Description
We investigate how accounting information and message framing jointly impact consumer
choice between brand name and generic drugs using a sample representative of the U.S.
adult population. We find that information about manufacturers’ profit margin/cost profile
in different frames predisposes consumers to develop more or less favorable attitudes
towards the firms and their products. The results suggest that two important accounting
constructs (profit margin and cost) are endowed with a descriptive valence that evokes
affective responses in consumers’ associative memory. This study adds to earlier work
on the role of accounting information’s connotative meaning in shaping user perceptions
by providing new evidence that the descriptive valence of accounting constructs can
impact consumer purchase decisions.
From bottom line to consumers’ mind: The framing effects
of accounting information
Yan Tian
a
, Hui Zhou
b,?
a
Department of Communication, University of Missouri-St. Louis, 1 University Blvd, St Louis, MO, United States
b
Melbourne Business School, The University of Melbourne, 200 Leicester Street, Carlton, VIC, Australia
a r t i c l e i n f o
Article history:
Available online 4 May 2015
a b s t r a c t
We investigate how accounting information and message framing jointly impact consumer
choice between brand name and generic drugs using a sample representative of the U.S.
adult population. We ?nd that information about manufacturers’ pro?t margin/cost pro?le
in different frames predisposes consumers to develop more or less favorable attitudes
towards the ?rms and their products. The results suggest that two important accounting
constructs (pro?t margin and cost) are endowed with a descriptive valence that evokes
affective responses in consumers’ associative memory. This study adds to earlier work
on the role of accounting information’s connotative meaning in shaping user perceptions
by providing new evidence that the descriptive valence of accounting constructs can
impact consumer purchase decisions. The ?ndings of this study also have direct implica-
tions for efforts aimed at reducing health care costs by promoting wider use of generic
drugs.
Ó 2015 Elsevier Ltd. All rights reserved.
Introduction
Accounting information affects not only stakeholders
who have ?nancial interests in the reporting company,
but also other stakeholder groups including regulatory
agencies, employees, and consumers. However, little
research has examined the effects of ?nancial reporting
on stakeholder groups other than investors (Kachelmeier,
Stephen, & Schadewald, 1991). In this study, we focus on
consumers and investigate how accounting information
and message framing jointly impact consumer judgment
and choice.
Accounting research has long established that the con-
notative meaning of accounting information plays a role
in shaping the perceptions and thoughts of those who are
exposed to the information (Flamholtz & Cook, 1978;
Haried, 1972, 1973; Houghton, 1987, 1988; Oliver, 1974).
These ?ndings can be better understood in light of the
development in psychological research that highlights the
importance of associative memory in the cognitive process.
Morewedge and Kahneman (2010) concluded that the pri-
mary mechanism underlying the well-documented prim-
ing and framing effects is the automatic operations of
associative memory. Associative memory is de?ned as ‘‘a
network of long-term memory for semantic information,
emotions and goals that is governed by the spread of acti-
vation, as determined by the strengths of interconnecting
weights (associations)’’ (Morewedge & Kahneman, 2010,
p. 435). The associative processes operate preconsciously
in the slow-learning memory system, quickly and auto-
matically generating intuitive and affective responses to
objects (Smith & Decoster, 2000; Uleman, Saribay, &
Gonzalez, 2008). Given the importance of semantic infor-
mation in associative memory, the ‘‘semantic halo effect’’
of accounting information documented in the aforemen-
tioned studies likely arises from the automatic activation
of compatible associations based on the way thehttp://dx.doi.org/10.1016/j.aos.2015.04.003
0361-3682/Ó 2015 Elsevier Ltd. All rights reserved.
?
Corresponding author. Tel.: +61 3 93498710.
E-mail address: [email protected] (H. Zhou).
Accounting, Organizations and Society 43 (2015) 56–66
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
accounting constructs are encoded in the slow-learning
memory system.
This study adds to earlier work on the role of accounting
constructs’ connotative meaning in shaping user percep-
tions by investigating the automatic affective reactions of
consumers to information about the manufacturers’ pro?t
margin/cost pro?le presented in different frames. We posit
that pro?t margin (cost) has a negative (positive) descrip-
tive valence, and thus information about the manufactur-
ers’ relative pro?t margins (costs) will evoke negative
(positive) affective responses. Drawing on direction of
comparison of asymmetries documented in previous
research (e.g., Bowdle & Medin, 2001; Roese, Sherman, &
Hur, 1998; Wanke, Schwarz, & Noelle-Neumann, 1995),
we further posit that comparison framing (i.e., which ?rm
is placed as the subject vs. referent in a comparison of two
?rms) can affect consumer attitudes, as consumers will
associate the affective responses with the subject ?rm
but not the referent ?rm. This implies that consumers will
be unfavorably predisposed towards the subject ?rm rela-
tive to the referent ?rm when a comparison of pro?t mar-
gins is presented. Conversely, consumers will be favorably
predisposed towards the subject ?rm relative to the refer-
ent ?rm when a comparison of costs is presented. The rel-
atively favorable (unfavorable) image of a ?rm, in turn, will
increase (decrease) consumer intention to choose that
?rm’s products over those of its competitor.
We use consumer choice between brand name and gen-
eric drugs as the setting to test our propositions.
Pharmaceuticals provide a suitable setting to investigate
our research question for two reasons. First, prescription
prices, as well as manufacturers’ pro?t margins and costs,
vary signi?cantly between brand name and generic drugs.
1
Bhat (2005) reports that the average prescription price of a
brandname drug is $65.29comparedto only $19.33for a gen-
eric drug in 2000, and net pro?t margin before taxes of ten
brand name pharmaceutical manufacturers was 23.6% com-
pared to only 17.2% for ten generic pharmaceutical manufac-
turers. Second, there is widespread belief that consumers
view the price difference between brand name and generic
drugs as a signal of difference in quality, despite the fact that
generic drugs have to meet the same rigid standards as brand
name drugs in the FDA approval process. For example, a 2008
New York Times editorial titled ‘‘Generic Drug Resistance’’
highlighted the dif?culty in persuading more patients to use
generic drugs, even in the presence of substantial evidence
con?rming the effectiveness and safety of generics. This is
consistent with the phenomenon of price reliance docu-
mentedinconsumer researchwhereprices are usedas quality
indicators, especially in purchase decisions of ‘‘credence
goods’’ such as pharmaceuticals (Zeithaml, 1988).
In addition to providing a setting for testing the joint
effect of accounting information and message framing on
consumers, consumer choice between generic and brand
name medications is an important policy issue. Policy
makers and insurers have long promoted wider use of gen-
eric drugs in place of their more expensive brand name
alternatives as an effective mechanism to contain prescrip-
tion drug spending. Using 1997–2000 Medical Expenditure
Panel data, Haas, Phillips, Gerstenberger, and Seger (2005)
estimated that broad generic substitution of outpatient
prescription drugs would result in a national savings of
$8.8 billion in the United States each year, approximately
11% of the annual drug expenditures in the United States
during 1997 to 2000. With the rapid increase in prescrip-
tion drug expenditures during the last decade, the absolute
savings from substituting generic drugs for brand name
drugs will be even more substantial going forward. The
magnitude of these ?gures underlies the efforts of policy
makers and insurers to increase the use of generic drugs.
A signi?cant number of such programs are speci?cally tar-
geted at improving consumer perceptions of generic drugs
because the perceptions of patients play an important role
in deciding which medications are prescribed to them
(Shrank et al., 2009).
We have designed an experiment to investigate whether
exposing consumers to information on manufacturers’
pro?t margin/cost pro?le can evoke affective responses that
affect consumer attitudes towards the ?rms and their prod-
ucts. The participants are a sample representative of the U.S.
adult population. To identify the affective responses to
pro?t margin (cost), we vary the comparison framing in
the statements that compare the pro?t margin (cost) of
brand name and generic drug manufacturers by manipulat-
ing the directionof comparisons. For simplicity in reference,
pro?t margin (cost) disclosures stating that brand name
drug manufacturers have higher pro?t margin (cost) than
generic drug manufacturers are referred to as the brand
name as subject condition. Conversely, pro?t margin (cost)
disclosures stating that generic drug manufacturers have
lower pro?t margin (cost) than brand name drug manufac-
turers are referred to as the generic as subject condition.
Consistent with our predictions, the results show that
disclosures about the manufacturers’ pro?t margin/cost
pro?le and the frame of disclosures have interactive effects
on consumer perceptions and purchase intentions.
Speci?cally, when information about relative pro?t margin
is presented, the brand name as subject frame is associated
with a more favorable image of generic drug manufacturers
than the generic as subject frame. In contrast, when informa-
tion about relative cost levels is presented, the generic as
subject frame is associated with a more favorable image of
generic drug manufacturers than the brand name as subject
frame. Finally, the image of generic drug manufacturers
mediates the effect of the disclosures on consumer inten-
tions to communicate about and purchase generic drugs.
This studycontributes toresearchthat examines the cog-
nitive mechanisms underlying the response to accounting
information and extends the literature on the role of
accounting information’s connotative meaning in shaping
user perceptions. We provide evidence that encountering
information about the manufacturers’ pro?t margin/cost
pro?le in different frames leads to selective attention and
cognitive search mechanisms that result in valence-consis-
tent knowledge being accessed in consumers’ associative
1
The research and development costs incurred in developing new drugs
are very high. This is behind the rationale for granting market exclusivity (in
the form of patents) to brand name drug manufacturers in order to
compensate for the drug development costs. It is only possible for a brand
name drug to have a generic counterpart after the patent on the drug expires,
by which time the drug development costs have supposedly been recovered.
Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66 57
memory. Our results demonstrate that two important
accounting constructs (pro?t margin and cost) are endowed
witha descriptive valence that evokes affective responses in
the consumers’ cognitive process. These ?ndings exemplify
how the descriptive valence of accounting constructs can
in?uence the decision outcome in a consumer setting and
illustrate the richness of accounting constructs.
In addition, this study is of interest to policy makers and
insurers concerned with increasing the effectiveness of
education campaigns aimed at improving patient accep-
tance of generic drugs. The results from our study highlight
the need to pay attention to the frame as well as to the
content of messages when designing such campaigns. As
promoting wider use of generic drugs plays an important
role in containing health care expenditures, our ?ndings
can have direct implications for efforts to reduce health
care costs without compromising quality and safety.
The remainder of the paper proceeds as follows. The
next two sections review the theoretical framework and
develop the research hypotheses. We then describe the
data and experimental design. The following two sections
present the empirical analysis and results. The last section
summarizes and concludes.
Theoretical framework
Cognitive representation of accounting information
Previous consumer research has established that con-
sumers respond to cues on both intrinsic and extrinsic attri-
butes of products when making purchase decisions
(Zeithaml, 1988). Extrinsic cues (e.g., brand name) can
impact purchase behavior when consumers adopt personi-
?ed products and form stereotypes about ?rms in a similar
fashion to person perception (Aaker, 1997; Aggarwal &
McGill, 2007; Fournier, 1998). Given that accounting con-
structs can be endowed with mental implications beyond
their explicit denotative meaning (Flamholtz & Cook,
1978; Haried, 1972, 1973; Houghton, 1987, 1988; Oliver,
1974), consumers can formimplicit impressions of the ?rm
based on the accounting constructs they encounter.
The model in Zeithaml (1988) characterizes consumers’
purchase decisions as a tradeoff between the bene?t/‘‘get’’
and sacri?ce/‘‘give’’ elements. This framework helps
account for how consumers interpret pro?t/cost informa-
tion. From the consumer’s perspective, the manufacturer’s
pro?t margin likely represents a ‘‘give’’ element because it
belongs to the manufacturer, thus carrying a negative con-
notation. In contrast, cost likely carries a positive connota-
tion because it can be considered a ‘‘get’’ element that
captures what is put into the product, which belongs to
the consumer.
2
Previous research provides some evidence that infor-
mation on the ?nancial pro?le of the manufacturers can
act as extrinsic cues that impact consumer perceptions
and purchase decisions. For example, Kachelmeier et al.
(1991) demonstrate in a laboratory market that consumers
justify price increases driven by increases in sellers’ costs,
but resist price increases that increase sellers’ pro?ts, as
they viewsuch price increases as unfair. On the other hand,
Aaker, Vohs, and Mogilner (2010) document that con-
sumers are less willing to buy products made by nonpro?t
organizations, as consumers associate not earning pro?ts
with the manufacturer being ineffective in achieving its
goal and thus less competent. These ?ndings suggest that
information on the ?nancial pro?le of the manufacturer
can generate affective responses in consumers’ associative
memory, which in turn in?uence purchase decisions.
The research on associative activation and valence-
based encoding provides a theoretical framework that
helps account for the way pro?t margin and cost are
encoded in the associative memory. As Levin, Schneider,
and Gaeth (1998) point out, attribute framing reveals that
the valence of a description at the most basic level can
in?uence the processing of that information. With attri-
bute framing, information is encoded relative to its
descriptive valence: A description of positive valence leads
to an encoding of the information that tends to evoke
favorable associations in memory, whereas the same infor-
mation of negative valence is likely to cause an encoding
that evokes unfavorable associations. Consequently, attri-
bute framing is characterized by a valence-consistent shift
wherein stimulus of positive (negative) valence leads to
knowledge with a more positive (negative) valence being
accessed in associative memory.
Consistent with the associative model of attribute fram-
ing, the negative valence of pro?t margin can result in selec-
tive cognitive search mechanisms for negatively valenced
knowledge being accessed in associative memory: That is,
high pro?t margin prompts suspicions of unfairness while
low pro?t margin indicates incompetence. Similarly, the
positive valence of cost can result in selective cognitive
search mechanisms for positively valenced knowledge
being accessed in associative memory: that is, associating
higher cost withhigher quality(as it costs moretomakebet-
ter products) while associating lower cost with better value
(as lower cost means less waste). As attribute framing high-
lights how the information is presented at the most basic
level can have a substantial in?uence on consumers’ mental
representation of the information, the information frame
likely plays a key role in determining how consumers
respondto disclosures regarding manufacturers’ pro?t mar-
gin/cost pro?le. The presentation of information is likely to
be particularly important when comparison is involved,
given that previous ?ndings show that simply changing
the direction of comparison leads to different comparative
judgments (Wanke et al., 1995). Below, we discuss the liter-
ature on this phenomenon and its implications for our
research setting.
Direction of comparison asymmetries
Areference point is the cognitive anchoring point inrela-
tion to which other items are seen (Rosch, 1975). The notion
that the human perceptual process tends to be structured in
terms of reference points has broad implications for
2
Lipe (1993) documents that evaluators with a cost frame provide
higher performance ratings than those with a loss frame. While the ?ndings
come from a non-consumer setting, they suggest that cost can act similarly
as a ‘‘gain’’ frame in the cognitive process.
58 Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66
cognitive and consumer behavior research (Klein and
Oglethorpe, 1987; Puto, 1987). As the reference point is
continuously modi?ed by a variety of environmental factors
up to the time of the purchase decision, a stimulus able to
shift the consumer’s reference point canin?uenceconsumer
perception and, in turn, consumer choice.
One well-documentedphenomenonthat exempli?es the
shift in reference point is direction of comparison asymme-
tries (Wanke et al., 1995). Roese et al. (1998), as well as
Bowdle and Medin (2001), provide evidence that direction
of comparison asymmetries can be attributed to the linguis-
tic norm that places the more common item in the referent
position. They conclude that the target term (subject) and
the base term (referent) of a comparison play different
semantic roles, with the subject acting as the ?gure and
the referent acting as the ground. Moreover, the syntax loca-
tions of the items tend to be assimilated as the conceptual
locations of the items. In other words, the referent of a com-
parison tends to act as the cognitive referent point against
which the subject of the comparison is evaluated.
In this study, we manipulate the direction of compar-
isons to shift consumers’ cognitive reference point in order
to identify the affective responses to the accounting con-
structs in which we are interested. Speci?cally, we vary
the comparison framing (i.e., which one is placed as the
subject vs. referent in the comparison) in the statements
that compare the pro?t margins and costs of brand name
and generic drug manufacturers and measure whether
there are signi?cant differences in consumer attitudes
and purchase intentions across groups. In the next section,
we develop the hypotheses to be tested through our exper-
imental design.
Hypothesis development
As discussed in the last section, manufacturers’ pro?t
margin likely represents a ‘‘give’’ element in the consumers’
mental model and thus carries a negative connotation. Due
to valence-based encoding in the consumers’ cognitive pro-
cess, the negative valence of pro?t margin can cause con-
sumers to selectively attend to negatively valenced
associations in memory (associating higher pro?t margin
with unfairness while associating lower pro?t margin with
incompetence). Giventhe directionof comparisonasymme-
tries discussed earlier (Bowdle & Medin, 2001; Roese et al.,
1998; Wanke et al., 1995), the subject of a comparison is
likely to be evaluated more intensively than the referent.
Accordingly, we expect that disclosure comparing manufac-
turers’ pro?t margin will predispose consumers to associate
the negative valence of pro?t margin with the subject ?rm
but not the referent ?rm. This implies that disclosure of
the relative pro?t margin will result in a more favorable
image of the referent ?rm relative to the subject ?rm.
Thus, we propose to test the following hypothesis:
H1. Disclosure of the relative pro?t margin and the frame
of the disclosure will jointly impact consumer perceptions
such that the ‘‘brand name as subject’’ condition will result
in a more favorable image of generic drug manufacturers
than the ‘‘generic as subject’’ condition.
As manufacturers’ cost likely represents a ‘‘get’’ element
in the consumers’ mental model, the positive valence of
cost can cause consumers to selectively attend to posi-
tively valenced associations in memory (associating higher
cost with better quality while associating lower cost with
better value). Similar to H1, varying the directions of com-
parison is hypothesized to result in signi?cant differences
in consumer perceptions since the subject of the compar-
ison acts as the ?gure being evaluated while the referent
plays the implicit role of the ground in consumers’ cogni-
tive process. Contrary to disclosures that compare manu-
facturers’ pro?t margin, disclosures that compare
manufacturers’ cost are likely to positively impact the sub-
ject of the comparison due to the positive valence of cost.
Accordingly, we propose to test the following hypothesis:
H2. Disclosure of relative cost levels and the frame of the
disclosure will jointly impact consumer perceptions such
that the ‘‘generic as subject’’ condition will result in a more
favorable image of generic drug manufacturers than the
‘‘brand name as subject’’ condition.
We apply the stimulus–orientation–response (S–O–R)
approach (Markus & Zajonc, 1985) to examine whether
exposure to information about the manufacturers’ pro?t
margin/cost pro?le impacts eventual consumer choice
between brand name and generic drugs through the per-
ceived image of manufacturers. With the S–O–R approach,
cognitive constructs (e.g., minds, internal representations,
and thoughts) are direct responses to stimulus, and they
are also antecedents of behavioral consequences, so these
cognitive constructs are the O in the S–O–R approach,
mediating the relationship between stimulus and ?nal
response (Markus & Zajonc, 1985).
In this study, we conceptualize image of generic drug
manufacturers as the orientation or cognitive variable, pre-
dicted by the stimuli and predicting the ?nal response
variables: intention to communicate with doctors and phar-
macists about generic drugs and intention to purchase generic
drugs. We focus on the intention variables as the ?nal
response for two reasons. First, theoretical work on rea-
soned action argues that an individual’s intention to per-
form a behavior is an important predictor of whether
that individual will actually perform the behavior (Ajzen,
Albarracín, & Hornik, 2007; Fishbein & Ajzen, 1975).
Second, empirical evidence provides strong support for
intention variables acting as precursors of actual behavior.
For example, Sheppard, Hartwick, and Warshaw (1988)
performed a large-scale meta-analysis and found that the
Intention-Behavior correlation is both strong and robust.
In the context of our research setting, we hypothesize that
the image of generic drug manufacturers has a direct effect
on the intention variables, and this image variable medi-
ates the relationship between the stimuli and the intention
variables. Accordingly, we propose to test the following
hypotheses:
H3a. The image of generic drug manufacturers will medi-
ate the effect of ?nancial disclosures and message frame on
consumer intentions to communicate with doctors and
pharmacists about generic drugs.
Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66 59
H3b. The image of generic drug manufacturers will medi-
ate the effect of ?nancial disclosures and message frame on
consumer intentions to purchase generic drugs.
Experimental method
We have designed a 2 (disclosure that compares costs
vs. disclosure that compares pro?t margins) ? 2 (brand
name as subject frame vs. generic as subject frame) factorial
experiment to test our research hypotheses. To obtain a
sample representative of the U.S. adult population, we
hired Knowledge Networks to conduct the experiment.
As part of the GfK Group (the ?fth largest market research
organization in the world), Knowledge Networks is special-
ized in providing data collection services for marketing,
media, health care, and social policy research. Sample pro-
jects by Knowledge Networks include JAMRS Advertising
Tracking Study (Fors Marsh Group LLC, funded by the U.S.
Department of Defense), Election 2008 & Beyond
(University of Chicago, funded by the Ford Foundation),
American National Election Surveys (Stanford University,
funded by the National Science Foundation), Public
Opinion & Deliberation (University of Pennsylvania, funded
by the National Institutes of Health), and Annenberg
National Health Communication Survey (University of
Pennsylvania and University of Southern California).
Knowledge Networks randomly recruits participants
through address-based probability sampling. Knowledge
Networks’ sample is more representative than samples
based on the traditional random-digit dialing methods or
an online ‘‘opt-in’’ convenient sample, since Knowledge
Networks’ sample represents listed and unlisted telephone
numbers, telephone and non-telephone households, cell-
phone-only households, and households with and without
Internet access. When needed, Knowledge Networks pro-
vided households with a netbook computer and Internet
access so that the households could complete online
experiments/surveys.
Knowledge Networks conducted the experiment over a
ten-day period. It invited 1007 participants representing
non-institutionalized adults aged 18 years or older resid-
ing in the United States to participate in the study, of
whom 608 completed the experiment online. To improve
the internal validity of the study, we deleted cases with
missing values for the variables that we used in this study.
Thus, our ?nal sample consists of 563 participants.
3
Pre-stimulus variables
All participants completed the experiment online.
Participants began by answering questions about their
prior experience with prescription medications.
Speci?cally, the participants ?rst answered the question:
‘‘During the past 12 months, how many times did you
buy prescription drugs?’’ The answer options to this ques-
tion are on a seven-point scale ranging from 1 (0 times) to
7 (26 times or more).
The participants then read the following background
information about brand name and generic drugs: ‘‘A
brand name medication can only be produced and sold
by the company that holds the patent for the drug. When
the patent of a brand name medication expires, a generic
version of the drug can be produced and sold by other
companies. All generic drugs must be reviewed and
approved by the U.S. Food and Drug Administration
(FDA). The FDA requires that a generic version of a drug
use the same active ingredient(s) as the brand name drug.
Additionally, the FDA requires that generic drug manufac-
turing plants meet the same high standards as plants for
brand name drugs.’’
Next the participants answered the following
questions:
(1) ‘‘During the past 12 months, what was the per-
centage of generic drugs you bought for prescription
drugs?’’ The answer options to this question are on a ?ve
point scale ranging from 1 (0%) to 5 (100%); (2) ‘‘On a
1–7 scale, how strongly do you agree with the following
statement: During the past 12 months, my doctor talked
to me about generic medications.’’ The answer options to
this question are on a seven-point Likert-type scale with
1 = strongly disagree and 7 = strongly agree; and (3)
‘‘On a 1 to 7 scale, how strongly do you agree with the
following statement: During the past 12 months, my
pharmacist talked to me about generic medications.’’
The answer options to this question are on a seven-point
Likert-type scale with 1 = strongly disagree and 7 = strongly
agree.
We use the answers to the pre-stimulus questions as
control variables for our analyses.
Stimuli
We designed the following four stimulus messages for
the four experimental conditions:
(1) Pro?t Margin and brand name as subject Frame
Condition: In general, brand name drugs are more
expensive than generic drugs. We also know that
brand name drug manufacturers tend to enjoy
higher pro?t margin than generic drug
manufacturers.
(2) Pro?t Margin and generic as subject Frame Condition:
In general, generic drugs are less expensive than
brand name drugs. We also know that generic drug
manufacturers tend to enjoy lower pro?t margin
than brand name drug manufacturers.
3
In order to address the potential non-response bias, we compare the
563 participants who provided complete data and the 45 who did not.
Speci?cally, we conducted four one-way ANOVA tests to examine whether
the two groups of participants differed in four pre-stimulus variables likely
to be correlated with our outcome variables. The results indicated that
there were no signi?cant differences between the two groups in previous
generic drug consumption (F = 0.53, p = 0.47), information from doctors
about generic medications (F = 0.19, p = 0.67), or information from phar-
macists about generic medications (F = 0.34, p = 0.56). On the other hand,
participants who provided complete data reported higher overall prescrip-
tion drug consumption (Mean = 2.75, SD = 1.64) than those who did not
provide complete data (Mean = 2.16, SD = 1.51), and the difference was
signi?cant (F = 5.09, p = 0.02). This seems to suggest that participants who
consume more prescriptions drugs are more likely to provide complete
data in the experiment. This is a good ‘‘bias’’ as the results of this study will
be of greater signi?cance if the sample is weighted towards the population
who consume more prescription drugs.
60 Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66
(3) Cost and brand name as subject Frame Condition: In
general, brand name drugs are more expensive than
generic drugs. We also know that brand name drug
manufacturers generally face higher costs than gen-
eric drug manufacturers.
(4) Cost and generic as subject Frame Condition: In gen-
eral, generic drugs are less expensive than brand
name drugs. We also know that generic drug manu-
facturers generally face lower costs than brand
name drug manufacturers.
Knowledge Networks randomly assigned each partici-
pant to one of the four above conditions. Distribution of
participants for each condition is approximately even, with
133 (23.6%) for Condition A, 141 (25%) for Condition B, 141
(25%) for Condition C, and 148 (26.3%) for Condition D. The
four experimental groups do not differ in general prescrip-
tion drug consumption (p = .54), percentage of generic
drug consumption (p = .84), information from doctors on
generic medications (p = .94), or information from pharma-
cists on generic medications (p = .95).
Outcome variables
After participants read their respective stimulus mes-
sage, they answered questions about their views of brand
name and generic drug manufacturers, their intention to
communicate with doctors and pharmacists about their
preference for generic and brand name drugs, and their
intention to purchase generic drugs over band name drugs.
We use the answers to these questions as the dependent
variables for our analyses.
As our study focuses on consumer acceptance of generic
drugs in substitution of brand name drugs, we measure all
the outcome measures as the within-subject differences in
attitudes towards generic vs. brand name drugs.
Speci?cally, we measure the image of generic drug manu-
facturers as the difference between the answers to the fol-
lowing two questions: ‘‘On a 1–7 scale, how do you view
generic pharmaceutical companies’ image?’’ (1 = very
unfavorable and 7 = very favorable), and ‘‘On a 1–7 scale,
how do you view brand name pharmaceutical companies’
image?’’ (1 = very unfavorable and 7 = very favorable).
We measure the intention to communicate on generic
drugs as the arithmetic average of the two variables
described below. The ?rst variable is the difference
between the answers to the following two questions: ‘‘I
intend to talk to my doctor about my preference for generic
drugs for my next prescription.’’ (1 = strongly disagree and
7 = strongly agree), and ‘‘I intend to talk to my doctor about
my preference for brand name drugs for my next prescrip-
tion.’’ (1 = strongly disagree and 7 = strongly agree). The
second variable is the difference between the answers to
the following two questions: ‘‘I intend to talk to my phar-
macist about my preference for generic drugs for my next
prescription.’’ (1 = strongly disagree and 7 = strongly
agree), and ‘‘I intend to talk to my pharmacist about my
preference for brand name drugs for my next prescription.’’
(1 = strongly disagree and 7 = strongly agree). We measure
the intention to purchase generic drugs as the difference
between the answers to the following two questions: ‘‘I
would rather buy generic than brand name medications
for my next prescription.’’ (1 = strongly disagree and
7 = strongly agree), and ‘‘I would rather buy brand name
than generic medications for my next prescription.’’
(1 = strongly disagree and 7 = strongly agree). The answer
options to all the intention questions are on a seven-point
Table 1
Demographic information.
Variable Value Frequency Percentage
Gender Female 299 53.1
Male 264 46.9
Age 18–29 87 15.5
30–44 142 25.2
45–59 156 27.7
60+ 178 31.6
Race White, Non-Hispanic 411 73
Black, Non-Hispanic 54 9.6
Other, Non-Hispanic 21 3.7
Hispanic 63 11.2
2+Races, Non-
Hispanic
14 2.5
Education Less than high school 56 9.9
High School 167 29.7
Some college 168 29.8
Bachelor’s degree or
higher
172 30.6
Household
income
We investigate how accounting information and message framing jointly impact consumer
choice between brand name and generic drugs using a sample representative of the U.S.
adult population. We find that information about manufacturers’ profit margin/cost profile
in different frames predisposes consumers to develop more or less favorable attitudes
towards the firms and their products. The results suggest that two important accounting
constructs (profit margin and cost) are endowed with a descriptive valence that evokes
affective responses in consumers’ associative memory. This study adds to earlier work
on the role of accounting information’s connotative meaning in shaping user perceptions
by providing new evidence that the descriptive valence of accounting constructs can
impact consumer purchase decisions.
From bottom line to consumers’ mind: The framing effects
of accounting information
Yan Tian
a
, Hui Zhou
b,?
a
Department of Communication, University of Missouri-St. Louis, 1 University Blvd, St Louis, MO, United States
b
Melbourne Business School, The University of Melbourne, 200 Leicester Street, Carlton, VIC, Australia
a r t i c l e i n f o
Article history:
Available online 4 May 2015
a b s t r a c t
We investigate how accounting information and message framing jointly impact consumer
choice between brand name and generic drugs using a sample representative of the U.S.
adult population. We ?nd that information about manufacturers’ pro?t margin/cost pro?le
in different frames predisposes consumers to develop more or less favorable attitudes
towards the ?rms and their products. The results suggest that two important accounting
constructs (pro?t margin and cost) are endowed with a descriptive valence that evokes
affective responses in consumers’ associative memory. This study adds to earlier work
on the role of accounting information’s connotative meaning in shaping user perceptions
by providing new evidence that the descriptive valence of accounting constructs can
impact consumer purchase decisions. The ?ndings of this study also have direct implica-
tions for efforts aimed at reducing health care costs by promoting wider use of generic
drugs.
Ó 2015 Elsevier Ltd. All rights reserved.
Introduction
Accounting information affects not only stakeholders
who have ?nancial interests in the reporting company,
but also other stakeholder groups including regulatory
agencies, employees, and consumers. However, little
research has examined the effects of ?nancial reporting
on stakeholder groups other than investors (Kachelmeier,
Stephen, & Schadewald, 1991). In this study, we focus on
consumers and investigate how accounting information
and message framing jointly impact consumer judgment
and choice.
Accounting research has long established that the con-
notative meaning of accounting information plays a role
in shaping the perceptions and thoughts of those who are
exposed to the information (Flamholtz & Cook, 1978;
Haried, 1972, 1973; Houghton, 1987, 1988; Oliver, 1974).
These ?ndings can be better understood in light of the
development in psychological research that highlights the
importance of associative memory in the cognitive process.
Morewedge and Kahneman (2010) concluded that the pri-
mary mechanism underlying the well-documented prim-
ing and framing effects is the automatic operations of
associative memory. Associative memory is de?ned as ‘‘a
network of long-term memory for semantic information,
emotions and goals that is governed by the spread of acti-
vation, as determined by the strengths of interconnecting
weights (associations)’’ (Morewedge & Kahneman, 2010,
p. 435). The associative processes operate preconsciously
in the slow-learning memory system, quickly and auto-
matically generating intuitive and affective responses to
objects (Smith & Decoster, 2000; Uleman, Saribay, &
Gonzalez, 2008). Given the importance of semantic infor-
mation in associative memory, the ‘‘semantic halo effect’’
of accounting information documented in the aforemen-
tioned studies likely arises from the automatic activation
of compatible associations based on the way thehttp://dx.doi.org/10.1016/j.aos.2015.04.003
0361-3682/Ó 2015 Elsevier Ltd. All rights reserved.
?
Corresponding author. Tel.: +61 3 93498710.
E-mail address: [email protected] (H. Zhou).
Accounting, Organizations and Society 43 (2015) 56–66
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
accounting constructs are encoded in the slow-learning
memory system.
This study adds to earlier work on the role of accounting
constructs’ connotative meaning in shaping user percep-
tions by investigating the automatic affective reactions of
consumers to information about the manufacturers’ pro?t
margin/cost pro?le presented in different frames. We posit
that pro?t margin (cost) has a negative (positive) descrip-
tive valence, and thus information about the manufactur-
ers’ relative pro?t margins (costs) will evoke negative
(positive) affective responses. Drawing on direction of
comparison of asymmetries documented in previous
research (e.g., Bowdle & Medin, 2001; Roese, Sherman, &
Hur, 1998; Wanke, Schwarz, & Noelle-Neumann, 1995),
we further posit that comparison framing (i.e., which ?rm
is placed as the subject vs. referent in a comparison of two
?rms) can affect consumer attitudes, as consumers will
associate the affective responses with the subject ?rm
but not the referent ?rm. This implies that consumers will
be unfavorably predisposed towards the subject ?rm rela-
tive to the referent ?rm when a comparison of pro?t mar-
gins is presented. Conversely, consumers will be favorably
predisposed towards the subject ?rm relative to the refer-
ent ?rm when a comparison of costs is presented. The rel-
atively favorable (unfavorable) image of a ?rm, in turn, will
increase (decrease) consumer intention to choose that
?rm’s products over those of its competitor.
We use consumer choice between brand name and gen-
eric drugs as the setting to test our propositions.
Pharmaceuticals provide a suitable setting to investigate
our research question for two reasons. First, prescription
prices, as well as manufacturers’ pro?t margins and costs,
vary signi?cantly between brand name and generic drugs.
1
Bhat (2005) reports that the average prescription price of a
brandname drug is $65.29comparedto only $19.33for a gen-
eric drug in 2000, and net pro?t margin before taxes of ten
brand name pharmaceutical manufacturers was 23.6% com-
pared to only 17.2% for ten generic pharmaceutical manufac-
turers. Second, there is widespread belief that consumers
view the price difference between brand name and generic
drugs as a signal of difference in quality, despite the fact that
generic drugs have to meet the same rigid standards as brand
name drugs in the FDA approval process. For example, a 2008
New York Times editorial titled ‘‘Generic Drug Resistance’’
highlighted the dif?culty in persuading more patients to use
generic drugs, even in the presence of substantial evidence
con?rming the effectiveness and safety of generics. This is
consistent with the phenomenon of price reliance docu-
mentedinconsumer researchwhereprices are usedas quality
indicators, especially in purchase decisions of ‘‘credence
goods’’ such as pharmaceuticals (Zeithaml, 1988).
In addition to providing a setting for testing the joint
effect of accounting information and message framing on
consumers, consumer choice between generic and brand
name medications is an important policy issue. Policy
makers and insurers have long promoted wider use of gen-
eric drugs in place of their more expensive brand name
alternatives as an effective mechanism to contain prescrip-
tion drug spending. Using 1997–2000 Medical Expenditure
Panel data, Haas, Phillips, Gerstenberger, and Seger (2005)
estimated that broad generic substitution of outpatient
prescription drugs would result in a national savings of
$8.8 billion in the United States each year, approximately
11% of the annual drug expenditures in the United States
during 1997 to 2000. With the rapid increase in prescrip-
tion drug expenditures during the last decade, the absolute
savings from substituting generic drugs for brand name
drugs will be even more substantial going forward. The
magnitude of these ?gures underlies the efforts of policy
makers and insurers to increase the use of generic drugs.
A signi?cant number of such programs are speci?cally tar-
geted at improving consumer perceptions of generic drugs
because the perceptions of patients play an important role
in deciding which medications are prescribed to them
(Shrank et al., 2009).
We have designed an experiment to investigate whether
exposing consumers to information on manufacturers’
pro?t margin/cost pro?le can evoke affective responses that
affect consumer attitudes towards the ?rms and their prod-
ucts. The participants are a sample representative of the U.S.
adult population. To identify the affective responses to
pro?t margin (cost), we vary the comparison framing in
the statements that compare the pro?t margin (cost) of
brand name and generic drug manufacturers by manipulat-
ing the directionof comparisons. For simplicity in reference,
pro?t margin (cost) disclosures stating that brand name
drug manufacturers have higher pro?t margin (cost) than
generic drug manufacturers are referred to as the brand
name as subject condition. Conversely, pro?t margin (cost)
disclosures stating that generic drug manufacturers have
lower pro?t margin (cost) than brand name drug manufac-
turers are referred to as the generic as subject condition.
Consistent with our predictions, the results show that
disclosures about the manufacturers’ pro?t margin/cost
pro?le and the frame of disclosures have interactive effects
on consumer perceptions and purchase intentions.
Speci?cally, when information about relative pro?t margin
is presented, the brand name as subject frame is associated
with a more favorable image of generic drug manufacturers
than the generic as subject frame. In contrast, when informa-
tion about relative cost levels is presented, the generic as
subject frame is associated with a more favorable image of
generic drug manufacturers than the brand name as subject
frame. Finally, the image of generic drug manufacturers
mediates the effect of the disclosures on consumer inten-
tions to communicate about and purchase generic drugs.
This studycontributes toresearchthat examines the cog-
nitive mechanisms underlying the response to accounting
information and extends the literature on the role of
accounting information’s connotative meaning in shaping
user perceptions. We provide evidence that encountering
information about the manufacturers’ pro?t margin/cost
pro?le in different frames leads to selective attention and
cognitive search mechanisms that result in valence-consis-
tent knowledge being accessed in consumers’ associative
1
The research and development costs incurred in developing new drugs
are very high. This is behind the rationale for granting market exclusivity (in
the form of patents) to brand name drug manufacturers in order to
compensate for the drug development costs. It is only possible for a brand
name drug to have a generic counterpart after the patent on the drug expires,
by which time the drug development costs have supposedly been recovered.
Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66 57
memory. Our results demonstrate that two important
accounting constructs (pro?t margin and cost) are endowed
witha descriptive valence that evokes affective responses in
the consumers’ cognitive process. These ?ndings exemplify
how the descriptive valence of accounting constructs can
in?uence the decision outcome in a consumer setting and
illustrate the richness of accounting constructs.
In addition, this study is of interest to policy makers and
insurers concerned with increasing the effectiveness of
education campaigns aimed at improving patient accep-
tance of generic drugs. The results from our study highlight
the need to pay attention to the frame as well as to the
content of messages when designing such campaigns. As
promoting wider use of generic drugs plays an important
role in containing health care expenditures, our ?ndings
can have direct implications for efforts to reduce health
care costs without compromising quality and safety.
The remainder of the paper proceeds as follows. The
next two sections review the theoretical framework and
develop the research hypotheses. We then describe the
data and experimental design. The following two sections
present the empirical analysis and results. The last section
summarizes and concludes.
Theoretical framework
Cognitive representation of accounting information
Previous consumer research has established that con-
sumers respond to cues on both intrinsic and extrinsic attri-
butes of products when making purchase decisions
(Zeithaml, 1988). Extrinsic cues (e.g., brand name) can
impact purchase behavior when consumers adopt personi-
?ed products and form stereotypes about ?rms in a similar
fashion to person perception (Aaker, 1997; Aggarwal &
McGill, 2007; Fournier, 1998). Given that accounting con-
structs can be endowed with mental implications beyond
their explicit denotative meaning (Flamholtz & Cook,
1978; Haried, 1972, 1973; Houghton, 1987, 1988; Oliver,
1974), consumers can formimplicit impressions of the ?rm
based on the accounting constructs they encounter.
The model in Zeithaml (1988) characterizes consumers’
purchase decisions as a tradeoff between the bene?t/‘‘get’’
and sacri?ce/‘‘give’’ elements. This framework helps
account for how consumers interpret pro?t/cost informa-
tion. From the consumer’s perspective, the manufacturer’s
pro?t margin likely represents a ‘‘give’’ element because it
belongs to the manufacturer, thus carrying a negative con-
notation. In contrast, cost likely carries a positive connota-
tion because it can be considered a ‘‘get’’ element that
captures what is put into the product, which belongs to
the consumer.
2
Previous research provides some evidence that infor-
mation on the ?nancial pro?le of the manufacturers can
act as extrinsic cues that impact consumer perceptions
and purchase decisions. For example, Kachelmeier et al.
(1991) demonstrate in a laboratory market that consumers
justify price increases driven by increases in sellers’ costs,
but resist price increases that increase sellers’ pro?ts, as
they viewsuch price increases as unfair. On the other hand,
Aaker, Vohs, and Mogilner (2010) document that con-
sumers are less willing to buy products made by nonpro?t
organizations, as consumers associate not earning pro?ts
with the manufacturer being ineffective in achieving its
goal and thus less competent. These ?ndings suggest that
information on the ?nancial pro?le of the manufacturer
can generate affective responses in consumers’ associative
memory, which in turn in?uence purchase decisions.
The research on associative activation and valence-
based encoding provides a theoretical framework that
helps account for the way pro?t margin and cost are
encoded in the associative memory. As Levin, Schneider,
and Gaeth (1998) point out, attribute framing reveals that
the valence of a description at the most basic level can
in?uence the processing of that information. With attri-
bute framing, information is encoded relative to its
descriptive valence: A description of positive valence leads
to an encoding of the information that tends to evoke
favorable associations in memory, whereas the same infor-
mation of negative valence is likely to cause an encoding
that evokes unfavorable associations. Consequently, attri-
bute framing is characterized by a valence-consistent shift
wherein stimulus of positive (negative) valence leads to
knowledge with a more positive (negative) valence being
accessed in associative memory.
Consistent with the associative model of attribute fram-
ing, the negative valence of pro?t margin can result in selec-
tive cognitive search mechanisms for negatively valenced
knowledge being accessed in associative memory: That is,
high pro?t margin prompts suspicions of unfairness while
low pro?t margin indicates incompetence. Similarly, the
positive valence of cost can result in selective cognitive
search mechanisms for positively valenced knowledge
being accessed in associative memory: that is, associating
higher cost withhigher quality(as it costs moretomakebet-
ter products) while associating lower cost with better value
(as lower cost means less waste). As attribute framing high-
lights how the information is presented at the most basic
level can have a substantial in?uence on consumers’ mental
representation of the information, the information frame
likely plays a key role in determining how consumers
respondto disclosures regarding manufacturers’ pro?t mar-
gin/cost pro?le. The presentation of information is likely to
be particularly important when comparison is involved,
given that previous ?ndings show that simply changing
the direction of comparison leads to different comparative
judgments (Wanke et al., 1995). Below, we discuss the liter-
ature on this phenomenon and its implications for our
research setting.
Direction of comparison asymmetries
Areference point is the cognitive anchoring point inrela-
tion to which other items are seen (Rosch, 1975). The notion
that the human perceptual process tends to be structured in
terms of reference points has broad implications for
2
Lipe (1993) documents that evaluators with a cost frame provide
higher performance ratings than those with a loss frame. While the ?ndings
come from a non-consumer setting, they suggest that cost can act similarly
as a ‘‘gain’’ frame in the cognitive process.
58 Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66
cognitive and consumer behavior research (Klein and
Oglethorpe, 1987; Puto, 1987). As the reference point is
continuously modi?ed by a variety of environmental factors
up to the time of the purchase decision, a stimulus able to
shift the consumer’s reference point canin?uenceconsumer
perception and, in turn, consumer choice.
One well-documentedphenomenonthat exempli?es the
shift in reference point is direction of comparison asymme-
tries (Wanke et al., 1995). Roese et al. (1998), as well as
Bowdle and Medin (2001), provide evidence that direction
of comparison asymmetries can be attributed to the linguis-
tic norm that places the more common item in the referent
position. They conclude that the target term (subject) and
the base term (referent) of a comparison play different
semantic roles, with the subject acting as the ?gure and
the referent acting as the ground. Moreover, the syntax loca-
tions of the items tend to be assimilated as the conceptual
locations of the items. In other words, the referent of a com-
parison tends to act as the cognitive referent point against
which the subject of the comparison is evaluated.
In this study, we manipulate the direction of compar-
isons to shift consumers’ cognitive reference point in order
to identify the affective responses to the accounting con-
structs in which we are interested. Speci?cally, we vary
the comparison framing (i.e., which one is placed as the
subject vs. referent in the comparison) in the statements
that compare the pro?t margins and costs of brand name
and generic drug manufacturers and measure whether
there are signi?cant differences in consumer attitudes
and purchase intentions across groups. In the next section,
we develop the hypotheses to be tested through our exper-
imental design.
Hypothesis development
As discussed in the last section, manufacturers’ pro?t
margin likely represents a ‘‘give’’ element in the consumers’
mental model and thus carries a negative connotation. Due
to valence-based encoding in the consumers’ cognitive pro-
cess, the negative valence of pro?t margin can cause con-
sumers to selectively attend to negatively valenced
associations in memory (associating higher pro?t margin
with unfairness while associating lower pro?t margin with
incompetence). Giventhe directionof comparisonasymme-
tries discussed earlier (Bowdle & Medin, 2001; Roese et al.,
1998; Wanke et al., 1995), the subject of a comparison is
likely to be evaluated more intensively than the referent.
Accordingly, we expect that disclosure comparing manufac-
turers’ pro?t margin will predispose consumers to associate
the negative valence of pro?t margin with the subject ?rm
but not the referent ?rm. This implies that disclosure of
the relative pro?t margin will result in a more favorable
image of the referent ?rm relative to the subject ?rm.
Thus, we propose to test the following hypothesis:
H1. Disclosure of the relative pro?t margin and the frame
of the disclosure will jointly impact consumer perceptions
such that the ‘‘brand name as subject’’ condition will result
in a more favorable image of generic drug manufacturers
than the ‘‘generic as subject’’ condition.
As manufacturers’ cost likely represents a ‘‘get’’ element
in the consumers’ mental model, the positive valence of
cost can cause consumers to selectively attend to posi-
tively valenced associations in memory (associating higher
cost with better quality while associating lower cost with
better value). Similar to H1, varying the directions of com-
parison is hypothesized to result in signi?cant differences
in consumer perceptions since the subject of the compar-
ison acts as the ?gure being evaluated while the referent
plays the implicit role of the ground in consumers’ cogni-
tive process. Contrary to disclosures that compare manu-
facturers’ pro?t margin, disclosures that compare
manufacturers’ cost are likely to positively impact the sub-
ject of the comparison due to the positive valence of cost.
Accordingly, we propose to test the following hypothesis:
H2. Disclosure of relative cost levels and the frame of the
disclosure will jointly impact consumer perceptions such
that the ‘‘generic as subject’’ condition will result in a more
favorable image of generic drug manufacturers than the
‘‘brand name as subject’’ condition.
We apply the stimulus–orientation–response (S–O–R)
approach (Markus & Zajonc, 1985) to examine whether
exposure to information about the manufacturers’ pro?t
margin/cost pro?le impacts eventual consumer choice
between brand name and generic drugs through the per-
ceived image of manufacturers. With the S–O–R approach,
cognitive constructs (e.g., minds, internal representations,
and thoughts) are direct responses to stimulus, and they
are also antecedents of behavioral consequences, so these
cognitive constructs are the O in the S–O–R approach,
mediating the relationship between stimulus and ?nal
response (Markus & Zajonc, 1985).
In this study, we conceptualize image of generic drug
manufacturers as the orientation or cognitive variable, pre-
dicted by the stimuli and predicting the ?nal response
variables: intention to communicate with doctors and phar-
macists about generic drugs and intention to purchase generic
drugs. We focus on the intention variables as the ?nal
response for two reasons. First, theoretical work on rea-
soned action argues that an individual’s intention to per-
form a behavior is an important predictor of whether
that individual will actually perform the behavior (Ajzen,
Albarracín, & Hornik, 2007; Fishbein & Ajzen, 1975).
Second, empirical evidence provides strong support for
intention variables acting as precursors of actual behavior.
For example, Sheppard, Hartwick, and Warshaw (1988)
performed a large-scale meta-analysis and found that the
Intention-Behavior correlation is both strong and robust.
In the context of our research setting, we hypothesize that
the image of generic drug manufacturers has a direct effect
on the intention variables, and this image variable medi-
ates the relationship between the stimuli and the intention
variables. Accordingly, we propose to test the following
hypotheses:
H3a. The image of generic drug manufacturers will medi-
ate the effect of ?nancial disclosures and message frame on
consumer intentions to communicate with doctors and
pharmacists about generic drugs.
Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66 59
H3b. The image of generic drug manufacturers will medi-
ate the effect of ?nancial disclosures and message frame on
consumer intentions to purchase generic drugs.
Experimental method
We have designed a 2 (disclosure that compares costs
vs. disclosure that compares pro?t margins) ? 2 (brand
name as subject frame vs. generic as subject frame) factorial
experiment to test our research hypotheses. To obtain a
sample representative of the U.S. adult population, we
hired Knowledge Networks to conduct the experiment.
As part of the GfK Group (the ?fth largest market research
organization in the world), Knowledge Networks is special-
ized in providing data collection services for marketing,
media, health care, and social policy research. Sample pro-
jects by Knowledge Networks include JAMRS Advertising
Tracking Study (Fors Marsh Group LLC, funded by the U.S.
Department of Defense), Election 2008 & Beyond
(University of Chicago, funded by the Ford Foundation),
American National Election Surveys (Stanford University,
funded by the National Science Foundation), Public
Opinion & Deliberation (University of Pennsylvania, funded
by the National Institutes of Health), and Annenberg
National Health Communication Survey (University of
Pennsylvania and University of Southern California).
Knowledge Networks randomly recruits participants
through address-based probability sampling. Knowledge
Networks’ sample is more representative than samples
based on the traditional random-digit dialing methods or
an online ‘‘opt-in’’ convenient sample, since Knowledge
Networks’ sample represents listed and unlisted telephone
numbers, telephone and non-telephone households, cell-
phone-only households, and households with and without
Internet access. When needed, Knowledge Networks pro-
vided households with a netbook computer and Internet
access so that the households could complete online
experiments/surveys.
Knowledge Networks conducted the experiment over a
ten-day period. It invited 1007 participants representing
non-institutionalized adults aged 18 years or older resid-
ing in the United States to participate in the study, of
whom 608 completed the experiment online. To improve
the internal validity of the study, we deleted cases with
missing values for the variables that we used in this study.
Thus, our ?nal sample consists of 563 participants.
3
Pre-stimulus variables
All participants completed the experiment online.
Participants began by answering questions about their
prior experience with prescription medications.
Speci?cally, the participants ?rst answered the question:
‘‘During the past 12 months, how many times did you
buy prescription drugs?’’ The answer options to this ques-
tion are on a seven-point scale ranging from 1 (0 times) to
7 (26 times or more).
The participants then read the following background
information about brand name and generic drugs: ‘‘A
brand name medication can only be produced and sold
by the company that holds the patent for the drug. When
the patent of a brand name medication expires, a generic
version of the drug can be produced and sold by other
companies. All generic drugs must be reviewed and
approved by the U.S. Food and Drug Administration
(FDA). The FDA requires that a generic version of a drug
use the same active ingredient(s) as the brand name drug.
Additionally, the FDA requires that generic drug manufac-
turing plants meet the same high standards as plants for
brand name drugs.’’
Next the participants answered the following
questions:
(1) ‘‘During the past 12 months, what was the per-
centage of generic drugs you bought for prescription
drugs?’’ The answer options to this question are on a ?ve
point scale ranging from 1 (0%) to 5 (100%); (2) ‘‘On a
1–7 scale, how strongly do you agree with the following
statement: During the past 12 months, my doctor talked
to me about generic medications.’’ The answer options to
this question are on a seven-point Likert-type scale with
1 = strongly disagree and 7 = strongly agree; and (3)
‘‘On a 1 to 7 scale, how strongly do you agree with the
following statement: During the past 12 months, my
pharmacist talked to me about generic medications.’’
The answer options to this question are on a seven-point
Likert-type scale with 1 = strongly disagree and 7 = strongly
agree.
We use the answers to the pre-stimulus questions as
control variables for our analyses.
Stimuli
We designed the following four stimulus messages for
the four experimental conditions:
(1) Pro?t Margin and brand name as subject Frame
Condition: In general, brand name drugs are more
expensive than generic drugs. We also know that
brand name drug manufacturers tend to enjoy
higher pro?t margin than generic drug
manufacturers.
(2) Pro?t Margin and generic as subject Frame Condition:
In general, generic drugs are less expensive than
brand name drugs. We also know that generic drug
manufacturers tend to enjoy lower pro?t margin
than brand name drug manufacturers.
3
In order to address the potential non-response bias, we compare the
563 participants who provided complete data and the 45 who did not.
Speci?cally, we conducted four one-way ANOVA tests to examine whether
the two groups of participants differed in four pre-stimulus variables likely
to be correlated with our outcome variables. The results indicated that
there were no signi?cant differences between the two groups in previous
generic drug consumption (F = 0.53, p = 0.47), information from doctors
about generic medications (F = 0.19, p = 0.67), or information from phar-
macists about generic medications (F = 0.34, p = 0.56). On the other hand,
participants who provided complete data reported higher overall prescrip-
tion drug consumption (Mean = 2.75, SD = 1.64) than those who did not
provide complete data (Mean = 2.16, SD = 1.51), and the difference was
signi?cant (F = 5.09, p = 0.02). This seems to suggest that participants who
consume more prescriptions drugs are more likely to provide complete
data in the experiment. This is a good ‘‘bias’’ as the results of this study will
be of greater signi?cance if the sample is weighted towards the population
who consume more prescription drugs.
60 Y. Tian, H. Zhou/ Accounting, Organizations and Society 43 (2015) 56–66
(3) Cost and brand name as subject Frame Condition: In
general, brand name drugs are more expensive than
generic drugs. We also know that brand name drug
manufacturers generally face higher costs than gen-
eric drug manufacturers.
(4) Cost and generic as subject Frame Condition: In gen-
eral, generic drugs are less expensive than brand
name drugs. We also know that generic drug manu-
facturers generally face lower costs than brand
name drug manufacturers.
Knowledge Networks randomly assigned each partici-
pant to one of the four above conditions. Distribution of
participants for each condition is approximately even, with
133 (23.6%) for Condition A, 141 (25%) for Condition B, 141
(25%) for Condition C, and 148 (26.3%) for Condition D. The
four experimental groups do not differ in general prescrip-
tion drug consumption (p = .54), percentage of generic
drug consumption (p = .84), information from doctors on
generic medications (p = .94), or information from pharma-
cists on generic medications (p = .95).
Outcome variables
After participants read their respective stimulus mes-
sage, they answered questions about their views of brand
name and generic drug manufacturers, their intention to
communicate with doctors and pharmacists about their
preference for generic and brand name drugs, and their
intention to purchase generic drugs over band name drugs.
We use the answers to these questions as the dependent
variables for our analyses.
As our study focuses on consumer acceptance of generic
drugs in substitution of brand name drugs, we measure all
the outcome measures as the within-subject differences in
attitudes towards generic vs. brand name drugs.
Speci?cally, we measure the image of generic drug manu-
facturers as the difference between the answers to the fol-
lowing two questions: ‘‘On a 1–7 scale, how do you view
generic pharmaceutical companies’ image?’’ (1 = very
unfavorable and 7 = very favorable), and ‘‘On a 1–7 scale,
how do you view brand name pharmaceutical companies’
image?’’ (1 = very unfavorable and 7 = very favorable).
We measure the intention to communicate on generic
drugs as the arithmetic average of the two variables
described below. The ?rst variable is the difference
between the answers to the following two questions: ‘‘I
intend to talk to my doctor about my preference for generic
drugs for my next prescription.’’ (1 = strongly disagree and
7 = strongly agree), and ‘‘I intend to talk to my doctor about
my preference for brand name drugs for my next prescrip-
tion.’’ (1 = strongly disagree and 7 = strongly agree). The
second variable is the difference between the answers to
the following two questions: ‘‘I intend to talk to my phar-
macist about my preference for generic drugs for my next
prescription.’’ (1 = strongly disagree and 7 = strongly
agree), and ‘‘I intend to talk to my pharmacist about my
preference for brand name drugs for my next prescription.’’
(1 = strongly disagree and 7 = strongly agree). We measure
the intention to purchase generic drugs as the difference
between the answers to the following two questions: ‘‘I
would rather buy generic than brand name medications
for my next prescription.’’ (1 = strongly disagree and
7 = strongly agree), and ‘‘I would rather buy brand name
than generic medications for my next prescription.’’
(1 = strongly disagree and 7 = strongly agree). The answer
options to all the intention questions are on a seven-point
Table 1
Demographic information.
Variable Value Frequency Percentage
Gender Female 299 53.1
Male 264 46.9
Age 18–29 87 15.5
30–44 142 25.2
45–59 156 27.7
60+ 178 31.6
Race White, Non-Hispanic 411 73
Black, Non-Hispanic 54 9.6
Other, Non-Hispanic 21 3.7
Hispanic 63 11.2
2+Races, Non-
Hispanic
14 2.5
Education Less than high school 56 9.9
High School 167 29.7
Some college 168 29.8
Bachelor’s degree or
higher
172 30.6
Household
income