Content uploaded by Dara O'Rourke
Author content
All content in this area was uploaded by Dara O'Rourke on Oct 31, 2017
Content may be subject to copyright.
RESEARCH AND ANALYSIS
The Impact of Sustainability Information
on Consumer Decision Making
Dara O’Rourke and Abraham Ringer
Summary
This article presents an empirical analysis of the impact of sustainability information on
consumer purchase intentions and how this influence varies by issue (health, environment,
and social responsibility), product category, type of consumer, and type of information. We
assess over 40,000 online purchase interactions on the website GoodGuide.com and find
a significant impact of certain types of sustainability information on purchase intentions,
varying across different types of consumers, issues, and product categories. Health ratings
in particular showed the strongest effects. Direct users—those who intentionally sought
out sustainability information—were most strongly influenced by sustainability information,
with an average purchase intention rate increase of 1.15 percentage points for each point
increase in overall product score, reported on a zero to ten scale. However, sustainability
information had, on average, no impact on nondirect users, demonstrating that simply
providing more or better information on sustainability issues will likely have limited impact
on changing mainstream consumer behavior unless it is designed to connect into existing
decision-making processes.
Keywor ds :
attitudes-behavior gap
behavior change
consumers
decision making
industrial ecology
sustainability information
Introduction
In the United States, almost 70% of gross domestic product
(GDP) is driven, in some way, by “personal consumption ex-
penditures” (Emmons 2012). Markets for food, housing, apparel,
transportation, electronics, and even cosmetics drive major as-
pects of the economy. The specifics of these purchases—the
kinds of cars we drive, the food we eat, and the type of housing
we live in—are central to the state of our economy, as well as
to some of our largest social, environmental, and human health
challenges.
The industrial ecology (IE) community has recognized the
critical role of consumer decisions in sustainability issues for
many years (Hertwich 2005; Jackson 2005a; Tukker et al. 2006a,
2006b, 2010). IE research has been extremely important in iden-
tifying the most significant consumer impact areas: mobility,
food, housing, and energy-using products (Tukker et al. 2010).
However, most IE research has remained focused on technical
Address correspondence to: Dara O’Rourke, Department of Environmental Science, Policy, and Management, University of California, Berkeley, 130 Mulford Hall, Berkeley,
CA 94720, USA. Email: orourke@berkeley.edu Web: http://nature.berkeley.edu/orourke
© 2015 by Yale University
DOI: 10.1111/jiec.12310 Editor managing review: Charles Corbett
Volume 00, Number 0
strategies to change products and production systems through
the advancement of green chemistry, green product design, de-
materialization, process efficiency improvements, supply-chain
footprinting, life cycle thinking, end-of-life management, and
so on. This has been critical work and has served to undergird
recent initiatives related to sustainable development and the
“circular economy.”
However, much less research has been conducted on how
this work might motivate consumers to change behaviors to-
ward more sustainable products and systems, or even to un-
derstand what types of interventions (information, incentives,
mandates, and “nudges”) are most effective in influencing con-
sumer decisions. As Tukker and colleagues (2006a) pointed
out in the Oslo Declaration on Sustainable Consumption,
“simply providing information to consumers does not lead to
marked changes in behavior” (p. 12). And yet, information
provision—through reports, articles, databases, public service
www.wileyonlinelibrary.com/journal/jie Journal of Industrial Ecology 1
RESEARCH AND ANALYSIS
announcements, education campaigns, and so on—remains a
primary strategy of governments, international agencies, aca-
demics, and nongovernmental organizations (NGOs).
Almost a decade after the Oslo Declaration, significant ques-
tions remain around how to communicate sustainability infor-
mation to the public in a way that will motivate a shift toward
“greener” or reduced consumption. Though the IE community
has continued to refine the science and presentation of life cycle
assessments (LCAs), eco-footprints, eco-labels, and so on, we
still know very little about which of these metrics and indica-
tors, in which forms, delivered at what point in the consumer
decision-making process, have the most impact.
As Tukker and colleagues (2010, 20) have noted, there also
remains “considerable ambiguity in the contemporary under-
standing of how consumers actually choose products and ser-
vices.” Further, research over the last 10 years has identified an
important gap between what consumers say they care about and
what they actually do when faced with trade-offs around sustain-
ability. Surveys have shown that 30% to 70% of consumers say
they want to buy greener, healthier, more socially responsible
products, but only 1% to 5% actually do (Mintel 2011; Packaged
Facts 2011; Devinney et al. 2010). This gap between peoples’
stated preferences and actual purchases not only presents an
interesting empirical puzzle for consumer decision making, but
it is also represents an important impediment to behavioral
change that can support solutions to sustainability problems.
A growing body of research examines this “attitudes-
behaviors gap” (Kollmuss and Agyeman 2002; Staats 2003;
Jackson 2005b; Carrington et al. 2012; Young et al. 2010) by
analyzing why consumers make the decisions they do and, more
specifically, how values, attitudes, and knowledge regarding
sustainability issues influence purchase behavior. Recent
research (Carrington et al. 2012; Peattie 2010; Bray et al. 2011;
Prothero et al. 2011) has identified a range of contributing
factors that intervene between sustainability preferences and
purchasing decisions. These include: a lack of credible infor-
mation on environmental and social performance of products
(US EPA 1998; Ipsos Public Affairs 2010); consumers’ “willful
ignorance” about product ethicality (Ehrich and Irwin 2005);
negative beliefs about the quality and performance of “greener”
products (Luchs et al. 2010; Chang 2011); brand loyalties that
inhibit change (Seyfang 2009); skepticism about corporate
claims regarding product and supply-chain sustainability
(Gibbs and Soell 2010); and skepticism about the impact that
an individual consumer can make (Hanss and B¨
ohm 2010).
There has also been important recent research on the limi-
tations of focusing on individuals as the unit of change around
consumption. The emergence of “practice theory” (Brand 2010;
Shove and Walker 2010; Spaargaren 2011) argues for the need
to analyze the contexts and “systems of provision” within which
consumer choices are made. Choices we make about how to
commute, travel, and feed our families are often constrained by
existing cultural and physical systems.
Although all of these dynamics may impede consumers act-
ing on their values, we focus in this article on the foundational
role of information in consumer decision making. People first
need to know, and believe, that there are differing impacts from
their decisions and they need to be able to access meaningful
information that distinguishes one choice from another. Provid-
ing sustainability-related information is often the first strategy
deployed to advance more sustainable lifestyles and consump-
tion patterns. However, it is not clear what role this information
plays in actually changing behaviors.
There has, of course, been significant research conducted on
consumer decision making that spans the fields of psychology,
marketing, and behavioral economics. Among the theoretical
foundations that motivate this research is the long-standing
exploration of individual decision making, particularly in the
context of consumer behavior. From Schwartz’s (1977) “Norm-
Activation Theory” to Ajzen’s (1991) “Theory of Planned
Behavior” to Stern’s more recent (1999) “Attitude-Behavior-
Context” model, psychologists have sought to explain the
processes that mediate between people’s values and behaviors.
This research has challenged “rational choice” theories in
economics that model consumers as rational actors optimizing
their utility through calculated trade-offs in price, quality, and
so on. This rich field of research has shown that even the
most straightforward acts of consumption can be complicated,
conflicted, and appear irrational. People’s behavior is not
motivated solely by utility maximization (Kahneman 2011;
Carlsson and Johansson-Stenman 2012); people sometimes
make “automatic” and habitual decisions (Duhigg 2012), act
within a social context, and are motivated by social approval
and status (Jackson 2005b).
Bettman and colleagues (1998) expand this research to de-
velop what they term “constructive” choice theory. Under-
standing how consumers construct a decision (often in real
time) takes into account the goals of the decision maker, the
complexity of the decision task, the context and framing within
which the decision is made, the need for accuracy in the deci-
sion, the availability, completeness, and format of information
presented, and the emotional and cognitive costs of making the
decision. A key question that emerges out of this research—
for both academics and business—is how and when consumers
change their decisions and, further, whether information can
lead to changed behavior (rather than just changed attitudes).
Recent research (Janssen and Hamm 2012; Hainmueller et al.
2015) evaluates the impacts of information systems such as
eco-labels and fair trade certifications on consumer behavior.
This builds on questions around how consumers interpret “cre-
dence qualities” of products—such as environmental and labor
impacts of supply chains—that are not immediately discern-
able. These types of product and company attributes—which
are central to questions of sustainability and IE—require an in-
dependent source of information (such as from a government
agency or NGO) to attest that a product or company meets
certain environmental performance characteristics (Thompson
et al. 2010).
However, as Thøgersen and colleagues (2010) argue, despite
a growing number of eco-labels, there is still very little research
analyzing consumer decision-making processes in response to
eco-labels. Early research in this area was often based on
2Journal of Industrial Ecology
RESEARCH AND ANALYSIS
surveys centered around two basic questions: Would a person
buy environmentally certified products, and would they pay
more (Aguilar and Cai 2010)? More recent research involves
“real choice” experiments and analyses of sales data that study
purchases of products labeled, for example, with carbon rat-
ings (Vanclay et al. 2011). There has also been considerable
research on labeling of food products, as well as specific reasons
why people are buying more certified organic products. This re-
search assesses “altruistic” concerns for the environmental im-
pact of food production, as well as more “selfish” concerns about
personal health impacts of products (Michaud et al. 2013).
Consumer decision making is clearly variable and multi-
faceted. Adding to this complexity are the major changes un-
derway in where and how consumers access information, what
sources they trust, and what they do with information once
they have it. Given the complexities of shopping processes and
the importance of context, it is desirable to move out of the
controlled setting of the laboratory to study consumers in situ
as they experience real trade-offs in purchasing decisions.
Our research seeks to build on this past research on consumer
decision making and the impacts of information systems (such
as eco-labels) to assess how sustainability information might
influence purchase decisions.
Methodology
This research was designed to examine the impact of sus-
tainability information—environmental, social, and health rat-
ings of products and companies—on consumer purchases while
they shop online. Specifically, this study analyzed the impact of
sustainability ratings reported by the GoodGuide.com website.
GoodGuide’s staff, comprised of chemists, toxicologists, nutri-
tionists, sociologists, LCA experts, and computer scientists, has
rated more than 200,000 food, personal care, and household
chemical products, as well as apparel, appliances, automobiles,
and electronic devices. The website emphasizes both the expert-
driven and third-party nature of the scoring process and pro-
vides an explanation of its methodologies, in order to maximize
trust and legitimacy with users. The GoodGuide sustainability
ratings combine product- and company-level metrics to assess:
company transparency; environmental impacts; water, energy,
and material use; environmental management practices; regu-
latory violations; presence of chemical hazards within products;
as well as broad social and health impacts.1
All of this information, collected from more than 1,000
data sources, has been compiled into three scores, each ranging
from zero to ten, addressing health, environment, and social
impacts, respectively. The “health” rating has been modeled
on human health hazard assessment, chemical risk assessments,
and nutritional evaluations; the “environment” rating has been
modeled on a simplified LCA of the product category; and the
“society” rating has been modeled on corporate social responsi-
bility reporting (along the lines of the Global Reporting Initia-
tive). These subscores, along with an aggregated overall score,
are reported on each product page, along with opportunities
for users to explore the elements of subscore components, to
compare similar products and buy products by clicking a “Buy
Now” button, which brings them to one of several e-commerce
sites selling the product. The GoodGuide product pages also
display leading eco-certifications (such as USDA Organic, Fair
Trade, Energy Star, Design for Environment, Cradle to Cra-
dle, Leaping Bunny, and so on). In this regard, the GoodGuide
information is much more comprehensive and detailed than a
traditional eco-label.
GoodGuide incorporates and moves beyond eco-labels, pre-
senting a multiattribute decision support tool that employs life
cycle thinking to identify “hotspots” at both the product and
company level and then translates data for these criteria into
a form that consumers can understand. GoodGuide works to
present information on what matters most in a product cate-
gory, as well as how different products and companies compare
along those dimensions. GoodGuide also moves beyond one
of the key limitations of eco-labels—that they generally only
cover the top 1% to 5% of products in a category—by working
to rate the top 80% of products sold in each category, as well as
the leading green brands.
Figure 1 shows an example of a GoodGuide product page
with the breakdown of health, environmental, and social ratings
for one product, along with a Buy Now button in the lower left,
linking users to the product on one of several e-commerce sites.
The goal of our research was to study the impact of these
types of quantitative sustainability ratings on consumer pur-
chase intentions while shopping. We studied 12 months of
consumers who visited the website GoodGuide.com in what
is essentially a “field observation.” We recorded data from all
product pages on the website, 41,398 of which were viewed dur-
ing the study period, measuring the number of times each page
was viewed and the number of purchase events that occurred on
each, analyzing these with respect to the sustainability ratings
of each product (see table 1). We also examined the impact of
product rankings and found very similar results. In none of the
research presented here did we vary any conditions or subject
any subgroups to control or varying treatments.
However, the analysis for this article moves beyond past re-
search that has depended on surveys and focus groups to assess
how much consumers care about sustainability issues. As Auger
and Devinney (2007) have shown, research utilizing “uncon-
strained survey responses” regarding concerns about an issue
are often plagued by “social desirability biases” and do not cor-
relate well with what people actually do while operating in
contexts where information is limited, trade-offs exist in price,
quality, and so on, and no interviewer is judging them on their
responses.
All data were collected retrospectively for the 12-month
period from 1 November 2011 to 31 October 2012. All product
ratings remained static throughout this time period.
For the main analysis, two data sets were combined. We
first collected the number of purchase intention (PI) events
that occurred on each product page over the study period (see
table 1). Web analytics software was programmed to record a
purchase intention event each time a user clicked the button
labeled Buy Now and marked with a product’s current price,
sending them to the corresponding product page on an
e-commerce site. The second data set measured the number
O’Rourke and Ringer, Impact of Sustainability Information on Consumers 3
RESEARCH AND ANALYSIS
Figure 1 Example of GoodGuide product rating page.
4Journal of Industrial Ecology
RESEARCH AND ANALYSIS
Ta b l e 1 Visitors and purchase intention events
Metric Value
Unique visitor sessions 269,287
Total purchase intention events 44,605
Unique product pages viewed 41,398
of times each product page had been viewed over the study
period (see table 1). For each data set, user sessions of 5 sec-
onds or less were excluded because these contained high rates
of “bounces”—users leaving the site before engaging in any
interaction.
These two data sets were then combined, and for each prod-
uct, the purchase intention rate (PIR) was determined by cal-
culating the ratio of Buy Now events to page views. Product
pages were excluded from the analysis if missing subscores or an
e-commerce site product match had caused the presentation of
sustainability information or Buy Now button in a way incon-
sistent with other product pages.2This filtering yielded 41,398
product observations for analysis (see table 1).
The relationship between PIR and sustainability score was
examined using increasingly narrow sets of product pages, mea-
suring the response to both overall score and the three subscores.
We first examined all pages together, then broke these down
by broad product category (those for which there was adequate
data included personal care products, household chemicals, sta-
ple foods, convenience foods, and pet food). Finally, we broke
apart these broad categories into narrower categories in order
to explore potential differences in the effect of scores across
products that were broadly similar, but distinct, in certain char-
acteristics. Additional tests were conducted in order to explore,
first, potential nonlinearities in effect along the range of scores,
and second, whether there was a response to changes in color
distinct from that of numerical score.
To begin, users were stratified into two broad categories based
on how they had arrived at the website: direct and nondirect
users. Direct users comprised those who arrived having directly
entered the URL GoodGuide.com, made use of a browser book-
mark, or by a search engine having queried either: “GoodGuide”
or some close variant; a product search including the words
“healthiest,” “safest,” “greenest,” and so on; or a product search
including an issue of concern such as “parabens,” “sweatshop-
free,” or “pollution.” All others were considered nondirect
users, including those who arrived by a search engine query
for a generic topic or weblink without intentionally seeking to
visit GoodGuide.com. The majority of nondirect users came
to GoodGuide.com by search engine, having queried a phrase
such as “best mouthwash” or “top baby shampoo,” given that
GoodGuide is often on the first page of search engine results for
these types of terms. We assume that these shoppers were inter-
ested to purchase a top-performing product in these categories.
An analysis of PIRs between the two groups indicates signif-
icant differences in their overall purchase intention rates (see
table 2).3
A simple linear model was fit in order to analyze the overall
purchase rate for GoodGuide users:
Ta b l e 2 Purchase intention rates (PIR)
PIR (weighted) n
Direct users 3.8 21,712
Nondirect users 5.5 38,570
z=−9.2876
Ta b l e 3 Correlations between subscores
Pair-wise correlations HealthScore EnvScore
EnvScore 0.034
SocialScore −0.006 0.853
VIF values
HealthScore 1.08
EnvScore 2.52
SocialScore 2.52
Note:VIF=variance inflation factors.
Model1 : Purchase Intention Rate =β0
+βOVERALL(Overall Score)+ε
where β0is the intercept, that is, the expected PIR when overall
score =0andβ1is the coefficient of overall score, the expected
increase in PIR for each point increase in overall score. εis the
residual unexplained by the model. A multiple linear regression
model was also employed, in order to estimate the impact of the
subscores on purchase rates:
Model2 : Purchase Intention Rate =γ0
+γHEALTH(Health Score)+γENV(Env Score)
+γSOCIAL(Social Score)+ε
A Breusch-Pagan test for heteroskedasticity indicated that,
as expected, the two-stage sampling design introduced con-
siderable nonconstant variance among PIR estimates. That is,
data points based on fewer page views exhibited systematically
higher variance (χ2=20.42). Thus, analytical weighting was
employed throughout, according to the inverse of page view
values (Gould 1994; Baum 2006).
A test of pair-wise correlations among the three subscore
variables indicated a high level of correlation between the en-
vironmental score and social score, which leads us to be cautious
of the reported effects of these variables. However, the variance
inflation factors (VIF) did not indicate serious multicolinearity
(Chatterjee and Price 1991), so we proceeded with the above
model (see table 3).
Results
We analyzed the data for four primary questions:
1. Does sustainability information have an influence on pur-
chase intent? If so, how much?
O’Rourke and Ringer, Impact of Sustainability Information on Consumers 5
RESEARCH AND ANALYSIS
Ta b l e 4 Effect of overall scores: direct vs. nondirect users
βOVERALL r2
Direct users (n =21,712) 1.2*** 0.0494
Nondirect users (n =38,570) −0.12*** 0.0003
Note: Asterisks indicate p values <.005 (***).
Ta b l e 5 Effects of environment, health, and social subscores
User type Constant γHEALTH γENV γSOCIAL n Adj r2
Direct
users
−2.51 0.65*** 0.06 0.22*** 17587 0.0653
Nondirect
users
6.46 −0.11*** −0.30*** 0.32*** 29340 0.0013
Note: Asterisks indicate p values <.005 (***).
2. Does this vary for different types of consumers (direct vs.
nondirect users)?
3. Does this vary for different types of information (health,
environment, and social ratings)?
4. Does this vary for different types of products?
In the aggregate, the data suggest that GoodGuide sustain-
ability scores did influence users’ purchasing behavior. Analyz-
ing the two main user types separately, we see that this appears
to be driven almost entirely by direct users (table 4).
Among direct users, each one-point increase in a product’s
overall score on a zero to ten scale is, on average, associated
with a PIR increase of 1.2 PI events per 100 page views. Among
nondirect users, there is essentially no response, particularly in
consideration of the exceedingly low r2. These results suggest
real differences between direct and nondirect users in the degree
to which sustainability information influences their purchase
decisions.
We next explored whether there were differences in effect of
the three subscores among the different types of users. As seen
in table 5, among direct users, both health and social subscores
are associated with increases in PIR, whereas environment
subscores show no significant association. Among nondirect
users, health and environment scores show significant negative
relationships with PIR, whereas social scores show a significant
positive relationship.
It should also be noted that the r2for direct users, though
still low, are an order of magnitude higher than nondirect users.
Breaking down the analysis further, we hypothesize that users
have different responses according to product types. Table 6
shows results for direct and nondirect users, stratified across
product categories. As can be seen in table 6, among direct users,
health scores had the strongest association with increased PIR,
particularly among personal care products and pet food. En-
vironmental scores had positive association only for pet food,
whereas social scores showed a positive association only for
personal care. Social scores for pet food showed negative as-
sociation. Among nondirect users, results were more mixed.
Positive associations were found on household chemicals and
pet food health scores and personal care social scores, whereas
negative associations were found in a host of other categories,
with very low r2values throughout.
As we break down the broad categories into narrower cat-
egories, we see that the influence of scores appears to be quite
variable, even among related products. Table 7 shows, for in-
stance, that among direct users, the influence of health scores
in sun care, skin care, and baby care is considerably stronger
than in deodorant and makeup.
Ta b l e 6 Multiple linear regression: direct and nondirect users, by broad product category
Product category User type constant γHEALTH γENV γSOCIAL n Adj r2
Personal care Direct −2.72 0.77*** −0.02 0.27*** 13,122 0.078
Nondirect 6.96 −0.08*** −0.53*** 0.49*** 2,1640 0.002
Household chemicals Direct −3.73 0.39*** 0.30 0.28 978 0.096
Nondirect 4.27 0.35*** −0.26 0.03 1,304 0.007
Food: staples Direct −1.89 0.33*** −0.04 0.38 1,478 0.018
Nondirect 7.38 −0.18** 0.09 −0.39 2,904 0.002
Food: convenience Direct 0.92 0.31*** −0.36 0.28 1,284 0.008
Nondirect 15.89 −0.38*** −1.22*** 0.23 2,642 0.019
Pet food Direct −5.78 0.72*** 1.24*** −0.98*** 725 0.058
Nondirect −3.12 0.50*** −0.09 0.34 850 0.031
Note: Asterisks indicate p values <.01 (**) and <.005 (***).
6Journal of Industrial Ecology
RESEARCH AND ANALYSIS
Ta b l e 7 Direct users: personal care products
Product category Constant γHEALTH γENV γSOCIAL n Adj r2
Baby care −5.27 0.86*** −1.42*** 2.12*** 374 0.188
Bath shower and soap −6.64 1.05*** 0.95*** −0.41 1,001 0.112
Feminine hygiene 9.64 0.46 −0.75 −0.54 104 0.013
Hair care −3.58 1.03*** 0.49*** −0.39* 2,512 0.0123
Medicine cabinet −6.36 0.86*** −2.62*** 4.04*** 325 0.113
Men’s grooming −5.13 0.57*** −0.40 1.68*** 242 0.136
Deodorants and antiperspirants 0.34 0.15* 0.75*** −0.57 545 0.020
Eye and ear care 1.20 0.84*** −0.78 0.61 191 0.134
Foot and nail care 7.19 0.52*** −0.41 −0.94 410 0.020
Fragrance and perfumes −4.49 0.84*** −0.56 1.26 226 0.085
Makeup −0.12 0.37*** 0.32* −0.32 3,156 0.016
Skin care −0.93 0.87*** 0.16 −0.41* 3,127 0.080
Sun care −6.23 1.02*** −0.97* 1.81*** 479 0.222
Oral care 2.08 0.74*** 0.79 −1.57*** 430 0.085
Note: Asterisks indicate p values <.05 (*), <.01 (**), and <.005 (***).
Ta b l e 8 Direct users: household chemicals
Product
category
Constant γHEALTH γENV γSOCIAL n Adj r2
Insect
repellents
−1.85 1.19*** 1.50 −1.92 56 0.047
Air
fresheners
−0.71 1.01*** 0.16 −0.59 144 0.110
Dishwashing 0.02 0.15* 1.35*** −1.31 139 0.194
Household
cleaners
−6.22 0.40*** 0.10 0.86* 367 0.077
Laundry −3.78 0.58*** 0.17 0.25 266 0.132
Note: Asterisks indicate p values <.05 (*), <.01 (**), and <.005 (***).
In table 8, we see similar variability among subcategories of
household chemicals, such as the difference in health score ef-
fect between insect repellents and dishwashing products. Note
as well the difference in the association demonstrated between
health score and environmental score among dishwashing
products.
Several further tests were undertaken to explore the dynam-
ics within these results. Following indications of nonparamet-
ric behavior evident using kernel-weighted local polynomial
smoothing (Fan and Gijbels 1996), we conducted spline analy-
sis (Greene 2012) of the influence within ranges of scores, with
nodes at 3.0, 6.0, and 8.0 reflecting broad ranges of scores as
presented to website users.4These results, shown in table 9, in-
dicate that the overall associations between score and PIR are
driven predominantly by the effects of high-end scores. That
is, the difference between a poor rating and a very poor rating
matters little, whereas differences at the high end are strongly
associated with PIR differences, at least among health scores.
We next set out to test whether there was any effect produced
by the colors assigned to different scores. That is, behind each
score on a product page, users see a color: red (for scores less
than 3.1); orange (for scores higher than 3.1, but less than 6.1);
Ta b l e 9 Multivariate spline analysis: direct users all products
(n =17,587)
Subscore Spline Coefficient
Health 0–3 0.15
3–6 −0.05
6–8 1.03***
8–10 1.22***
Environment 0–3 −4.09
3–6 0.75***
6–8 0.37***
8–10 −2.38***
Social 0–3 5.06
3–6 −0.94***
6–8 1.91***
8–10 −4.49***
Note: Asterisks indicate p values <.005 (***).
light green (higher than 6.1, but lower than 8.1); or dark green
(higher than 8.1). Using regression discontinuity analysis, we
tested whether discontinuities occurred at the color changes of
3.1, 6.1, and 8.1, indicating an effect separate from that of the
numerical scores.
We were interested in this question because there have been
a number of studies recently showing the impact of “traffic
light” labeling systems that show consumers red, yellow, and
green symbols on products (Hallstein and Villas-Boas 2013).
However, we did not observe evidence of discontinuities at any
of the color changes. All Wald statistics were nonsignificant.
Discussion
This research represents a preliminary examination of the
impact of sustainability information on consumer behavior
in the natural conditions of a live website, rather than an
O’Rourke and Ringer, Impact of Sustainability Information on Consumers 7
RESEARCH AND ANALYSIS
experimental or survey setting. From these data, we draw a
number of conclusions.
First, it is clear that many consumers are unaffected by sus-
tainability information, and for some, a product deemed more
“sustainable” may actually decrease purchase intent. On the
other hand, consumers who have expressed previous commit-
ment to sustainability issues appear to make use of this infor-
mation as part of their purchasing process. Among these direct
users, we found a statistically significant relationship between
sustainability information on purchase intent, with more pos-
itive information associated with greater purchase intent. In
examining subscores, we show that health ratings, those most
related to self-interest, are the predominant source of this effect.
Interestingly, we observe evidence that sustainability mea-
sures can resonate differently across different product types,
even within broad categories. Focusing on direct users and
health scores, for instance, sun care, where there is an intrin-
sic connection to health, shows an increase of 1.02 PI events
per 100 visitors for each point increase in health score and an
r2of 0.22. Deodorants, on the other hand, represent a product
type where consumers are particularly concerned about efficacy,
and here we detected a much weaker impact of health scores
(an increase of 0.15 PI events per 100 visitors for each point
increase in health score). Many product types fall somewhere
in between, with a weighted mean increase of 0.77 PI events
per 100 visits for each point increase in health score across
all personal care products. Health concerns also weigh heavily
in consumer considerations of household chemicals. However,
environmental concerns appear as well. In dishwashing soaps,
for instance, there is a relatively strong effect of environmental
ratings. These data suggest a number of hypotheses that might
be explored in further research.
Some results are surprising. For example, among several
product types, environment scores are negatively associated
with PI among direct users, even where health ratings show a
strong positive impact. That is, as environment scores increase,
PI decreases. This might be ascribed to the altruistic nature
of environment scores compared with the more self-interested
health scores. Among nondirect users, both health and envi-
ronment scores are, in some cases, associated with decreased PI,
as with personal care products and foods. This may indicate an
existing bias against “green” products.
As Chang (2011) points out, research has shown “ambiva-
lent” attitudes from consumers toward green products. Negative
perceptions are based around concerns that green products are
more expensive and lower quality than equivalent conventional
products. Consumers also express ambivalence around whether
switching to greener products will have any appreciable impact
on environmental problems.
With only a few exceptions, environmental and social scores
show no positive association with purchase behavior. The only
broad category where significant positive results were measured
for environmental ratings was in pet food among direct users
(which may reflect animal focused consumers equating environ-
mental impact with being “animal friendly”). The only positive
significant results measured for broad-category social ratings
were in personal care products. These issues, as compared with
health concerns, are, to a much greater degree, altruistic, and so
this is an interesting demonstration of the source of behavioral
motivations.
Finally, our statistical analyses show that consumer responses
to sustainability ratings are not linear across the range of scores.
Focusing on the health scores, where most of the significant
responses occur, a change in scores among low-scoring products,
that is, scores of 0 through 6, had very little impact on PI,
whereas above this point, consumers showed a responsiveness to
increasing scores. This is somewhat surprising, given that recent
research (Hallstein and Villas-Boas 2013) has shown a stronger
impact of “red lights” over “green lights” in consumer labeling.
We hypothesize that, at least on this platform, positive health
ratings help reinforce a purchase decision more than negative
information dissuades a consumer away from purchase.
These empirical findings contribute to several important on-
going debates. First, in considering efforts to motivate changes
in consumer behavior to advance sustainability, it is clear that
consumers think of different types of products very differently.
Information on the health and environmental impacts of these
varied products thus influences consumer decisions differently
as well. As Dhar and Wertenbroch (2000) and others have
shown, consumers consider “hedonic” products very differently
than “utilitarian” products. We would go further to posit that
decisions around products that “just have to work,” such as
deodorant or hair dye, are very resistant to influence by sus-
tainability information. Products that are more discretionary,
or where it is difficult for consumers to determine differences in
quality, are much better targets for informational campaigns to
shift behavior. Similarly, products that are closely tied to “sta-
tus” and that you consume publicly—even products displayed
on your kitchen counter, such as dish soap—may be more open
to influence from information campaigns. Finally, products that
have clear personal health impacts (vs. those that only have en-
vironmental or social impacts removed from the consumer) are
better targets for information influence.
It may equally be the case that complexity of sustainability
information poses a major challenge for the use of IE infor-
mation (such as life cycle, footprinting, and rating data) in
informing consumer decisions. As such, it is critical to consider
consumer perceptions and responses in the design of sustain-
ability indicators. We hypothesize that, particularly in the en-
vironmental arena, consumers have difficulties understanding
some measures of performance. Metrics for biodiversity loss, cli-
mate change, air quality, and so on, are difficult for GoodGuide
to communicate simply to consumers. Complex data need to be
translated into a format that works within “bounded” decision-
making contexts. Further, learning from behavioral psychology,
we believe there are important lessons from framing issues as
losses (which are much more motivating), anchoring, and as-
sisting consumers in comparing complex trade-offs.
It should be noted that this study is not without lim-
itations. First, this study only looked at online decision
making. While e-commerce is a rapidly growing segment of
consumption, it is still much smaller than “brick-and-mortar”
8Journal of Industrial Ecology
RESEARCH AND ANALYSIS
purchasing. Second, consideration of environmental, health,
and social performance—the underlying components of the
sustainability ratings we evaluated—account for only a small
portion of the overall decision calculus for consumers. The fact
that the regression goodness-of-fit measures (r2) are quite low,
generally below 0.1, indicate that, when making purchasing
decisions, consumers are focused primarily on other issues, such
as price and quality, variables we were not able to include in
this analysis.
Further, our outcome variable measured purchase intent,
rather than actual purchases. Whereas purchase intent is a be-
havior unto itself and valuable from a research perspective, we
do recognize that it may introduce bias. In particular, a user
who sees herself as sustainability minded may not express in-
tention to buy a low-rated product while on a website where
such issues are dominant, but may still purchase the item in a
store where such concerns become less immediate and a variety
of considerations come into play.
Implications and Future Research
Our research raises a number of questions for public pol-
icy and for strategies related to product transparency and con-
sumer information. First, it is clear that we need to know more
about how and under what conditions information can be de-
signed and delivered to influence consumers that are not already
seeking out information on the sustainability impacts of their
choices (what we call nondirect users). More research and ex-
perimentation is needed that focuses on designing information
with this public specifically in mind.
Along with efforts aimed at the provision of information,
policy makers might consider a two-stage strategy, with efforts
first directed at raising consumers’ awareness of issues, and then
presenting these same consumers with product-level sustain-
ability information. However, simply providing more (or bet-
ter) scientific information on sustainability and health measures
appears to have limited impact on changing these consumers’
behavior.
Further research is needed on designing information
systems—not just the information visualization, but also the
entire process of information delivery—to maximize its im-
pact (Froehlich et al. 2010; DiSalvo et al. 2010; Kalnikaite
et al. 2011). For instance, how can sustainability information
be connected into peoples’ actual purchasing processes? How
can information systems leverage the lessons of behavioral psy-
chology regarding the power of peer influence, status, and habit
(Fogg 2009; Michie et al. 2013)? Further, of particular import
for sustainability, how can information systems help overcome
existing biases against greener products? For the IE community,
it is critical to consider how the design and conduct of different
forms of analysis interact with information delivery systems and,
in particular, assess how information might be communicated
and utilized from the earliest stages of IE research.
This is particularly relevant in light of recent government
initiatives on “open data.” Although government efforts to open
information to the public are absolutely laudable, it is unlikely to
have much impact on consumer behavior unless it is connected
to the actual decision processes of consumers. This connects to
a broader challenge in IE of how best to communicate complex
sustainability information to different types of decision makers.
Additional research and experimentation is needed around pre-
senting rigorous science and LCA-related indicators in a form
that can be understood and resonate with average consumers.
Translation is required between LCA indicators, such as global
warming potential, eutrophication, ecotoxicity, and so on, and
how the public thinks about product and company impacts.
Regarding future research, we should note that though our
method of data collection, by website monitoring, has presented
opportunities previously unavailable for the observation of con-
sumer behavior during normal consumption processes, it also
introduced certain limitations. In particular, all data are retro-
spective and neither the website nor the analytics software were
designed ex ante with the conduct of these analyses in mind. As
such, certain data may be unmeasured or ambiguous, reflected
in our low goodness-of-fit measures.
Nonetheless, we believe that an extension of this work to
Web and mobile applications (apps) that connect to actual pur-
chase decisions may provide a fruitful platform for experiments
and learning (in place of notoriously inaccurate surveys and
focus groups). As Tukker and colleagues (2010) note, there is
still a need to study better ways to communicate sustainability
information and present “a careful balance of persuasive and dis-
suasive strategies” that might drive consumer change (Tukker
et al. 2010, 21).
There are a number of changes and additions we plan for fu-
ture analyses in order to improve the explanatory power of our
research and address more specific underlying dynamics of con-
sumer decision making. These will take two forms: (1) technical
changes that allow more reliable, statistically powerful hypoth-
esis testing and (2) design changes that allow for controlled
experiments.
We plan to study additional covariates in this research. For
instance, preconceived impressions of product quality and effi-
cacy may have a significant influence on consumers’ propensity
to purchase a product (Luchs et al. 2010). Clearly, there is a
need to control for price and quality in future research. A num-
ber of further experiments are also suggested by the growing
body of theory and empirical work on sustainable consumerism
(Hainmueller et al. 2015). We plan to draw on recent research
from behavioral psychology and information design to analyze
what influences consumers to change their purchases (i.e., to
switch between products) (Fogg 2009; Lee et al. 2011; Carlsson
and Johansson-Stenman 2012).
Although preliminary, we believe this research has laid an
important base for future experiments that study consumer be-
havior change underlying efforts to advance more sustainable
consumption.
Acknowledgments and Disclosure
The authors thank Daniel Schmidt, Ryan Aipperspach, and
Pedro Vieira from GoodGuide for assistance in accessing data
O’Rourke and Ringer, Impact of Sustainability Information on Consumers 9
RESEARCH AND ANALYSIS
for this analysis. The authors also thank Dr. Maureen Lahiff
for input on our statistical analysis, as well as the three review-
ers, the associate editor, and the editor-in-chief for excellent
recommendations to make the article stronger.
One of the authors was a co-founder of GoodGuide.
GoodGuide was acquired by Underwriters Laboratories (UL)
in 2012. Although one of the authors serves as a consultant to
UL, the authors have no financial interest in the outcome of
this research or any other conflicts related to this research.
Notes
1. GoodGuide’s rating system combines product- and company-level
information to characterize a product’s health, environmental, and
social impacts. A detailed description of the methodology is avail-
able at www.goodguide.com/about/methodologies.
2. For some products, no price was returned from Amazon.com, Price-
grabber.com, and so on. When there was no price displayed, the
button that normally said “Buy Now” changed to “Check prices.”
We feel this is a quite different action than buying.
3. PIR =Purchase Intention Events/Pageviews
4. www.goodguide.com/about/ratings.
References
Aguilar, F. X. and Z. Cai. 2010. Conjoint effect of environmental
labeling, disclosure of forest of origin and price on consumer pref-
erences for wood products in the US and UK. Ecological Economics
70(2): 308–316.
Ajzen, I. 1991. The theory of planned behavior. Organizational Behavior
and Human Decision Processes 50(2): 179–211.
Auger, P. and T. M. Devinney. 2007. Do what consumers say mat-
ter? The misalignment of preferences with unconstrained ethical
intentions. Journal of Business Ethics 76(4): 361–383.
Baum, C. 2006. An introduction to modern econometrics using Stata.Col-
lege Station, TX, USA: StataCorp.
Bettman, J. R., M. F. Luce, and J. W. Payne. 1998. Constructive con-
sumer choice processes. Journal of Consumer Research 25(3): 187–
217.
Brand, K. W. 2010. Social practices and sustainable consumption: Ben-
efits and limitations of a new theoretical approach (pp. 217–235).
Dordrecht, the Netherlands: Springer.
Bray, J., N. Johns, and D. Kilburn. 2011. An exploratory study into the
factors impeding ethical consumption. Journal of Business Ethics
98(4): 597–608.
Carlsson, F. and O. Johansson-Stenman. 2012. Behavioral economics
and environmental policy. Annual Review of Resource Economics
4: 75–99.
Carrington, M. J., B. A. Neville, and G. J. Whitwell. 2012. Lost in
translation: Exploring the ethical consumer intention-behavior
gap. Journal of Business Research 67(1): 2759–2767.
Chang, C. 2011. Feeling ambivalent about going green. Journal of Ad-
vertising 40(4): 19–32.
Chatterjee, S. and B. Price. 1991. Regression analysis by example,2nd
ed. New York: Wiley.
Devinney, T., P. Auger, and G. Eckhardt. 2010. The myth of the ethical
consumer. Cambridge, UK: Cambridge University Press.
Dhar, R. and K. Wertenbroch. 2000. Consumer choice between he-
donic and utilitarian goods. Journal of Marketing Research 37(1):
60–71.
DiSalvo, C., P. Sengers, and H. Brynjarsd´
ottir. 2010. Mapping the
landscape of sustainable HCI. In Proceedings of the 28th Interna-
tional Conference on Human Factors in Computing Systems (pp.
1975–1984). New York: ACM.
Duhigg, C. 2012. The power of habit: Why we do what we do in life and
business. New York: Random House.
Ehrich, K. R. and J. R. Irwin. 2005. Willful ignorance in the request
for product attribute information. Journal of Marketing Research
42(3): 266–277.
Emmons, W. R. 2012. Don’t expect consumer spending to be the en-
gine of economic growth it once was. Federal Reserve Bank of St.
Louis. www.stlouisfed.org/publications/re/articles/?id=2201. Ac-
cessed 13 June 2013.
Fan, J. and I. Gijbels. 1996. Local polynomial modelling and its applications.
London: Chapman & Hall.
Fogg, B. J. 2009. A behavior model for persuasive design. In Proceedings
of the 4th International Conference on Persuasive Technology (p. 40).
New York: ACM.
Froehlich, J., L. Findlater, and J. Landay. 2010. The design of eco-
feedback technology. In Proceedings of the 28th International Con-
ference on Human Factors in Computing Systems (pp. 1999–2008).
New York ACM.
Gibbs & Soell. 2010. Sense & SustainabilityTM Study—Perspectives on
corporate sustainability among consumers and Fortune 1000 exec-
utives. www.gibbs-soell.com/wp-content/uploads/2010/08/2010-
Gibbs-Soell-Sense-Sustainability-Study.pdf. Accessed 6 June
2013.
Gould, W. 1994. Clarification on analytic weights with linear regres-
sion. Stata Technical Bulletin 20: 2–3.
Greene, W. H. 2012. Econometric analysis, 7th ed. Upper Saddle River,
NJ, USA: Prentice Hall.
Hainmueller, J., M. J. Hiscox, and S. Sequeira. 2015. Consumer de-
mand for the fair trade label: Evidence from a multi-store field
experiment. Review of Economics and Statistics. 97(2): 242–256.
Hallstein, E. and S. B. Villas-Boas. 2013. Can household consumers
save the wild fish? Lessons from a sustainable seafood advisory.
Journal of Environmental Economics and Management 66(1): 52–71.
Hanss, D. and G. B¨
ohm. 2010. Can I make a difference? The role of gen-
eral and domain-specific self-efficacy in sustainable consumption
decisions. Umweltpsychologie 14(2): 46–74.
Hertwich, E. 2005. Consumption and the rebound effect: An industrial
ecology perspective. Journal of Industrial Ecology 9(1–2): 85–98.
Ipsos Public Affairs. 2010. Business cleaning sustainability
study. www.ipsos-na.com/news-polls/pressrelease.aspx?id=4857.
Accessed 15 November 2013.
Jackson, T. 2005a. Live better by consuming less? Is there a “double
dividend” in sustainable consumption? Journal of Industrial Ecology
9(1–2): 19–36.
Jackson, T. 2005b. Motivating sustainable consumption. A review of
evidence on consumer behaviour and behavioural change. Are-
port to the Sustainable Development Research Network. Surrey, UK:
Center for Environmental Strategies.
Janssen, M. and U. Hamm. 2012. Product labelling in the market for
organic food: Consumer preferences and willingness-to-pay for
different organic certification logos. Food Quality and Preference
25(1): 9–22.
10 Journal of Industrial Ecology
RESEARCH AND ANALYSIS
Kahneman, D. 2011. Thinking, fast and slow. New York: Farrar, Straus
and Giroux.
Kalnikaite, V., Y. Rogers, J. Bird, N. Villar, K. Bachour, S. Payne, and
S. Kreitmayer. 2011. How to nudge in situ: Designing lambent
devices to deliver salient information in supermarkets. In Proceed-
ings of the 13th International Conference on Ubiquitous Computing
(pp. 11–20). New York: ACM.
Kollmuss, A. and J. Agyeman. 2002. Mind the gap: Why do people act
environmentally and what are the barriers to pro-environmental
behavior? Environmental Education Research 8(3): 239–260.
Lee, M. K., S. Kiesler, and J. Forlizzi. 2011. Mining behavioral eco-
nomics to design persuasive technology for healthy choices. In
Proceedings of the 2011 Annual Conference on Human Factors in
Computing Systems (pp. 325–334). New York: ACM.
Luchs, M. G., R. W. Naylor, J. R. Irwin, and R. Raghunathan. 2010.
The sustainability liability: Potential negative effects of ethicality
on product preference. Journal of Marketing 74(5): 18–31.
Michaud, C., D. Llerena, and I. Joly. 2013. Willingness to pay for
environmental attributes of non-food agricultural products: A
real choice experiment. European Review of Agricultural Economics
40(2): 313–329.
Michie, S., M. Richardson, M. Johnston, C. Abraham, J. Francis, W.
Hardeman, M. F. Eccles, J. Cane, and C. E. Wood. 2013. The be-
havior change technique taxonomy of 93 hierarchically clustered
techniques: Building an international consensus for the reporting
of behavior change interventions. Annals of Behavioral Medicine
46(1): 81–95.
Mintel. 2011. Green living—US. http://academic.mintel.com/. Ac-
cessed 6 June 2013.
Packaged Facts. 2011. Consumers and sustainability: Food & bev-
erage, personal care, household cleaners, and OTC medicines.
Rockville, MD, USA: Packaged Facts.
Peattie, K. 2010. Green consumption: Behavior and norms. Annual
Review of Environment and Resources 35: 195–228.
Prothero, A., S. Dobscha, J. Freund, W. E. Kilbourne, M. G. Luchs,
L. K. Ozanne, and J. Thøgersen. 2011. Sustainable consumption:
Opportunities for consumer research and public policy. Journal of
Public Policy & Marketing 30(1): 31–38.
Seyfang, G. 2009. The new economics of sustainable consumption: Seeds
of change. London: Palgrave.
Shove, E. and G. Walker. 2010. Governing transitions in the sustain-
ability of everyday life. Research Policy 39(4): 471–476.
Spaargaren, G. 2011. Theories of practices: Agency, technology, and
culture: Exploring the relevance of practice theories for the gov-
ernance of sustainable consumption practices in the new world-
order. Global Environmental Change 21(3): 813–822.
Staats, H. 2003. Understanding pro-environmental attitudes and be-
haviour: An analysis and review of research based on the theory
of planned behaviour. In Psychological theories for environmental
issues, edited by M. Bonnes et al. Burlington, VT, USA: Ashgate.
Stern, P. C. 1999. Information, incentives, and proenvironmental con-
sumer behavior. Journal of Consumer Policy 22(4): 461–478.
Schwartz, S. H. 1977. Normative influences on altruism. Advances in
Experimental Social Psychology 10: 221–279.
Thøgersen, J., P. Haugaard, and A. Olesen. 2010. Understanding con-
sumer responses to ecolabels. European Journal of Marketing 44(11–
12): 1787–1810.
Thompson, D. W., R. C. Anderson, E. N. Hansen, and L. R. Kahle.
2010. Green segmentation and environmental certification: In-
sights from forest products. Business Strategy and the Environment
19(5): 319–334.
Tukker, A., M. J. Cohen, U. Zoysa, E. Hertwich, P. Hofstetter, A. Inaba,
S. Lorek, and E. Stø. 2006a. The Oslo declaration on sustainable
consumption. Journal of Industrial Ecology 10(1–2): 9–14.
Tukker, A., G. Huppes, J. Guin´
ee, R. Heijungs, A. de Koning, L.
van Oers, S. Suh, et al. 2006b. Environmental impact of products
(EIPRO): Analysis of the life cycle environmental impacts related to
the final consumption of the EU-25. Seville, Spain: European Com-
mission, Joint Research Center, Institute for Prospective Tech-
nological Studies (IPTS).
Tukker, A., M. J. Cohen, K. Hubacek, and O. Mont. 2010. The impacts
of household consumption and options for change. Journal of
Industrial Ecology 14(1): 13–30.
US EPA (U.S. Environmental Protection Agency). 1998. Environ-
mental labeling issues, policies and practices worldwide. Publica-
tion prepared by Abt Associates and Gary Davis. US EPA contract
no. 68-W6-0021. Washington, DC: U.S. Environmental Protec-
tion Agency.
Vanclay, J. K., J. Shortiss, S. Aulsebrook, A. M. Gillespie, B. C. Howell,
R. Johanni, and J. Yates. 2011. Customer response to carbon
labelling of groceries. Journal of Consumer Policy 34(1): 153–160.
Young, W., K. Hwang, S. McDonald, and C. J. Oates. 2010. Sustain-
able consumption: Green consumer behaviour when purchasing
products. Sustainable Development 18(1): 20–31.
About the Authors
Dara O’Rourke is an associate professor in the Depart-
ment of Environmental Science, Policy, and Management at
the University of California–Berkeley in Berkeley, CA, USA.
Abraham Ringer is a doctoral student at the University of
California–Berkeley.
O’Rourke and Ringer, Impact of Sustainability Information on Consumers 11