Real-time: Cognitive rationale -- 1
Why Researchers Should Think “Real-Time”:
A Cognitive Rationale
University of Michigan
Preliminary draft for M. R. Mehl & T. S. Conner (eds.),
Handbook of Research Methods for Studying Daily Life.
New York: Guilford
Version: 12 Dec 2010
Preparation of this chapter was supported by a fellowship from the Center for Advanced Study
in the Behavioral Sciences and NIH Grant P30 AG024928. Address correspondence to Norbert
Schwarz, ISR, University of Michigan, 426 Thompson St, Ann Arbor, MI 48106-1248, USA;
firstname.lastname@example.org; +1-734-272-4677 voice
Real-time: Cognitive rationale -- 2
When we want to know what people think, feel, and do, we ask them. This reliance on
self-reports is based on the tacit assumption that people know their thoughts, feelings, and
behaviors and can report on them “with candor and accuracy”, as Angus Campbell (1981), a
pioneer of survey research, put it. From this perspective, problems arise when the research
situation discourages candor and accuracy, when the questions are ambiguous and difficult to
understand, or when the task exceeds participants’ knowledge and the limits of memory. A
large methodological literature addresses these concerns and what to do about them (for
reviews see Bradburn, Sudman, & Wansink, 2004; Sudman, Bradburn, & Schwarz, 1996;
Tourangeau, Rips, & Rasinski, 2000). The lessons learned from this work highlight that many
self-report problems can be attenuated by asking questions in close temporal proximity to the
event of interest. Doing so constrains the multiple meanings of questions, reduces memory and
estimation problems, and facilitates access to episodic detail, all of which can improve self-
report. The real-time or close-in-time measures discussed in this handbook take advantage of
However, these (largely uncontroversial) methodological issues are only some of the
reasons why researchers should think real-time. At a more fundamental level, recent research
across many areas of psychological science highlights that every aspect of human cognition,
emotion, motivation, and behavior is situated and highly context-sensitive, thwarting attempts
to understand it in a decontextualized way (see the contributions in Mesquita, Barrett, & Smith,
2010). As this work progresses, it becomes increasingly clear that our methods should
acknowledge this insight. They rarely do. This issue goes beyond the familiar methodological
questions of “How to ask about X” and presents a fundamental (and controversial) challenge to
bring our empirical operations into line with our theoretical assumptions. Studying
psychological phenomena in the context of daily life can make important contributions to this
development by shedding new light on the situated and embedded nature of human behavior
This chapter elaborates on these themes. The first section summarizes basic insights
into how respondents answer questions and sets the stage for later sections. To date, research
into the cognitive and communicative processes underlying self-reports has rarely addressed
real-time (or close-in-time) measurement, which presents its own set of self-report problems. I
draw attention to some of them and offer pertinent conjectures. The second section addresses
reports of past behavior and reviews issues of autobiographical memory, highlighting the role
of inference strategies and lay theories in determining what “must have been” (Ross, 1989). It
pays particular attention to what respondents can, or cannot, report on with some accuracy.
The third section turns to reports of emotions and physical symptoms. It compares
prospective reports of expected future feelings and retrospective reports of past feelings with
concurrent reports of momentary experience. Of particular interest are systematic
Real-time: Cognitive rationale -- 3
convergences and divergences between these reports. On the one hand, predicted feelings
usually converge with remembered feelings and the behavioral choices people make; on the
other hand, all of these variables are often poorly related to actual experience as assessed by
real-time measures (Schwarz, Kahneman, & Xu, 2009). These dynamics illustrate that feelings
are fleeting and poorly represented in memory (Robinson & Clore, 2002); once they dissipated,
respondents need to reconstruct what their feeling may have been. Shortly after the
experience, episodic reconstruction can result in relatively accurate reports, as indicated by
convergence with concurrent assessments (Kahneman, Krueger, Schkade, Schwarz, & Stone,
2004). But as time passes, respondents resort to general knowledge to infer the past
experience, which is also the knowledge used for predicting future feelings; these predictions,
in turn, are the basis for intention and choice (Would this be good for me?). Hence, prediction,
intention, choice, and later global memories converge because they are based on similar inputs
– and this convergence seems to confirm that one’s predictions and choices were right all
along. Unfortunately, concurrent measures often suggest otherwise but this lesson is missed
with the fading feeling (Schwarz et al., 2009). These dynamics impair learning from daily
experience and challenge researchers’ reliance on the consistency of respondents’ reports as an
indicator of validity.
The final section turns to reports of attitudes and preferences. It reviews the promises
and pitfalls of the traditional conceptualization of attitudes as enduring dispositions and notes
the malleable nature of attitude reports. Whereas this malleability is usually considered
deplorable measurement error, a situated cognition approach suggests that it may reflect
something more laudable and adaptive. If evaluation stands in the service of current action, it is
likely to benefit from sensitivity to one’s current goals and close attention to the affordances
and constraints of one’s current context (Schwarz, 2007). From this perspective, the context
“dependency” that frustrates observers and researchers, who both want to predict an actor’s
behavior, reflects an adaptive context “sensitivity” that may serve the actor well. Real-time
measurement in situ can shed new light on the underlying dynamics, in particular when it
adopts the actor’s rather than the observer’s perspective.
ANSWERING QUESTIONS: THE PSYCHOLOGY OF SELF-REPORT
Answering a question in a research context requires that respondents (1) interpret the
question to understand what the researcher wants to know and (2) retrieve and select relevant
information to (3) form an answer. In most cases, they cannot provide an answer in their own
words but (4) need to map it onto a set of response alternatives provided by the researcher.
Finally, (5) respondents may wish to "edit" their answer before they communicate it for reasons
of social desirability and self-presentation. Respondents' performance at each of these steps is
context sensitive and profoundly influenced by characteristics of the research setting and
instrument. Extensive reviews of these issues are available (Schwarz, Knäuper, Oyserman, &
Real-time: Cognitive rationale -- 4
Stich, 2008; Sudman, Bradburn, & Schwarz, 1996; Tourangeau, Rips, & Rasinski, 2000); I
summarize key points and draw attention to some implications for real-time measurement.
The key issue at the question comprehension stage is whether respondents'
understanding of the question matches the meaning the researcher had in mind. As all
textbooks note, writing simple questions and avoiding unfamiliar or ambiguous terms helps
(see Bradburn, Sudman & Wansink, 2004, for good practical advice). But ensuring that
respondents understand the words is not enough. When asked, "What have you done today?"
respondents will understand the words – but they still need to determine what kind of activities
the researcher is interested in. Should they report, for example, that they took a shower, or
not? Providing an informative answer requires inferences about the questioner's intention to
determine the pragmatic meaning of the question (Clark & Schober, 1992; Schwarz, 1996).
Question context and order
To infer the pragmatic meaning, respondents draw on contextual information, from the
purpose of the study and the researcher’s affiliation to the content of adjacent questions and
the nature of the response alternatives. Their use of this information is licensed by the tacit
assumptions that govern the conduct of conversation in daily life (Grice, 1975), which
respondents bring to the research situation (for reviews see Schwarz, 1994, 1996). Hence, they
interpret a given question in the thematic context of the overall interview and a term like
"drugs" acquires different meanings when presented in a survey pertaining to respondents'
medical history rather than to crime in the neighborhood. Similarly, they attend to the
researchers’ affiliation to infer the likely epistemic interest behind their questions. Taking this
interest into account, their explanations emphasize personality variables when asked by a
personality psychologist, but social context variables when asked by a social scientist
(Norenzayan & Schwarz, 1999). Respondents further assume that adjacent questions are
meaningfully related to one another, unless otherwise indicated, and interpret their intended
meaning accordingly (e.g., Strack, Schwarz, & Wänke, 1991).
When the data collection method enforces a strict sequence, as is the case for personal
and telephone interviews and computer administered questionnaires that do not allow
respondents to return to earlier questions, preceding questions can influence the interpretation
of subsequent questions but not vice versa. In contrast, preceding as well as following
questions can exert an influence when respondents can become aware of all questions prior to
answering them, as is the case for paper-and-pencil questionnaires and computer programs
without strict sequencing (Schwarz & Hippler, 1995). Most real-time studies probably fall into
the latter category, given that they repeat a small number of questions with high frequency,
thus allowing respondents to know what’s coming even when the instrument enforces a strict
Real-time: Cognitive rationale -- 5
The maxims of cooperative conversational conduct further ask speakers to provide
information the recipient needs and not to reiterate information the recipient already has
(Grice, 1975). Respondents observe this norm and hesitate to reiterate information they have
already provided in response to an earlier question (for a review see Schwarz, 1996). For
example, Strack and colleagues (1991) observed a correlation of r = .95 when respondents were
asked to report their overall happiness and their overall life-satisfaction in two separate
questionnaires, attributed to different researchers. However, the correlation dropped to r = .75
when the same two questions were presented in the same questionnaire, attributed to the
same researcher. In everyday discourse, the same questioner would not request the same
information twice, in somewhat different words; hence, respondents differentiate between
similar questions when they are presented by the same researcher. Two different researchers,
on the other hand, may very well ask the same thing in different words, so identical answers
Note that the repetition of very similar, if not identical, questions is a key feature of
many real-time measurement procedures. At present, we do not know how this affects
respondents’ question interpretation. Do respondents hesitate to repeat information at 4:05pm
that they already provided at 3:40pm? If they hesitate to provide the same answer, does their
attempt to provide new information increase meaningful differentiation between episodes or
does it foster differentiations that go beyond respondents’ actual experience in situ?
Formal characteristics of questions
From a conversational perspective, every contribution is assumed to be related to the
ongoing conversation, unless marked otherwise. In research settings, the researcher’s
contributions include formal characteristics of the question, which respondents use in inferring
the question’s pragmatic meaning (Schwarz, 1994, 1996). Suppose, for example, that
respondents are asked how frequently they felt "really irritated" recently. Does this question
refer to major or minor annoyances? The numeric values of the frequency scale provide
relevant information. When the scale presents low frequencies respondents infer that the
researcher is interested in less frequent events than when it presents high frequencies; as a
result they report on major annoyances (which are relatively rare) in the former, but on minor
annoyances in the latter case (Schwarz, Strack, Müller, & Chassein, 1988). The same logic
applies to the length of reference periods (Winkielman, Knäuper, & Schwarz, 1998). Given that
major annoyances are less frequent than minor annoyances, respondents infer that the
question pertains to minor irritations when it is presented with a short reference period (e.g.,
“yesterday”), but to major annoyances when presented with a long reference period (e.g., “last
six months”). Accordingly, questions with reference periods of differential length assess
substantively different experiences, e.g., “minor” rather than “major” episodes of anger.
This has potentially important implications for real-time measurement, which usually
includes very short and recent reference periods. When asked at 4:05pm how often they have
Real-time: Cognitive rationale -- 6
been angry since the last measurement at 3:40pm, respondents may report on very minor
episodes, which they would not consider worth mentioning for any longer reference period.
Moreover, once they assume that this is what the questioner has in mind, they may evaluate
each minor episode relative to other minor episodes. Consistent with this shift in the frame of
reference, they may then assign each minor episode a high intensity rating, leading the
researcher to conclude that intense anger is very frequent. To date, these possibilities have not
been addressed and little is known about the potential impact of high density measurement on
Recall and Judgment
Once respondents determined what the researcher is interested in, they need to recall
relevant information to form a judgment. In some cases, they may have direct access to a
previously formed relevant judgment that they can offer as an answer. More likely, however,
they will need to form a judgment when asked, taking the specifics of the question and the
questioner’s inferred epistemic interest into account. The processes pertaining to different
types of reports are discussed in the sections on behaviors, feelings, and attitudes.
Formatting the Response
Unless the question is asked in an open response format, respondents need to format
their answer to fit the response alternatives provided by the researcher (for a review see
Schwarz & Hippler, 1991; Sudman et al., 1996). Respondents observe these question
constraints and avoid answers that are not explicitly offered. Moreover, their selection of
response alternatives is influenced by the order in which they are presented. In most cases, a
given response alternative is more likely to be chosen when presented early rather than late on
the list under visual presentation conditions, reflecting the sequence of reading. Conversely, a
given alternative is more likely to chosen when presented late rather than early on the list
under auditory presentation conditions; respondents need to wait for the interviewer to finish
reading the list and work backward, beginning with the last alternative heard (Krosnick & Alwin,
1985; Sudman et al., 1996, chapter 6). This suggests that real-time data capture through
Interactive Voice Responding, where the response alternatives are presented auditorily, may
facilitate the emergence of recency effects, whereas the visual presentation formats typical for
ESM and daily diaries may facilitate primacy effects.
Finally, respondents’ use of rating scales reflects two regularities familiar from
psychophysical research; both have been conceptualized in Parducci’s (1965) range-frequency
theory (see Daamen & de Bie, 1992, for social science examples). First, respondents use the
most extreme stimuli to anchor the endpoints of the scale. Accordingly, they will rate a given
episode of anger as less intense when the high end of the scale is anchored by an extreme
rather than a moderate anger episode. This has important implications for the comparability of
Real-time: Cognitive rationale -- 7
retrospective and real-time reports. When asked to rate a single past episode, the recalled
episode is likely to be compared to other memorable instances – which are often memorable
because they were extreme. But when asked to rate multiple episodes over the course of a
single day, previously rated moderate episodes may still be highly accessible. Hence, the same
episode of anger may be rated as more extreme in real-time than in retrospective reports,
reflecting the use of differentially extreme scale anchors and comparison standards.
Second, psychophysical research further shows that respondents attempt to use all
categories of the rating scale about equally often when the number of to-be-rated stimuli is
large. Hence, two similar stimuli may receive notably different ratings when only a few stimuli
are presented, but identical ratings when many stimuli have to be located along the same scale.
In many real-time studies, respondents are asked to rate a large number of episodes along
identical scales over the course of a few hours, which is likely to elicit similar shifts in ratings.
Both of these regularities predict systematic differences between retrospective and concurrent
ratings as well as between concurrent ratings assessed with differential frequency. Future
research may fruitfully test this possibility.
Editing the Response: Social Desirability
As the final step of the question answering sequence, respondents have to
communicate their answer. Due to social desirability and self-presentation concerns they may
edit their response (see DeMaio, 1984, for a review). This is more likely in face-to-face
interviews than under the more confidential conditions of self-administered questionnaires,
with telephone interviews falling in between. This is good news for real-time data capture,
which predominantly relies on self-administered formats.
The literature further indicates that influences of social desirability are limited to
potentially threatening questions and typically modest in size (DeMaio, 1984). Note, however,
that a behavior that may seem only mildly unfavorable when reported once for a single specific
episode (e.g., “I don’t enjoy being with my spouse right now”) may become a major self-
presentation concern when the same answer would need to be provided over several episodes.
If so, high density measurement in real-time studies may accentuate self-presentation concerns
relative to retrospective reporting conditions though the cumulative impact of social
desirability concerns over multiple similar episodes. Finally, respondents’ self-presentation
concerns can be reliably reduced through techniques that ensure the anonymity and
confidentiality of the answer (see Bradburn et al., 2004, for detailed advice).
REPORTING ON BEHAVIORS
This section focuses on the recall stage of the question answering process and highlights
what respondents can and cannot remember and report. It is organized by the type of
information the researcher wants to assess.
Real-time: Cognitive rationale -- 8
Some questions pertain to historical information. Examples include, Have you ever had
an episode of heart burn? In what year did you first have an episode of heart burn?
Respondents’ memories are usually the only available source of information and real-time
measurement is not feasible. The best a researcher can do is to use interviewing techniques
that take the structure of autobiographical memory into account to facilitate recall (for advice
see Belli, 1998; Schwarz & Oyserman, 2001; Tourangeau, et al., 2000).
Current models of autobiographical memory conceptualize it as a hierarchical network
that includes extended periods (e.g., “the years I lived in New York”) at the highest level of the
hierarchy. Nested within each extended period are lower-level extended events (e.g., “my first
job” or “the time I was married to Lucy”). Further down the hierarchy are summarized events,
which take the form of knowledge-like representations that lack episodic detail (e.g., “During
that time, I was frequently ill.”). Specific events, like a particular episode of illness, are
represented at the lowest level of the hierarchy; to be represented at this level of specificity,
the event has to be unique. As Belli (1998, p. 383) notes, this network, organized by time (“the
years in New York”) and relatively global themes (“first job;” “first marriage;” “illness”), enables
“the retrieval of past events through multiple pathways that work top-down in the hierarchy,
sequentially within life themes that unify extended events, and in parallel across life themes
that involve contemporaneous and sequential events.” Such searches take time and their
outcome is somewhat haphazard, depending on the entry point into the network at which the
search started. Building on these insights, Event History Calendars improve recall by using
multiple entry points and forming connections across different periods and themes (see Belli,
1998, for a review and detailed advice).
In the absence of such (costly) efforts, respondents are likely to apply extensive
inference strategies to the few bits and pieces they remember to infer what “must have” been
(Ross, 1989). Suppose, for example, that respondents are asked how much alcohol they drank
five years ago. Having no direct access to this information, they are likely to consider their
current alcohol consumption as a benchmark and to make adjustments if they see a need to do
so. In most cases, their adjustments are insufficient because people assume an unrealistically high
degree of stability in their behavior. This results in retrospective reports that are more similar to
the present than is warranted, as observed for reports of income (Withey, 1954), pain (Eich et al.,
1985) or tobacco, marijuana, and alcohol consumption (Collins, Graham, Hansen, & Johnson,
1985). However, when respondents have reason to believe things were different in the past, they
will “remember” change (Ross, 1998), as discussed next.
Reports of Change, Covariation, and Causation
Some questions go beyond mere retrospective reports and ask respondents to report on
change over time (Do you smoke more or less now than you did when you were 30?) or to
Real-time: Cognitive rationale -- 9
assess the covariation of their behavior with other variables (Do you smoke more when you are
stressed?). Respondents can rarely retrieve the information that would be needed to answer
such questions and rely on extensive inference and estimation strategies to determine what
might have been. Their answers are useful to the extent that the underlying lay theories
happen to be correct, which is usually unknown.
Although most people assume an unwarranted amount of stability in their behavior,
they will readily detect change when their lay theory suggests that change must have occurred.
This is particularly likely --and problematic-- when the context suggests change, as is often the
case in medical studies: Believing that things get better with treatment (or why else would one
undergo it?), patients are likely to infer that their past condition must have been worse than
their present condition (e.g., Linton & Melin, 1982; for a review see Ross, 1989). From a
cognitive perspective, asking patients whether they feel better now than before their treatment
is the most efficient way to “improve” the success rate of medical interventions, which may
explain the recent popularity of “patient reported outcomes”. Unfortunately, there is no
substitute for appropriate study design. If change over time is of crucial interest, concurrent
measures at different points in time are the only reliable way to assess it.
Similar problems arise when respondents are asked to report on covariation (Under
which circumstances…?) or causation (Why…?). To arrive at an observation-based answer to
these questions, respondents would need to have an accurate representation of the frequency
of their behaviors, the different contexts of these behaviors, and the intensity of related
experiences. Respondents are often unable to provide accurate reports on any of these
components, making their joint consideration an unrealistically complex task.
Covariation and causation are best assessed with real-time data capture. Experience
sampling methods excel at this task by prompting respondents to report on their behavior,
experiences, and circumstances, allowing researchers to collect all the data needed for
appropriate analyses. However, an important caveat needs attention. While real-time or close-
in-time measures improve the accurate assessment of covariation, causation, and change,
respondents’ own behavioral decisions are based on their own perceptions, which may differ
from reality. Hence, erroneous lay theories of covariation are often better predictors of
behavior than accurate measures of covariation, as reviewed in the section on feelings.
Frequency questions ask respondents to report on the frequency of a behavior or
experience during a specified reference period, often last week or last month. Researchers
typically hope that respondents will identify the behavior of interest, search the reference
period, retrieve all instances that match the target behavior, and finally count these instances
to determine the overall frequency of the behavior. However, such a recall-and-count strategy
is rarely feasible. Respondents usually need to rely on extensive inference and estimation
Real-time: Cognitive rationale -- 10
strategies to arrive at an answer; which strategy they use depends on the frequency,
importance, and regularity of the behavior (e.g., Brown, 2002; Menon, 1993, 1994; Sudman et
Questions about rare and important behaviors can be answered on the basis of
autobiographical knowledge or a recall-and-count strategy. When asked “How often did you get
divorced?” most people know the answer without extended memory search. When asked “How
often did you relocate to another city?” many people will not know immediately, but can
compute an appropriate answer by reviewing their educational and job history, following a
recall-and-count strategy. Respondents’ task is more demanding when the behavior is
frequent. High frequency of a behavior makes it unlikely that detailed representations of
numerous individual episodes are stored in memory; instead, different instances blend into one
global, knowledge-like representation that lacks specific time or location markers (see Linton,
1982; Strube, 1987). Frequent doctor visits, for example, result in a well-developed knowledge
structure for the general event, allowing respondents to report in considerable detail on what
usually goes on during their doctor visits. But the highly similar individual episodes become
indistinguishable and irretrievable, making it difficult to report on any specific one. In these cases,
respondents need to resort to estimation strategies to arrive at a plausible frequency report.
Which estimation strategy they use depends on the regularity of the behavior and the context in
which the frequency question is presented.
When the behavior is highly regular, frequency estimates can be computed on the basis
of rate information (Menon, 1994; Menon, Raghubir, & Schwarz, 1995). Respondents who go to
church every Sunday have little difficulty in arriving at a weekly or monthly estimate. However,
exceptions are likely to be missed and the estimates are only accurate when exceptions are
rare. A related strategy relies on extrapolation from partial recall. When asked how often she
took pain medication during the last week, for example, a respondent may reason, “I took pain
killers three times today, but this was a bad day. So probably twice a day, times 7 days, makes
14 times a week.” The accuracy of this estimate will depend on the accuracy of the underlying
assumptions, the regularity of the behavior, and the day that served as input into the chain of
Other estimation strategies may even bypass any effort to recall specific episodes. For
example, respondents may simply rely on information provided by the research instrument
itself. As an example, consider the frequency scales shown in Table 1. Consistent with the
maxims of cooperative conversational conduct (Grice, 1975) respondents assume that the
researcher constructed a meaningful scale that is relevant to their task (Schwarz, 1996).
Presumably, the range of response alternatives reflects the researcher's knowledge about the
distribution of the behavior, with values in the middle range of the scale corresponding to the
"usual" or "average" behavior and values at the extremes of the scale corresponding to the
extremes of the distribution. Drawing on these assumptions, respondents use the range of the
Real-time: Cognitive rationale -- 11
response alternatives as a frame of reference in estimating their own behavioral frequency. This
results in higher frequency estimates when the scale presents high rather than low frequency
response alternatives, as Table 1 illustrates.
Table 1 about here
Such scale-based estimation effects have been observed for a wide range of behaviors (for
a review see Schwarz, 1996); they are more pronounced, the more poorly the respective behavior
is represented in memory (Menon, Raghubir, & Schwarz, 1995). When behaviors of differential
memorability are assessed, this can either exaggerate or cloud actual differences in the relative
frequency of the behaviors, undermining comparisons across behaviors. Moreover, respondents
with poorer memory are more likely to be influenced by frequency scales than respondents with
better memory (e.g., Knäuper, Schwarz, & Park, 2004), which can undermine comparisons across
groups. Finally, frequency scales also invite systematic underestimates of the variance in
behavioral frequencies because all respondents draw on the same frame of reference in
computing an estimate, resulting in reports that are more similar than reality warrants.
SELF-REPORTS OF FEELINGS
Feelings are subjective phenomena to which the person who has them has privileged
access. While this does not imply that feelings are always easy to identify for the experiencer
(see Clore, Conway, & Schwarz, 1994; Ellsworth & Scherer, 2003, for a discussion of different
types of feelings and the underlying appraisal processes), most researchers consider the
experiencer the final arbiter of what his or her feeling is. Unfortunately, that final arbiter is
likely to tell us different things at different points in time and numerous studies documented
profound discrepancies between people’s concurrent and retrospective reports of emotions
(for a comprehensive review see Robinson and Clore, 2002). This section reviews why this is the
case, presents some illustrative biases, and highlights distinct patterns of convergence and
divergence between prospective, concurrent, and retrospective reports as well as the choices
people make (for further discussion of emotion measurement see Larsen & Augustine, this
Accessibility Model of Emotion Report
To conceptualize the processes underlying emotion reports, Robinson and Clore (2002)
proposed an accessibility model. When people report on their current feelings, the feelings
themselves are accessible to introspection, allowing for accurate reports on the basis of
experiential information. But affective experiences are fleeting and not available to
introspection once the feeling dissipated. Accordingly, the opportunity to collect emotion
reports that are based on introspective access is limited to methods of real-time data capture,
like experience sampling (Stone et al., 1999; see also Larsen & Augustine, this volume). Once
the feeling dissipated, the affective experiences need to be reconstructed on the basis of
Real-time: Cognitive rationale -- 12
episodic or semantic information. When the report pertains to a specific recent episode, people
can draw on episodic memory, retrieving specific moments and details of the recent past.
Detailed episodic recall can often re-elicit a similar feeling (and is therefore a popular mood
manipulation); it can also provide sufficient material for relatively accurate reconstruction.
Hence, episodic reports often recover the actual experience with some accuracy, as indicated
by convergence with concurrent reports (e.g., Kahneman, Krueger, Schkade, Schwarz, & Stone,
2004; Robinson & Clore 2002; Stone, Schwartz, Schwarz, Schkade, Krueger, & Kahneman, 2006).
One method that facilitates episodic reporting is the Day Reconstruction Method (DRM;
Kahneman et al., 2004), discussed below. At present, it remains unclear how far in the past an
episode can be to still allow reasonably accurate episodic reconstruction. Most likely the
answer depends on the uniqueness and memorability of the episode, paralleling the above
discussion of behavioral frequency reports.
In contrast episodic reports, global reports of past feelings are based on semantic
knowledge. When asked how they “usually” feel during a particular activity, people draw on
their general beliefs about the activity and its attributes to arrive at a report. The actual
experience does not figure prominently in global reports because the experience itself is no
longer accessible to introspection and episodic reconstruction is not used to answer a global
Extending this accessibility model of emotion report, Schwarz, Kahneman and Xu (2009;
Xu & Schwarz, 2009) noted that the same semantic knowledge serves as a basis for predicting
future feelings, for which episodic information is not available to begin with. Such predictions
are usually more extreme than people’s actual experiences (for a review see Wilson & Gilbert,
2003) because the predictor focuses on core attributes of the activity at the expense of other
information, resulting in a “focusing illusion” (Schkade & Kahneman, 1997). For example,
Midwesterners who predict how happy they would be if they moved to California may focus on
the pleasant Californian climate, missing, for example, that they would still have to spend most
of the day in an office cubicle. Finally, hedonic predictions play an important role in people’s
daily lives because they serve as input into choice (March, 1978; Mellers & McGraw, 2001) and
influence which course of action people will or will not take.
Convergences and Divergences
The above rationale predicts a systematic pattern of convergences and divergences,
which results directly from the inputs on which the respective reports are based. First,
concurrent reports and retrospective reports pertaining to a specific and recent episode are
likely to show good convergence, provided that the episode is sufficiently recent to allow
detailed and vivid reinstantiation in episodic memory. Second, retrospective global reports of
past feelings and predictions of future feelings are also likely to converge, given that both are
based on the same semantic inputs. Third, choices are based on predicted hedonic
Real-time: Cognitive rationale -- 13
consequences, and hence converge with predictions and global memories. One unfortunate
side-effect of these convergences is that people’s global memories seem to “confirm” the
accuracy of their predictions and the wisdom of their choices, thus impairing the opportunity to
learn from experience (Schwarz & Xu, in press). However, fourth, concurrent and episodic
reports will often diverge from prediction, choice, and global memories. As a result, different
measures can paint very different pictures of a person’s affective experience with the same
situation, as a few examples may illustrate (see Schwarz et al., 2009, for a review).
How enjoyable are vacations?
Not surprisingly, people believe that vacations are very enjoyable and this belief shapes
their predictions, choices, and global memories, even when their actual recent experience was
less rosy. Assessing prospective, concurrent, and retrospective reports of vacation enjoyment,
Mitchell and colleagues (1997) found that prospective reports converged with retrospective
reports; however, both the predicted and remembered affect was more positive than the affect
reported concurrently during the vacation. In a later study, Wirtz and colleagues (2003) tracked
college students before, during, and after their spring-break vacations and compared their
predicted, concurrent, and remembered affect. They found that predicted and remembered
experiences were more intense (i.e., both more positive and more negative) than reported
concurrently during the vacation. However, the (biased) remembered experience predicted the
desire to repeat the vacation better than the actual experience, illustrating that we learn from
our memories, not from our experiences.
How bad was that colonoscopy?
Particularly memorable examples of learning from memory rather than experience have
been reported in the medical domain. For example, Redelmeier and Kahneman (1996) observed
that retrospective evaluations of pain are dominated by two moments that may be of particular
adaptive relevance (Fredrickson, 2000): the peak (“How bad does it get?”) and the end (“How
does it end?”). Other aspects, like the overall duration of pain, exert little influence. In fact,
extending the duration of a colonoscopy by adding a few moments of discomfort at the end
improves the overall evaluation of the episode by adding a better ending. It also improves the
likelihood of future compliance, again highlighting how memory beats experience in predicting
future behavior (Redelmeier, Katz, & Kahneman, 2003).
How much do parents enjoy spending time with their children?
Several decades ago, Juster and colleagues (1975) asked a representative sample of
Americans to rate 28 activities from "dislike very much" (0) to "enjoy a great deal" (10). They
found that activities with one's children consistently topped the list (ranks 1-4), whereas
grocery shopping and cleaning the house were reported as least enjoyable (rank 27 and 28;
Juster, 1985, p.336). In stark contrast to these reports, other studies indicate that parents’
marital satisfaction drops when children arrive, reaches a life-time low when the children are
teenagers, and recovers after the children leave the house (for a review see Argyle, 1999). Are
Real-time: Cognitive rationale -- 14
the children a more mixed pleasure than global reports of enjoyment convey? Close-in-time
measures of affective experience, collected with the Day Reconstruction Method, suggest so.
Specifically, 909 employed women in Texas recalled their activities during the preceding day
and reported how they felt during each specific episode (Kahneman et al., 2004). In these
episodic reports, activities coded as “taking care of my children” ranked just above the least
enjoyable activities of the day, namely working, housework, and commuting; data from other
samples replicated this pattern.
Several processes contribute to this divergence between global and episodic reports.
First, global judgments of enjoyment are based on general beliefs ("I enjoy my kids"), which are
often supported by belief-consistent memories of great vividness (like fond memories of shared
activities). Yet most mundane episodes of a given day are less enjoyable than the episodes that
make for fond memories. Second, activities are organized in memory by their focal features.
Attempts to recall memories pertaining to one’s interactions with one’s children will therefore
result in an overrepresentation of child focused activities, at the expense of numerous other
episodes of the day in which the children were present. The reconstruction of a whole day in
the DRM avoids many of these problems of selective recall and provides a fuller assessment of
the affective impact of children throughout the day. Hence, the findings suggest that part of the
reason that children seem more enjoyable in general than on any given day is simply that
parents do not consider the full range of child-related time use when providing global reports.
Finally, global reports are subject to higher social desirability pressures than episodic reports. A
parent who reports, “I don’t enjoy spending time with my children” is clearly a bad parent; but
noting that “They were a pain last night” is perfectly legitimate.
Several methodological implications are worth emphasizing. Researchers who want to
assess peoples’ actual hedonic experiences should preferably do so with concurrent reports,
using experience sampling methodologies (Stone et al., 1999). If this is not feasible, episodic
reporting methods, like the Day Reconstruction Method (Kahneman et al., 2004), provide a less
burdensome alternative that can capture the experience with some accuracy, provided that the
relevant episodes are recent. In contrast, global reports of affect are theory-driven, not
experience-driven. They capture respondents’ beliefs about their experience rather than the
experience itself and are subject to pronounced focusing effects.
However, people’s behavioral choices are based on their expected hedonic
consequences (March, 1978). These expectations converge with global memories, but often
diverge from actual experience. Hence, predictions, choices, and global memories are poor
indicators of experience. Yet when people make behavioral decisions, global memories and
expectations are likely to figure prominently in the information they consider. Ironically, this
turns faulty indicators of experience into good predictors of future choices and behaviors (e.g.,
Wirtz et al., 2003). It also suggests that optimizing a study for accurate description of what
Real-time: Cognitive rationale -- 15
people do and feel does not optimize it for accurate prediction of what they will do next (and
vice versa) – description and prediction are different goals and their optimization requires
An Example of Episodic Reconstruction: The Day Reconstruction Method
The Day Reconstruction Method (Kahneman et al., 2004) is designed to collect data that
describe a person’s time use and affect on a given day through a systematic reconstruction
conducted on the following day. In a self-administered questionnaire, respondents first
reinstantiate the previous day into working memory by producing a short diary consisting of a
sequence of episodes, usually covering the time from when they got up to when they went to
bed. The diary’s format draws on insights from cognitive research with Event History Calendars
(Belli, 1998) and facilitates retrieval from autobiographical memory through multiple pathways.
Its episodic reinstantiation format attenuates biases commonly observed in retrospective
reports (Robinson & Clore, 2002; Schwarz & Sudman, 1994). Respondents’ diary entries are
confidential and the diary is not returned to the researcher, which allows respondents to use
idiosyncratic notes, including details they may not want to share.
Next, respondents draw on their diary to answer a series of questions about each
episode, including (1) when the episode began and ended, thus providing time use data; (2)
what they were doing; (3) where they were; (4) whom they were interacting with; and (5) how
they felt, assessed on multiple affect dimensions. The details of this response form can be
tailored to the specific issues under study; only this form is returned to the researcher for
analysis. For methodological reasons, it is important that respondents complete the diary
before they are aware of the specific content of the later questions about each episode. Early
knowledge of these questions may affect the reconstruction of the previous day and may
introduce selection bias. The DRM can be administered individually or in group settings and
respondents can report on a complete day in 45 to 75 minutes. DRM reports have been
validated against experience sampling data and Krueger, Kahneman, Schkade, Schwarz, and
Stone (2009) provide a comprehensive review of the methodology and available findings.
SELF-REPORTS OF ATTITUDES: EVALUATION IN CONTEXT
Another common type of self-report question asks people to report on their likes and
dislikes. Psychologists commonly assume that these reports reflect a predisposition to evaluate
some object in a favorable or unfavorable manner; this predisposition is referred to as an attitude
(Eagly & Chaiken, 1993, 2005). Attitudes are hypothetical constructs that cannot be directly
observed and need to be inferred from individuals' responses to the attitude object. As Gordon
Allport (1935, p. 836) put it, “How does one know that attitudes exist at all? Only by necessary
inference. There must be something to account for the consistency of conduct” (italics added).
From this perspective, it is not surprising that attitude questions are often asked without reference
Real-time: Cognitive rationale -- 16
to any specific context – what makes the construct appealing is exactly the promise of predictive
power across contexts. Empirically, attitude research never delivered on this promise. In an early
review of attitude-behavior consistency, Wicker (1969, p. 65) concluded that “only rarely can as
much as 10% of the variance in overt behavioral measures be accounted for by attitudinal data.”
Even the attitude reports themselves proved highly malleable and minor variations in question
wording, question order or response format can elicit profound shifts in reported attitudes, even
on familiar and important topics (for early examples see Cantril, 1944; Payne, 1951; for reviews
see Schwarz, 1999; Schwarz, Groves, & Schuman, 1998; Schuman & Presser, 1981). Attempts to
overcome these disappointments took one of two general paths; one focused on improving
measurement of the attitude itself and the other on improving the predictive power of the
attitude measure by taking context variables into account.
Stalking the “True” Attitude
Mirroring Campbell’s (1981) convictions, many researchers assumed that context effects
on attitude reports and low attitude-behavior consistency can be traced to participants’ hesitation
to report their true feelings “with candor and accuracy”. This focused efforts on attempts to
reduce respondents’ self-presentation concerns (e.g., techniques that ensure respondents’
anonymity and the confidentiality of their answers; see Bradburn et al., 2004, for
recommendations) or to convince them that “lying” was futile – thanks to sophisticated
machinery, the researcher would learn their “true” attitude anyway (e.g., Jones and Sigall’s, 1971,
“bogus pipeline”). Empirically, such techniques have been found to increase the frequency of
socially undesirable answers. For example, people are more likely to admit that they enjoy
pornography when they cannot be identified as the source of the answer (Himmelfarb &
Lickteig, 1982) and White participants are more likely to report that they dislike African
Americans under bogus pipeline conditions (e.g., Allen, 1975). However, external validation of
the reports is not available and the procedures themselves may invite correction of one’s
spontaneous answer in light of the concern about bias that is clearly conveyed.
Whereas these developments assumed that people know their own attitudes but may
not want to report them, later developments considered the possibility that people may
sometimes not be aware of their own attitudes or may not even want to “admit” them to
themselves. Implicit measures of attitudes address this possibility (for overviews see the
contributions in Wittenbrink & Schwarz, 2007). These procedures range from evaluative and
conceptual priming techniques (for a review see Wittenbrink, 2007) and response competition
procedures (e.g., the IAT; for a review see Lane, Banaji, Nosek, & Greenwald, 2007) to low-tech
paper-and-pencil measures (e.g., word completion tasks; for a review see Vargas,
Sekaquaptewa, & von Hippel, 2007). To many researchers’ disappointment, implicit measures
did not deliver the robust, context-independent assessment of attitudes that theorists have
long hoped for. To the contrary, implicit measures of attitudes are subject to the same context
Real-time: Cognitive rationale -- 17
effects that have been observed with explicit self-reports (for extensive reviews see Blair, 2002;
Ferguson & Bargh, 2007). For example, Dasgupta and Greenwald (2001) found that exposure to
pictures of liked African Americans and disliked European Americans resulted in shifts on a
subsequent IAT that paralleled previously observed effects of exposure to liked or disliked
exemplars on explicit measures of attitudes (e.g., Bodenhausen, Schwarz, Bless, & Wänke,
1995). Similarly, Wittenbrink, Judd, and Park (2001) found that the same Black face primes
elicited more negative automatic responses when the faces were presented on the background
of an urban street scene rather than a church scene. Moreover, automatic evaluations have
also been obtained for novel objects, for which no previously acquired object-attitude links
could have been stored in memory (e.g., Duckworth, Bargh, Garcia, & Chaiken, 2002).
Such findings make it unlikely that implicit measures provide a “bona fide pipeline”
(Fazio, Jackson, Dunton, & Williams, 1995) to people’s true and enduring attitudes, formed on
the basis of past experience and stored in memory as object-evaluation associations. However,
the findings are fully compatible with an alternative conceptualization of attitudes as
evaluations in context (for variants on this theme see Ferguson & Bargh, 2007; Lord & Lepper,
1999; Schwarz, 2007).
Attitude Construal: Evaluation in Context
As William James (1890, p. 333) observed, “My thinking is first and last and always for
the sake of my doing.” Few psychologists doubt this truism, but even fewer heed its
implications. To serve action in a given context, any adaptive system of evaluation should be
informed by past experience, but highly sensitive to the specifics of the present. It should
overweight recent experience at the expense of more distant experience, and experience from
similar situations at the expense of experience from dissimilar situations. In addition, it should
take current goals and concerns into account to ensure that the assessment is relevant to what
we attempt to do now, in this context. In short, only context-sensitive evaluation can guide
behavior in adaptive ways by alerting us to problems and opportunities when they exist; by
interrupting ongoing processes when needed (but not otherwise); and by rendering
information highly accessible that is relevant now, in this situation. From this perspective, it is
no coincidence that any list of desirable context sensitivities reads like a list of the conditions
that give rise to context effects in attitude judgment (Schwarz, 1999; 2007).
Close attention to context also improves the predictive value of attitude reports as
reflected in increased attitude-behavior consistency. This was first highlighted in the seminal
work of Fishbein and Ajzen (1975), who considered it a measurement issue, not a conceptual
issue. However, the underlying principle follows directly from attitude construal models: an
evaluation reported at time 1 will map onto an evaluation or behavioral decision at time 2 to
the extent that the person draws on the same inputs at both points in time. This matching
principle (Lord & Lepper, 1999) offers a coherent conceptualization of the conditions of
Real-time: Cognitive rationale -- 18
stability as well as change in attitude reports and predicts when attitude judgments will or will
not relate to later behavioral decisions (for reviews see Lord & Lepper, 1999; Schwarz, 2007).
Numerous variables – from the person’s current goals to the nature of the context and the
frequency and recency of previous exposure – can influence the temporary construal of the
attitude object and hence the observed consistencies and inconsistencies across time and
Taking the Actor’s Perspective
Construal models of attitudes are compatible with broader current developments in
psychological science, most notably our increasing understanding of the situated and embodied
nature of cognition, emotion, and motivation (for recent reviews see Barsalou, 2005;
Niedenthal et al., 2006; and the contributions in Mesquita et al., 2010). But much as social
psychologists would expect, construal models lack the intuitive appeal of dispositional attitude
models. After all, the logic of dispositional models is fully compatible with observers’ robust
preference for dispositional rather than situational explanations, also known as the
“fundamental attribution error” (Ross, 1977). In contrast, construal models emphasize the role
of contextual variables, which are usually more attended to by the actor (Jones & Nisbett,
1971), who benefits from the context-sensitivity of evaluation in situ. From this perspective,
Allport’s (1935) hope that enduring attitudes can account for an actor’s “consistency of
conduct” in the present is an observer’s dream, but an actor’s nightmare. After decades of
conducting attitude research predominantly from the perspective of an observer who tries to
predict an actor’s behavior, the increasing interest in how people live and experience their lives
on a moment-to-moment basis may contribute to a more systematic exploration of evaluation-
in-context from an actor’s perspective (see also Mehl & Robbins, this volume).
CODA: QUESTIONS IN SITU
As this selective discussion of the complexities of self-report indicates, retrospective
questions often ask respondents for information that they cannot provide with any validity, as
discussed in the sections on self-reports of behaviors and feelings. Other questions ask for
generic answers that may be incompatible with the contextualized and situated nature of
human experience. In the case of attitude measurement, much of the appeal of the enterprise
rests on the hope of predicting behavior across contexts, leading researchers to discount the
context sensitivity of evaluative judgment as undesirable noise. Methods of real-time or close-
in-time measurement attenuate these problems by assessing information in situ, thus allowing
(at least potentially) for the simultaneous assessment of contextual and experiential variables,
and by posing more realistic tasks in the form of questions about respondents’ current
behavior, experiences, and circumstances. These are promising steps.
At the same time, asking questions in situ raises new self-report issues, which have so
far received limited attention. Central to these new issues is the high density of most real-time
Real-time: Cognitive rationale -- 19
data capture procedures, which require respondents to answer the same questions multiple
times within a relatively short time. As noted in the section on question comprehension, this
introduces conversational issues of nonredundancy (Grice, 1975; Schwarz, 1994) that may
invite an emphasize on what is unique and new in each episode at the expense of attention to
what is shared across episodes and has therefore already been reported earlier, making its
repetition a violation of conversational norms. Similarly, rating many episodes along the same
scale invites attention to the frequency principle (Parducci, 1965) of rating scale use, eliciting
differentiation in the reports that may exceed differences in experience. Moreover, repeated
ratings make it likely that previous related episodes are still accessible and serve as scale
anchors or comparison standards. In most cases, these recent anchors would be less extreme
than the “memorable” events used to anchor rating scales in one-time ratings. If so, a given
episode would be rated as more intense in real-time assessment, where it is evaluated against a
less extreme anchor, than in retrospective assessment, where it is evaluated against a more
distant “memorable” episode. The cognitive and communicative processes underlying real-
time self-reports require the systematic exploration and experimentation that advanced the
understanding of self-reports in other domains (Schwarz & Sudman, 1996; Sudman et al., 1996).
Without such work, we run the risk of merely replacing known biases with unknown ones.
Finally, advocates of real-time measurement do probably not appreciate the conclusion
that accurate assessments of real-time experience are poorer predictors of future behavioral
choices than faulty memories of the same experience (e.g., Kahneman et al., 1993; Redelmeier
et al, 2003; Wirtz et al., 2003). As one reviewer of this chapter put it, “Why should we even
bother measuring experience if global or retrospective assessments are the ‘better’ predictors
of choice?” The answer is simple: there’s more to behavioral science than the observer’s desire
to predict others’ choices. A full understanding of the human experience requires attention to
the actor’s perspective and insight into how people live and experience their lives. Real-time
measurement in situ is ideally suited to illuminate the dynamics of human experience from the
actor’s perspective, balancing decades of research that privileged the observer’s goals.
Real-time: Cognitive rationale -- 20
Allen, B. P. (1975). Social distance and admiration reactions of “unprejudiced” Whites. Journal
of Personality, 43, 709-726.
Allport, G. W. (1935). Attitudes. In C. Murchison (ed.), A handbook of social psychology (pp.
789-994). Worcester, Mass: Clark University Press.
Argyle, M. (1999). Causes and correlates of happiness. In D. Kahneman, E. Diener, and N.
Schwarz (Eds.), Well-being: Foundations of hedonic psychology (pp. 353-373). New York:
Barsalou, L. W. (2005). Situated conceptualization. In H. Cohen & C. Lefebvre (eds.), Handbook
of categorization in cognitive science (pp. 619-650). Amsterdam: Elsevier.
Belli, R. (1998). The structure of autobiographical memory and the event history calendar:
Potential improvements in the quality of retrospective reports in surveys. Memory, 6, 383-
Blair, I. V. (2002). The malleability of automatic stereotypes and prejudice. Personality and
Social Psychology Review, 6, 242-261.
Bodenhausen, G. V., Schwarz, N., Bless, H., & Wänke, M. (1995). Effects of atypical exemplars
on racial beliefs: Enlightened racism or generalized appraisals? Journal of Experimental
Social Psychology, 31, 48-63.
Bradburn, N., Sudman, S., & Wansink, B. (2004). Asking questions (2nd ed.). San Francisco:
Brown, N. R. (2002). Encoding, representing, and estimating event frequencies: Multiple
strategy perspective. In P. Sedlmeier & T. Betsch (Eds.), Frequency processing and
cognition (pp. 37-54) New York: Oxford University Press.
Campbell, A. (1981). The sense of well-being in America. New York: McGraw-Hill.
Cantril, H. (1944) Gauging public opinion. Princeton, N.J.: Princeton University Press.
Clark, H. H., & Schober, M. F. (1992). Asking questions and influencing answers. In J. M. Tanur
(Ed.), Questions about questions (pp. 15-48). New York: Russel Sage.
Clore, G. L., Schwarz, N., & Conway, M. (1994). Affective causes and consequences of social
information processing. In R. S. Wyer, & T. K. Srull (Eds.), Handbook of social cognition
(2nd ed.; Vol. 1, pp. 323-418). Hillsdale, NJ: Erlbaum.
Collins, L.M., Graham, J.W., Hansen, W.B., & Johnson, C.A. (1985). Agreement between
retrospective accounts of substance use and earlier reported substance use. Applied
Psychological Measurement, 9, 301 - 309.
Daamen, D. D.,L., & de Bie, S. E. (1992). Serial context effects in survey items. In N. Schwarz & S.
Sudman (Eds.), Context effects in social and psychological research (pp. 97-114). New
York: Springer Verlag.
Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic attitudes: Combating
automatic prejudice with images of liked and disliked individuals. Journal of Personality
Real-time: Cognitive rationale -- 21
and Social Psychology, 81, 800-814.
DeMaio, T. J. (1984). Social desirability and survey measurement: A review. In C. F. Turner & E.
Martin (Eds.), Surveying subjective phenomena (Vol. 2, pp. 257-281). New York: Russell
Duckworth, K. L., Bargh, J. A., Garcia, M., & Chaiken, S. (2002). The automatic evaluation of
novel stimuli. Psychological Science, 13, 513 - 519.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth: Harcourt, Brace,
Eagly, A.H., & Chaiken, S. (2005). Attitude research in the 21st century: The current state of
knowledge. In D. Albarracin, B. T. Johnson, & M. P. Zanna (eds.), The handbook of
attitudes (pp. 743-768). Mahwah, NJ: Erlbaum.
Eich,E., Reeves, J.L., Jaeger, B., & Graff-Radford, S.B. (1985). Memory for pain: Relation between
past and present pain intensity. Pain, 23, 375-380.
Ellsworth, P.C., & Scherer, K.R. (2003). Appraisal processes in emotion. In R.J. Davidson, K.R.
Scherer, & H. H. Goldsmith (eds.), Handbook of affective sciences (pp. 572-595). New
York: Oxford University Press.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability in automatic
activation as an unobtrusive measure of racial attitudes. A bona fide pipeline? Journal of
Personality and Social Psychology, 69, 1013-1027.
Ferguson, M. J., & Bargh, J. A. (2003). The constructive nature of automatic evaluation. In J.
Musch & K. C. Klauer (Eds.), The psychology of evaluation: Affective processes in
cognition and emotion (pp.169-188). NJ: Erlbaum.
Ferguson, M. J., & Bargh, J.A. (2007). Beyond the attitude object: Automatic attitudes spring
from object-centered-contexts. In B. Wittenbrink & N. Schwarz (eds.), Implicit measures
of attitudes: Progress and controversies. New York: Guilford.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Fredrickson, B.L. (2000). Extracting meaning from past affective experiences: The importance of
peaks, ends, and specific emotions. Cognition & Emotion, 14, 577-606.
Gibbs, R. W. (2006). Embodiment and cognitive science. New York: Cambridge University Press.
Grice, H. P. (1975). Logic and conversation. In P. Cole, & J.L. Morgan (Eds.), Syntax and
semantics, 3: Speech acts. pp. 41-58. New York: Academic Press.
Himmelfarb, S., & Lickteig, C. (1982). Social desirability and the randomized response
technique. Journal of Personality and Social Psychology, 43, 710-717.
James, W. (1890). The principles of psychology (Vol. 2). New York: Henry Holt.
Jones, E. E., & Sigall, H. (1971). The bogus pipeline: A new paradigm for measuring affect and
attitude. Psychological Bulletin, 76, 349-364.
Juster, F. T. (1985). Preferences for work and leisure. In F. T. Juster & F. P. Stafford (eds.), Time,
Real-time: Cognitive rationale -- 22
goods, and well-being (pp. 397-414). Ann Arbor: Institute for Social Research.
Kahneman, D., Fredrickson, B.L., Schreiber, C.A., & Redelmeier, D. (1993). When more pain is
preferred to less: Adding a better end. Psychological Science, 4, 401-405.
Kahneman, D., Krueger, A. B., Schkade, D., Schwarz, N., & Stone, A. A. (2004). A survey method
for characterizing daily life experience: The Day Reconstruction Method (DRM). Science,
Kahneman, D., Krueger, A. B., Schkade, D., Schwarz, N., & Stone, A. A. (2006). Would you be
happier if you were richer? A focusing illusion. Science, 312, 1908-1910.
Knäuper, B., Schwarz, N., & Park, D.C. (2004). Frequency reports across age groups: Differential
effects of frequency scales. Journal of Official Statistics, 20, 91-96.
Krosnick, J.A. & Alwin, D. F. (1987). An evaluation of a cognitive theory of response order effects
in survey measurement. Public Opinion Quarterly, 51, 201-19.
Krueger, A., Kahneman, D., Schkade, D., Schwarz, N., & Stone, A. (2009). National time
accounting: The currency of life. In A. Krueger (ed.), Measuring the subjective well-being
of nations: National accounts of time use and well-being (pp. 9-86). Chicago: Chicago
Lane, K.A., Banaji, M.R., Nosek, B.A., & Greenwald, A.G. (2007). Understanding and using the
Implicit Association Test: IV. What we know (so far) about the method. In B.
Wittenbrink & N. Schwarz (eds.), Implicit measures of attitudes: Progress and
controversies. New York: Guilford.
Linton, S. J., & Melin, L. (1982). The accuracy of remembering chronic pain. Pain, 13, 281-285.
Loewenstein, G. F. & Adler, D. (1995). A bias in the prediction of tastes. Economic Journal, 105,
Loewenstein, G., & Schkade, D. A. (1999). Wouldn’t it be nice? Predicting future feelings. In D.
Kahneman, E. Diener, and N. Schwarz (Eds.), Well-being: Foundations of hedonic
psychology (pp. 85-105). New York: Russell-Sage.
Lord, C.G., & Lepper, M. R. (1999). Attitude representation theory. Advances in
Experimental Social Psychology, 31, 265-343.
March, J. (1978). Bounded rationality, ambiguity, and the engineering of choice. Bell Journal of
Economics, 9, 587-608.
Mellers, B. A., & McGraw, A. P. (2001). Anticipated emotions as guides to choice. Current
Directions in Psychological Science, 10, 210-214.
Menon, G. (1993). The effects of accessibility of information in memory on judgments of
behavioral frequencies. Journal of Consumer Research, 20, 431-440.
Menon, G. (1994). Judgments of behavioral frequencies: Memory search and retrieval
strategies. In N. Schwarz & S. Sudman, S. (Eds.) (1994). Autobiographical memory and
the validity of retrospective reports (pp. 161- 172). New York: Springer Verlag.
Menon, G., Raghubir, P., & Schwarz, N. (1995). Behavioral frequency judgments: An
Real-time: Cognitive rationale -- 23
accessibility-diagnosticity framework. Journal of Consumer Research, 22, 212-228.
Mesquita, B., Barrett, L.F., & Smith, E.R. (Eds.) (2010). The mind in context). New York: Guilford
Mitchell, T. R., Thompson, L., Peterson, E., & Cronk, R. (1997). Temporal adjustments in the
evaluation of events: The “rosy view.” Journal of Experimental Social Psychology, 33,
Niedenthal, P. M., Barsalou, L.W., Winkielman, P., Krauth-Gruber, S., & Ric, F. (2006).
Embodiment in attitudes, social perception, and emotion. Personality and Social
Psychology Review, 9, 184-211.
Norenzayan, A., & Schwarz, N. (1999). Telling what they want to know: Participants tailor
causal attributions to researchers' interests. European Journal of Social Psychology, 29,
Parducci, A. (1965). Category judgment: A range-frequency model. Psychological Review, 72,
Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of
Personality and Social Psychology, 46, 598-609.
Payne, S. L. (1951). The art of asking questions. Princeton: Princeton University Press.
Redelmeier, D., & Kahneman, D. (1996). Patients’ memories of painful medical treatments:
Real-time and retrospective evaluations of two minimally invasive procedures. Pain, 66,
Redelmeier D.A., Katz, J., and Kahneman, D. (2003). Memories of colonoscopy: A randomized
trial. Pain, 104, 187-194.
Robinson, J. P. (1977). How Americans use time. A social-psychological analysis of everyday
behavior. New York: Praeger.
Robinson, M. D., & Clore, G. L. (2002). Belief and feeling: Evidence for an accessibility model of
emotional self-report. Psychological Bulletin, 128, 934-960.
Ross, L. (1977). The intuitive psychologist and his shortcomings: Distortions in the attribution
process. Advances in Experimental Social Psychology, 10, 173-220.
Ross, M. (1989). The relation of implicit theories to the construction of personal histories.
Psychological Review, 96, 341-357.
Schkade, D. A., & Kahneman, D. (1997). Does living in California make people happy?: A
focusing illusion in judgments of life satisfaction. Psychological Science, 9, 340-346.
Schuman, H. (1983). Survey research and the fundamental attribution error. Personality and
Social Psychology Bulletin, 9, 103-104.
Schuman, H., & Presser, S. (1981). Questions and answers in attitude surveys. New York:
Schwarz, N. (1994). Judgment in a social context: Biases, shortcomings, and the logic of
conversation. Advances in Experimental Social Psychology, 26, 123-162.
Schwarz, N. (1996). Cognition and communication: Judgmental biases, research methods and
Real-time: Cognitive rationale -- 24
the logic of conversation. Hillsdale, NJ: Erlbaum.
Schwarz, N. (1999). Self-reports: How the questions shape the answers. American Psychologist,
Schwarz, N. (2002). Situated cognition and the wisdom of feelings: Cognitive tuning. In L.
Feldman Barrett & P. Salovey (Eds.), The wisdom in feelings (pp.144-166). New York:
Schwarz, N. (2007). Attitude construction: Evaluation in context. Social Cognition, XX
Schwarz, N., Groves, R., & Schuman, H. (1998). Survey methods. In D. Gilbert, S. Fiske, & G.
Lindzey (Eds.), Handbook of social psychology (4th edition, Vol. 1, pp. 143-179). New
Schwarz, N., & Hippler, H.J. (1991). Response alternatives: The impact of their choice and
ordering. In P. Biemer, R. Groves, N. Mathiowetz, & S. Sudman (Eds.), Measurement
error in surveys (pp. 41-56). Chichester, UK: Wiley.
Schwarz, N., & Hippler, H. J. (1995). Subsequent questions may influence answers to preceding
questions in mail surveys. Public Opinion Quarterly, 59, 93-97.
Schwarz, N., Kahneman, D., & Xu, J. (2009). Global and episodic reports of hedonic experience.
In R. Belli, D. Alwin, & F. Stafford (eds.), Using calendar and diary methods in life events
research (pp. 157-174). Newbury Park, CA: Sage.
Schwarz, N., Hippler, H.J., Deutsch, B. & Strack, F. (1985). Response categories: Effects on
behavioral reports and comparative judgments. Public Opinion Quarterly, 49, 388-395.
Schwarz, N., Knäuper, B., Oyserman, D., & Stich, C. (2008). The psychology of asking questions.
In E. de Leeuw, J. Hox, & D. Dillman (eds.), International handbook of survey
methodology (pp. 18-34). New York: Taylor & Francis.
Schwarz, N. & Oyserman, D. (2001). Asking questions about behavior: Cognition,
communication and questionnaire construction. American Journal of Evaluation, 22,
Schwarz, N., Strack, F., Müller, G., & Chassein, B. (1988). The range of response alternatives may
determine the meaning of the question: Further evidence on informative functions of
response alternatives. Social Cognition, 6, 107 - 117.
Schwarz, N., & Sudman, S. (Eds.) (1994). Autobiographical memory and the validity of
retrospective reports. New York, NY: Springer Verlag.
Schwarz, N., & Sudman, S. (Eds.) (1996). Answering questions: Methodology for
determining cognitive and communicative processes in survey research. San Francisco:
Schwarz, N., & Xu, J. (in press). Why don’t we learn from poor choices? The consistency of
expectation, choice, and memory clouds the lessons of experience. Journal of Consumer
Smith, E.R., & Conrey, F.R. (2007). Mental representations are states not things: Implications for
Real-time: Cognitive rationale -- 25
implicit and explicit measurement. In B. Wittenbrink & N. Schwarz (eds.), Implicit
measures of attitudes: Progress and controversies. New York: Guilford.
Smith, E. R., & Semin, G. R. (2004). Socially situated cognition: Cognition in its social context.
Advances in Experimental Social Psychology, 36, 53-117.
Stone, A. A., Schwartz, J. E., Schwarz, N., Schkade, D., Krueger, A., & Kahneman, D. (2006). A
population approach to the study of emotion: Diurnal rhythms of a working day
examined with the Day Reconstruction Method (DRM). Emotion, 6, 139-149.
Stone, A. A., Shiffman, S. S., & DeVries, M. W. (1999). In D. Kahneman, E. Diener, & N. Schwarz
(eds.), Well-being: The foundations of hedonic psychology (pp. 26-39). New York:
Strack, F., Schwarz, N., & Wänke, M. (1991). Semantic and pragmatic aspects of context effects
in social and psychological research. Social Cognition, 9, 111-125.
Sudman, S., Bradburn, N. M., & Schwarz, N. (1996). Thinking about answers: The
application of cognitive processes to survey methodology. San Francisco, CA: Jossey-
Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The psychology of survey response. Cambridge:
Cambridge University Press.
Vargas, P.T., Sekaquaptewa,D., & von Hippel, W. (2007). Armed only with paper and pencil:
“Low tech” measures of implicit attitudes. In B. Wittenbrink & N. Schwarz (eds.),
Implicit measures of attitudes: Progress and controversies (pp. 103-122). New York:
Wicker, A.W. (1969). Attitudes versus actions: The relationship of verbal and overt behavioral
responses to attitude objects. Journal of Social Issues, 25, 41–78.
Willis, G. B. (2004). Cognitive interviewing. A tool for improving questionnaire design. Thousand
Wilson, T. D., & Gilbert, D. T. (2003). Affective forecasting. In M. P. Zanna (Ed.), Advances in
experimental social psychology (Vol. 35, pp. 345-411). San Diego, CA: Academic
Winkielman, P., Knäuper, B., & Schwarz, N. (1998). Looking back at anger: Reference periods
change the interpretation of (emotion) frequency questions. Journal of Personality and
Social Psychology, 75, 719-728.
Wirtz, D., Kruger, J., Scollon, C. N., & Diener, E. (2003). What to do on spring break? The role of
predicted, on-line, and remembered experience in future choice. Psychological Science,
Withey, S. B. (1954). Reliability of recall of income. Public Opinion Quarterly, 18, 31 - 34.
Wittenbrink. B. (2007). Measuring attitudes through priming. In B. Wittenbrink & N. Schwarz
(eds.), Implicit measures of attitudes: Progress and controversies (pp. 17-58). New York:
Real-time: Cognitive rationale -- 26
Wittenbrink, B., Judd, C. M., & Park, B. (2001). Spontaneous prejudice in context: Variability in
automatically activated attitudes. Journal of Personality and Social Psychology, 81, 815-
Wittenbrink, B., & Schwarz, N. (eds.) (2007). Implicit measures of attitudes:
Procedures and controversies. New York: Guilford.
Xu, J., & Schwarz, N. (2009). Do we really need a reason to indulge? Journal of Marketing
Research, 46, 25-36.
Real-time: Cognitive rationale -- 27
Table 1. Reported Daily TV Consumption as a Function of Response Alternatives
Reported Daily TV Consumption
Low Frequency Alternatives High Frequency Alternatives
Up to 1/2 h 7.4% Up to 2 1/2h 62.5%
1/2 h to 1h 17.7% 2 1/2h to 3h 23.4%
1h to 1 1/2h 26.5% 3h to 3 1/2h 7.8%
1 1/2h to 2h 14.7% 3 1/2h to 4h 4.7%
2h to 2 1/2h 17.7% 4h to 4 1/2h 1.6%
More than 2 1/2h 16.2% More than 4 1/2h 0.0%
Note. N = 132. Adapted from Schwarz, N., Hippler, H.J., Deutsch, B., & Strack, F. (1985). Response
scales: Effects of category range on reported behavior and comparative judgments. Public Opinion
Quarterly, 49, 388-395. Reprinted by permission.