interaction Styles, and the Things They Leave
Behind Political Psychology, 29(6), 807–840 (2008).
25. J.T. Jost, B.A. Nosek, S.D. Gosling, Ideology: Its resurgence in social, personality,
and political psychology, Perspectives on Psychological Science, 3,126-136 (2008).
26. S. J. Heine, D. R. Lehman, H. R. Markus, S. Kitayama, Is there a universal need for
positive self-regard?, Psychological Review, 106, 766-794 (1999).
27. R. E. Nisbett , K. C. I. Peng , A. Norenzayan , Culture and systems of thought:
Holistic vs. analytic cognition, Psychological Review, 108, 291-310 (2001).
28. Mesquita, B. (2001). Emotions in collectivist and individualist contexts. Journal of
Personality and Social Psychology, 80(1), 68-74
29. S. S. Iyengar, M. R. Lepper, Rethinking the value of choice: A cultural perspective
on intrinsic motivation, Journal of Personality and Social Psychology, 76(3), 349-366
(1999).
30. H. Kim, H.R. Markus, Deviance or uniqueness, harmony or conformity? A cultural
analysis, Journal of Personality and Social Psychology, 77(4), 785-800 (1999)
31. Nisbett, R. E. (1993). Violence and U.S. regional culture. American Psychologist, 48,
441-449.
32. Plaut, V. C., Markus, H. R., & Lachman, M. E. (2002). Place matters: Consensual
features and regional variation in American well-being and self. Journal of Personality
and Social Psychology, 83, 160 –184.
33. Snibbe, A. C., & Markus, H. R. (2005). You can’t always get what you want:
Educational attainment, agency, and choice. Journal of Personality and Social
Psychology, 88, 703–720.
34. Rentfrow, P. J., Gosling, S. D. & Potter, J. (2008). A Theory of the emergence,
persistence, and expression of regional variation in basic traits. Perspectives on
Psychological Science, 3, 339-369
35. R. Shachar, T. Erdem, K. Cutright, and G.J. Fitzsimons. Brands: The opiate of the
non-religious masses? Articles in Advance, Marketing Science, 2010.
10
Table 1: Summary statistics for generic and new brand shares
Generic Brands Share New Products Share
Category Mean Std Dev # Products Mean Std Dev
Blades 14.72 7.91 83 1.94 7.91
Coffee 12.57 9.31 254 1.02 9.31
Deodorant 0.83 1.12 148 0.94 1.12
Diapers 21.33 11.35 22 4.41 11.35
Frozen Pizza 12.09 8.85 39 3.35 8.85
Frozen Dinner 1.39 1.86 183 1.20 1.86
Household Cleaner 6.79 5.45 117 1.05 5.45
Hot Dog 9.85 8.00 133 1.73 8.00
Laundry Detergent 6.49 5.83 180 2.16 5.83
Margarine/Butter 13.17 8.34 61 2.29 8.34
Mayonnaise 12.95 8.39 56 2.07 8.39
Milk 75.72 19.82 78 1.86 19.82
Mustard/Ketchup 23.29 8.78 192 1.31 8.78
Toothpaste 0.55 1.02 209 0.82 1.02
Peanut Butter 24.95 10.59 29 6.18 10.59
Photographs 20.12 15.79 45 2.08 15.79
Razors 6.00 6.91 30 7.96 6.91
Salty Snacks 9.73 6.89 708 0.28 6.89
Sauces 7.75 5.45 504 0.55 5.45
Soda 11.21 9.57 301 0.53 9.57
Soup 11.46 6.38 244 0.84 6.38
Sugar Substitute 10.96 9.73 45 4.20 9.73
Tissue Paper 24.88 11.91 38 3.14 11.91
Toilet Paper 18.39 11.19 162 1.30 11.19
Toothpaste 17.81 8.54 150 1.01 8.54
Yogurt 22.13 12.50 140 1.31 12.50
All categories 16.48 17.18 4,151 1.29 17.18
11
Table 2: Correlation between Ideology, Political Affiliation, and Religiosity. *
represents significant at a 95% confidence level.
Table 3: County level summary statistics
Correlation Liberal-Conservative Political Party
Source: General Social Survey
Political Party Affiliation 0.32*
Religiosity 0.18* 0.03*
Political Party Affiliation 0.39*
Religiosity 0.17* 0.000
Source: American National Election Survey
Variable Mean St. Dev
Religiosity (Adherence) 50% 11%
% Republican Votes 52% 12%
Demographics
Income 55,318 11,452
Elderly 11.63 2.60
% Unemployed 5.30 1.67
Education 13.52 0.86
HH Size 2.66 0.18
African American 12.89 13.04
ACV ($MM) 23.48 13.39
12
Table 4: Parameter estimates from the regression of conservative and controls on generics
share and new product share by category. Number in brackets represents the standard error
and the * represents significant at a 95% confidence level.
Generic Brand Share New Product Share
Religion Republican Religion Republican
All categories -0.26 (0.02) * -0.34 (0.02) * -0.43 (0.03) * -0.49 (0.04) *
By category
Blades -0.21 (0.07) * -0.61 (0.07) * -0.36 (0.11) * -0.47 (0.12) *
Coffee 0.00 (0.07) -0.34 (0.07) * -1.04 (0.16) * -0.59 (0.17) *
Deodorant 0.18 (0.11) -0.04 (0.11) -0.40 (0.09) * -0.54 (0.09) *
Diapers -0.65 (0.07) * -0.35 (0.07) * -0.41 (0.20) * -0.71 (0.21) *
Frozen Pizza -0.07 (0.07) -0.15 (0.07) * -0.77 (0.30) * 0.01 (0.31)
Frozen Dinner 0.14 (0.08) 0.15 (0.08) -0.50 (0.09) * -0.53 (0.10) *
Household Cleaner -0.15 (0.07) * -0.48 (0.07) * -0.52 (0.14) * -0.88 (0.14) *
Hot Dog -0.62 (0.07) * -0.49 (0.07) * -0.46 (0.18) * -0.18 (0.18)
Laundry Detergent -0.23 (0.07) * -0.46 (0.07) * -1.25 (0.26) * -0.45 (0.27)
Margarine/Butter -0.45 (0.07) * -0.29 (0.07) * -0.31 (0.24) -0.66 (0.24) *
Mayonnaise -0.59 (0.07) * -0.71 (0.07) * 0.10 (0.24) -0.17 (0.24)
Milk -0.26 (0.07) * 0.35 (0.07) * -0.56 (0.25) * -0.38 (0.25)
Mustard/Ketchup -0.56 (0.07) * -0.35 (0.07) * -0.02 (0.19) -0.86 (0.20) *
Toothpaste -0.25 (0.11) * 0.47 (0.10) * -0.54 (0.15) * -0.40 (0.15) *
Peanut Butter -0.25 (0.07) * -0.31 (0.07) * -0.21 (0.48) -0.06 (0.46)
Photographs -0.53 (0.07) * -0.27 (0.08) * -0.11 (0.22) -0.61 (0.23) *
Razors -0.49 (0.10) * -0.17 (0.10) -0.15 (0.13) -0.62 (0.14) *
Salty Snacks 0.08 (0.07) 0.00 (0.07) -0.06 (0.07) -0.64 (0.08) *
Sauces 0.14 (0.07) * -0.39 (0.07) * -0.78 (0.08) * -0.59 (0.08) *
Soda -0.37 (0.07) * -1.03 (0.07) * -0.31 (0.09) * -0.52 (0.10) *
Soup -0.30 (0.07) * -0.15 (0.07) * 0.02 (0.13) -0.12 (0.13)
Sugar Substitute -0.23 (0.07) * -0.43 (0.07) * -0.54 (0.32) -0.03 (0.32)
Tissue Paper -0.21 (0.07) * -0.66 (0.07) * -0.31 (0.29) -0.19 (0.28)
Toilet Paper -0.16 (0.07) * -0.59 (0.07) * -0.57 (0.09) * -0.33 (0.10) *
Toothbrush -0.33 (0.07) * -0.50 (0.07) * -0.73 (0.08) * -0.23 (0.09) *
Yogurt -0.12 (0.07) -0.11 (0.07) -0.24 (0.11) * -0.53 (0.12) *
Control variables
Fixed effectsCategory * Chain * State * Year Brand * State * Year
Demographics and
marketing mix
Promotion, Feature, Display, Income, Elderly, Unemployment, Education, HH
Size, African American, Metro, ACV (Store size)
Total observations 176,357 150,068
13
Figure 1: Marginal impact is the percentage change in market share with one standard
deviation increase in Conservativeness (Religiosity and Republican Party vote).
-10% -5% 0% 5%
Blades
Coffee
Deodorant
Diapers
Frozen Pizza
Frozen Dinner
Household Cleaner
Hot Dog
Laundry Detergent
Margarine/Butter
Mayonnaise
Milk
Mustard/Ketchup
Toothpaste
Peanut Butter
Photographs
Razors
Salty Snacks
Sauces
Soda
Soup
Sugar Substitute
Tissue Paper
Toilet Paper
Toothbrush
Yogurt
Marginal impact on Generic Brand Share
-12% -7% -2% 3%
Blades
Coffee
Deodorant
Diapers
Frozen Pizza
Frozen Dinner
Household Cleaner
Hot Dog
Laundry Detergent
Margarine/Butter
Mayonnaise
Milk
Mustard/Ketchup
Toothpaste
Peanut Butter
Photographs
Razors
Salty Snacks
Sauces
Soda
Soup
Sugar Substitute
Tissue Paper
Toilet Paper
Toothbrush
Yogurt
Marginal impact on New Product Share