Exploring Consumers’
Answers to Survey
Questions: Are Uninformed
Responses Truly
Uninformed?
Timothy R. Graeff
Middle Tennessee State University
ABSTRACT
Researchers have observed that consumers often give answers to
questions about which they are uninformed. Drawing from work in
cognitive psychology, this research explores why, how, and when
consumers answer survey questions about which they are
uninformed. Results from a telephone survey of 1348 consumers
suggest that the effects of stimulus factors designed to increase item
response rates (pressure to respond in the introduction to the
survey, absence of a DK option) are moderated by consumers’
familiarity with other similar-sounding attitude objects.
Implications for future research are discussed. 䉷 2003 Wiley
Periodicals, Inc.
Marketers routinely rely on the results from survey research to measure
consumers’ awareness of their brands as well as consumers’ attitudes
and feelings toward their brands. In the area of public policy, surveys
and poll results often guide decisions about important issues such as
taxation, defense, and diplomacy (Norton-Smith, 1994; B. I. Page & Sha-
piro, 1983; Worchester, 1997). Coupled with the increased reliance on
surveys is the recognition of a most troubling form of bias in survey
Psychology & Marketing, Vol. 20(7): 643–667 (July 2003)
Published online in Wiley InterScience (www.interscience.wiley.com)
䉷 2003 Wiley Periodicals, Inc. DOI: 10.1002/mar.10090
643
,research — the uninformed response bias. People often give opinions on
issues about which they are uninformed (Bishop, Oldendick, Tuchfar-
ber, & Bennett, 1980; Gilljam & Granberg, 1993; Hartley, 1946; Schu-
man & Presser, 1981). Survey respondents have freely given opinions
about fictitious governmental agencies and congressional bills (Bishop,
Oldendick, & Tuchfarber, 1983; Bishop, Tuchfarber, & Oldendick, 1986;
Schuman & Presser, 1980). Respondents have also given opinions about
fictitious political figures (Kolson & Green, 1970) and have given direc-
tions to places that do not exist (Collett & O’Shea, 1976). In each case,
those expressing opinions were by definition uninformed because the
issues were fictitious — thus the name uninformed response bias.
The importance of uniformed responses to decision makers is obvious.
If consumers provide answers (opinions) to questions about which they
are uninformed, marketing and business decisions might be based on
meaningless answers to survey questions. The increasing reliance on
the results of survey research to make marketing and public-policy de-
cisions begs for greater understanding of why, how, and when respon-
dents give opinions on issues about which they are uninformed.
An additional problem for marketers is that efforts to increase item
response rates might actually increase uninformed response rates. Un-
informed respondents can be pressured to provide meaningless answers
to survey questions (Converse, 1964, 1970, 1974; Kanuk & Berenson,
1975; Linsky, 1975; Yu & Cooper, 1983). Research demonstrates that
pressuring respondents to answer survey questions increases unin-
formed response rates, and reducing pressure to respond [offering a
“don’t know” (DK) option] reduces uninformed response rates (Graeff,
2002). However, offering a DK option does not eliminate uninformed
responses. Even when uninformed respondents are allowed to honestly
and easily admit that they lack the required knowledge or have no opin-
ion, the majority of respondents still offer an uninformed opinion (Haw-
kins & Coney, 1981).
Are Uninformed Responses Truly Uninformed?
To measure consumers’ uninformed responses, researchers are faced
with a methodological dilemma. It is necessary to use fictitious issues
to measure uninformed responses. The rationale is straightforward (see
Schneider, 1985). If consumers are asked questions about real issues or
brands, there is no way of determining if a response is informed or un-
informed. Thus, when consumers are asked a question about a fictitious
issue, any response other than DK is considered by definition to be un-
informed. Consumers cannot be informed and have knowledge or mem-
ory structures about something that does not exist.
The methodological dilemma is that the fictitious issues must sound
plausible. If not, respondents could easily identify them as fictitious.
For example, researchers have examined uninformed responses to
644 GRAEFF
, seemingly plausible fictitious agencies such as the National Bureau of
Consumer Complaint (Hawkins & Coney, 1981) and fictitious legislation
such as the Public Affairs Act, Agricultural Trade Act, and the Monetary
Control Bill (Bishop, Oldendick, & Tuchfarber, 1983; Bishop, Tuchfar-
ber, & Oldendick, 1986). Consumers have every right to expect that
questions asked of them are about real issues, and these appear to be
real pieces of legislation and governmental agencies. But are consumers’
responses to such seemingly plausible, albeit fictitious, issues truly un-
informed?
Unfortunately, the understanding of uninformed responses has been
limited to the effects of stimulus factors and contextual cues on the
probability of respondents offering uninformed responses (Bradburn &
Sudman, 1988; Gilljam & Granberg, 1993; Krosnick & Alwin, 1987;
Schuman & Presser, 1981). Such stimulus factors include the presence
of a DK option, pressure to respond, and mode of survey contact (phone
versus mail versus personal interview; Ayidiya & McClendon, 1990;
Bishop, Oldendick & Tuchfarber, 1983; Bishop, Oldendick, Tuchfarber
& Bennett, 1980; Bishop, Tuchfarber & Oldendick, 1986; Hawkins &
Coney, 1981; Schneider, 1985).
Previous research has not considered consumers’ level of familiarity
with similar-sounding attitude objects. What has been identified as un-
informed responses might actually consist of responses from consumers
who are unfamiliar with the attitude object (they have never encoun-
tered it, nor do they have any knowledge of it) as well as consumers who
are familiar with similar-sounding attitude objects. Truly unfamiliar
consumers might simply guess in order to respond. Alternatively, con-
sumers who are familiar with similar-sounding attitude objects might
think that they know the answer, but are incorrect in their knowledge
because they are truly mistaken about the object being evaluated. Con-
sumers in previous studies might have based their evaluations of the
fictitious government agencies on inferences drawn from their related
knowledge of similar-sounding real agencies. They might have been
guided by a heuristic that researchers do not ask fictitious questions,
and the name of the attitude object was close enough to confuse them.
Thus, they would not consider themselves to be uninformed. Rather,
their familiarity with a similar-sounding attitude object guided their
survey response.
Knowledge Schemas and Survey Responses
A cognitive explanation of uninformed responses suggests that consum-
ers use knowledge schemas to answer survey questions. A knowledge
schema is a mental structure containing general expectations and
knowledge of the world (Augoustinos & Walker, 1995). Schemas influ-
ence the encoding of new information, memory for old information, and
inferences about missing information (Fiske & Taylor, 1991; Taylor &
UNINFORMED RESPONSES 645
Answers to Survey
Questions: Are Uninformed
Responses Truly
Uninformed?
Timothy R. Graeff
Middle Tennessee State University
ABSTRACT
Researchers have observed that consumers often give answers to
questions about which they are uninformed. Drawing from work in
cognitive psychology, this research explores why, how, and when
consumers answer survey questions about which they are
uninformed. Results from a telephone survey of 1348 consumers
suggest that the effects of stimulus factors designed to increase item
response rates (pressure to respond in the introduction to the
survey, absence of a DK option) are moderated by consumers’
familiarity with other similar-sounding attitude objects.
Implications for future research are discussed. 䉷 2003 Wiley
Periodicals, Inc.
Marketers routinely rely on the results from survey research to measure
consumers’ awareness of their brands as well as consumers’ attitudes
and feelings toward their brands. In the area of public policy, surveys
and poll results often guide decisions about important issues such as
taxation, defense, and diplomacy (Norton-Smith, 1994; B. I. Page & Sha-
piro, 1983; Worchester, 1997). Coupled with the increased reliance on
surveys is the recognition of a most troubling form of bias in survey
Psychology & Marketing, Vol. 20(7): 643–667 (July 2003)
Published online in Wiley InterScience (www.interscience.wiley.com)
䉷 2003 Wiley Periodicals, Inc. DOI: 10.1002/mar.10090
643
,research — the uninformed response bias. People often give opinions on
issues about which they are uninformed (Bishop, Oldendick, Tuchfar-
ber, & Bennett, 1980; Gilljam & Granberg, 1993; Hartley, 1946; Schu-
man & Presser, 1981). Survey respondents have freely given opinions
about fictitious governmental agencies and congressional bills (Bishop,
Oldendick, & Tuchfarber, 1983; Bishop, Tuchfarber, & Oldendick, 1986;
Schuman & Presser, 1980). Respondents have also given opinions about
fictitious political figures (Kolson & Green, 1970) and have given direc-
tions to places that do not exist (Collett & O’Shea, 1976). In each case,
those expressing opinions were by definition uninformed because the
issues were fictitious — thus the name uninformed response bias.
The importance of uniformed responses to decision makers is obvious.
If consumers provide answers (opinions) to questions about which they
are uninformed, marketing and business decisions might be based on
meaningless answers to survey questions. The increasing reliance on
the results of survey research to make marketing and public-policy de-
cisions begs for greater understanding of why, how, and when respon-
dents give opinions on issues about which they are uninformed.
An additional problem for marketers is that efforts to increase item
response rates might actually increase uninformed response rates. Un-
informed respondents can be pressured to provide meaningless answers
to survey questions (Converse, 1964, 1970, 1974; Kanuk & Berenson,
1975; Linsky, 1975; Yu & Cooper, 1983). Research demonstrates that
pressuring respondents to answer survey questions increases unin-
formed response rates, and reducing pressure to respond [offering a
“don’t know” (DK) option] reduces uninformed response rates (Graeff,
2002). However, offering a DK option does not eliminate uninformed
responses. Even when uninformed respondents are allowed to honestly
and easily admit that they lack the required knowledge or have no opin-
ion, the majority of respondents still offer an uninformed opinion (Haw-
kins & Coney, 1981).
Are Uninformed Responses Truly Uninformed?
To measure consumers’ uninformed responses, researchers are faced
with a methodological dilemma. It is necessary to use fictitious issues
to measure uninformed responses. The rationale is straightforward (see
Schneider, 1985). If consumers are asked questions about real issues or
brands, there is no way of determining if a response is informed or un-
informed. Thus, when consumers are asked a question about a fictitious
issue, any response other than DK is considered by definition to be un-
informed. Consumers cannot be informed and have knowledge or mem-
ory structures about something that does not exist.
The methodological dilemma is that the fictitious issues must sound
plausible. If not, respondents could easily identify them as fictitious.
For example, researchers have examined uninformed responses to
644 GRAEFF
, seemingly plausible fictitious agencies such as the National Bureau of
Consumer Complaint (Hawkins & Coney, 1981) and fictitious legislation
such as the Public Affairs Act, Agricultural Trade Act, and the Monetary
Control Bill (Bishop, Oldendick, & Tuchfarber, 1983; Bishop, Tuchfar-
ber, & Oldendick, 1986). Consumers have every right to expect that
questions asked of them are about real issues, and these appear to be
real pieces of legislation and governmental agencies. But are consumers’
responses to such seemingly plausible, albeit fictitious, issues truly un-
informed?
Unfortunately, the understanding of uninformed responses has been
limited to the effects of stimulus factors and contextual cues on the
probability of respondents offering uninformed responses (Bradburn &
Sudman, 1988; Gilljam & Granberg, 1993; Krosnick & Alwin, 1987;
Schuman & Presser, 1981). Such stimulus factors include the presence
of a DK option, pressure to respond, and mode of survey contact (phone
versus mail versus personal interview; Ayidiya & McClendon, 1990;
Bishop, Oldendick & Tuchfarber, 1983; Bishop, Oldendick, Tuchfarber
& Bennett, 1980; Bishop, Tuchfarber & Oldendick, 1986; Hawkins &
Coney, 1981; Schneider, 1985).
Previous research has not considered consumers’ level of familiarity
with similar-sounding attitude objects. What has been identified as un-
informed responses might actually consist of responses from consumers
who are unfamiliar with the attitude object (they have never encoun-
tered it, nor do they have any knowledge of it) as well as consumers who
are familiar with similar-sounding attitude objects. Truly unfamiliar
consumers might simply guess in order to respond. Alternatively, con-
sumers who are familiar with similar-sounding attitude objects might
think that they know the answer, but are incorrect in their knowledge
because they are truly mistaken about the object being evaluated. Con-
sumers in previous studies might have based their evaluations of the
fictitious government agencies on inferences drawn from their related
knowledge of similar-sounding real agencies. They might have been
guided by a heuristic that researchers do not ask fictitious questions,
and the name of the attitude object was close enough to confuse them.
Thus, they would not consider themselves to be uninformed. Rather,
their familiarity with a similar-sounding attitude object guided their
survey response.
Knowledge Schemas and Survey Responses
A cognitive explanation of uninformed responses suggests that consum-
ers use knowledge schemas to answer survey questions. A knowledge
schema is a mental structure containing general expectations and
knowledge of the world (Augoustinos & Walker, 1995). Schemas influ-
ence the encoding of new information, memory for old information, and
inferences about missing information (Fiske & Taylor, 1991; Taylor &
UNINFORMED RESPONSES 645
