Developmental Psychology

1990, Vol. 26, No. 6, 978-986

Copyright 1990 by the American Psychological Association, Inc.

0012-1649/90/$00.75

Predicting Adolescent Cognitive and Self-Regulatory Competencies From

Preschool Delay of Gratification: Identifying Diagnostic Conditions
Yuichi Shoda and Walter Mischel
Columbia University

Philip K. Peake

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Smith College

Variations of the self-imposed delay-of-gratification situation in preschool were compared to determine when individual differences in this situation may predict aspects of cognitive and self-regulatory competence and coping in adolescence. Preschool children from a university community
participated in experiments that varied features of the self-imposed delay situation. Experimental
analyses of the cognitive-attentional processes that affect waiting in this situation helped identify
conditions in which delay behavior would be most likely to reflect relevant cognitive and attentional competencies. As hypothesized, in those conditions, coherent patterns of statistically significant correlations were found between seconds of delay time in such conditions in preschool and
cognitive and academic competence and ability to cope with frustration and stress in adolescence.

To be able to delay immediate satisfaction for the sake of

future consequences has long been considered an essential
achievement of human development. After a series of investigations into the individual differences associated with the choice
to delay gratification (e.g., Klineberg, 1968; Mischel, 1958,
1961a, 1961b, 1966; Mischel & Metzner, 1962; Schack & Massari, 1973; Walls & Smith, 1970), research turned to the processes underlying the ability to sustain self-imposed delay of
gratification after the initial choice has been made (e.g, Mischel, 1974,1981; Toner & Smith, 1977). In a recent follow-up
study, preschool children who delayed gratification longer in
the self-imposed delay paradigm (e.g., Mischel, Ebbesen, &
Zeiss, 1972) were described more than 10 years later by their
parents as adolescents who were significantly more competent
(Mischel, Shoda, & Peake, 1988). Specifically, when these children became adolescents, their parents rated them as more academically and socially competent, verbally fluent, rational, attentive, planful, and able to deal well with frustration and
stress. The study suggested that long-term prediction may be

possible from the self-imposed delay paradigm, adding to a

growing tradition of research devoted to the identification of
stability and coherence throughout development (e.g., Block,
1971; Caspi, Elder, & Bern, 1987; Erickson, Sroufe, & Egeland,
1985; Kagan & Moss, 1962). However, the small sample size
required combining the different experimental situations in
which delay behavior was measured. Therefore, it was not possible to compare major variations of the self-imposed delay situation to examine the characteristics that might render it more or
less predictive of the obtained long-term outcomes. The present
study is an effort to overcome this constraint.
In the present study, we attempted to identify the particular
psychological conditions in which children's delay of gratification behavior is more likely to predict relevant individual differences in developmental outcomes. The identification of
these conditions, which may be considered "diagnostic" (Quattrone & Tversky, 1984; Tversky & Hutchinson, 1986), is derived
directly from the theoretical and experimental analyses of the
cognitive-attentional processes that enable the young child to
delay (e.g, Mischel, 1974,1981,1984). For this reason, we supplemented the original follow-up sample with a second, larger
wave of outcome data collected about 3 years later, and thereby
almost doubled the available number of respondents. The new
follow-up wave also added expanded rating measures of both
cognitive and coping competence, as well as Scholastic Aptitude Test (SAT) scores. These new data allowed us to compare
long-term correlates of delay behavior in the major variations of
the Mischel et al. (1972) self-imposed delay situations.
An important variation in the original preschool delay situations was whether or not the reward objects were more salient
(i.e, exposed vs. obscured) during the delay period. In an extensive series of experiments (Mischel, 1974,1981) to clarify the
basic processes allowing young children to delay gratification,
it was found that for children at this age (about 4.5 years old),
physically exposing the rewards appears to increase the tendency to have arousing, consumatory thoughts about them.

Preparation of this article and the research were supported in part

by Grants MH45994 and MH39349 to Walter Mischel from the National Institute of Mental Health.
We are grateful to Antonette Zeiss for her role in facilitating the
follow-up mailing for this study and in compiling and preparing the
experimental data for computer analysis; without her generous help,
this research would have been much more difficult to conduct. We
would also like to acknowledge the essential contributions of the
former students of Walter Mischel, who were crucial in conducting the
original experiments in which children's preschool delay time was assessed. Finally, we thank the children, parents, and staff of the Bing
School, who were both subjects and friends. Without all of these participants, this longitudinal study would not have been possible.
Correspondence concerning this article should be addressed to Yuichi Shoda or Walter Mischel, Department of Psychology, Columbia
University, New York, New York 10027.
978

979

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DELAY OF GRATIFICATION

This tendency could be overcome, however, when children

used effective cognitive strategies during the delay period, for
example, by distracting themselves from the arousing qualities
of the rewards or by transforming them cognitively (e.g, Mischel & Baker, 1975; Mischel & Moore, 1980). Furthermore, although these strategies were suggested as manipulations in the
experiments, they also seemed to be used spontaneously by
children who delayed longer.
Thus, when the rewards were exposed, preschoolers tended
to wait longer when they were given effective strategies, or when
they generated their own, for reducing the arousal while sustaining their goal-directed delay. Such strategies seemed to involve "a combination of avoiding excessive frustration by not
focusing on the actual rewards or by transforming them to minimize motivational arousal, and attending instead to the symbolic representation of the outcomes" (Mischel, 1974, p. 288).
Therefore, when preschoolers were not given these strategies,
their behavior in the exposed-rewards situation should have
more readily reflected any naturally occurring individual differences in their spontaneous use of such strategies. To the degree that individual differences in the child's ability to generate
effective cognitive-attentional strategies to cope with delay of
gratification are enduring and have consequences for facilitating adaptation, we expected children's preschool delay time
measured with the rewards exposed to predict later outcomes
and indices of cognitive and social competence relevant to this
ability. On the basis of the experimental analysis and the results
from thefirstfollow-up, we expected these outcomes to include
such qualities as attentional-cognitive resourcefulness and flexibility and, more generally, academic competencies, as well as
more effective, mature coping with frustration and stress (see
Mischel et al, 1988).
In comparison with the exposed-rewards condition, when
the rewards were obscured, no special strategies seemed to be
required for preschoolers to delay (e.g, Mischel, 1974). For example, instructions to "think fun" and to self-distract did not
significantly increase delay time when the rewards were already
obscured, indicating that children in this condition were able to
spontaneously wait without such help (Mischel et al, 1972). We
therefore reasoned that children's delay behavior in this situation would be less related to their ability to generate effective
strategies for coping with the conflict because the situation did
not demand such ability. Observed variance in delay behavior,
which was still substantial, might then reflect specific motivational or other situational considerations (e.g, how much they
wanted the particular objects) rather than their ability to generate strategies for sustaining self-imposed delay. This expectation
is consistent with the view that individual differences in strategiesforcoping with frustration or stress may be especially visible in situations that strain the coping competencies of the
individuals in them (Wright & Mischel, 1987). It is also consistent with a research strategy that has begun to show considerable value in attachment research: Early behavioral antecedents
seem more predictive when they are assessed in situations that
tax the coping skills of the individual (Ainsworth, 1979; Waters
& Sroufe, 1983).
So far, we have considered the diagnosticity of delay behavior
when children had to rely on their own coping strategies. The
available data also include conditions in which various strate-

gies and self-instructions had been suggested to the children to

use during the waiting period (e.g, Mischel & Baker, 1975). In
those conditions, children's behavior reflected not just their
own spontaneous coping with delay but also their reactions to
the suggested strategiesstrategies that sometimes were helpful for delay but that sometimes made delay more difficult. For
example, when the experimenter suggested an effective strategy
(e.g, to self-distract by "thinking fun" while waiting, as in Mischel et al, 1972), children who followed the instruction should
have been able to wait even if they had not spontaneously distracted themselves. In contrast, when poor strategies (e.g, to
think about rewards) were suggested, children were exposed to
a potentially confusing set of strategies (e.g, following the experimenter-suggested strategies vs. those they spontaneously
thought would be helpful, such as self-distraction). Thus, their
delay time should be less clearly reflective of the ability to generate effective strategies. However, complex and unpredictable
interactions may occur between the children's own spontaneous
delay strategies and the particular type of strategies suggested
to them, making it difficult to predict the potential meaning of
the child's delay behavior in those conditions. Our theoretical
predictions of greater diagnosticity therefore focused primarily
on comparisons between conditions in which the rewards were
exposed versus obscured, with no ideation suggested to the
children (i.e, their strategies were entirely spontaneous). For the
sake of completeness, however, the data for conditions in which
strategies were suggested will also be presented.

Method
Overview
Preschool children's delay of gratification behavior was assessed
during a period of approximately 6 years (1968-1974) in a series of
experiments conducted at the Bing School at Stanford University (e.g.,
Mischel & Ebbesen, 1970; Mischel, et al, 1972). In those studies, a
total of 653 children (316 boys, 337 girls) participated in at least one
experiment. About 10 years later (1981-1982), a short questionnaire
concerning the coping and cognitive competence of the children and
the California Child Q-set (CCQ) were mailed to the 125 parents
whose addresses could be located, yielding 95 respondents (see Mischel et al, 1988). To expand the sample of respondents, a second followup based on a more extensive address search was conducted in 1984. In
the second mailing, all parents were sent a new expanded questionnaire about coping and competence (the Adolescent Coping Questionnaire; ACQ) and a biographical information sheet on which they indicated their children's SAT scores. The CCQs were also sent to those
who either did not respond to the previous mailing or did not receive
the mailing because their addresses were not available at that time.
Materials were mailed to parents of 506 subjects. This yielded responses from parents of 90 subjects whose data were not available in
thefirst-wavefollow-up, as well as additional data from the parents
who had responded in the first mailing. As a result, the sample for
which CCQ was available increased from 67 to 165, and we obtained
parental ratings on the new 14-item ACQ for 134 children, as well as
reports of 94 children's SAT scores. Together with thefirst-wavemailing, a sample of 185 children was now available on whom there was at
least one follow-up measure.1
1

Of the 653 original subjects, 103 were not tested in a standard selfimposed delay situation and therefore were not used. Of the remaining
550, no address was known for 114, and follow-up materials were re-

980

Y. SHODA, W MISCHEL, AND P. PEAKE

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Subjects
The subjects of the present study were the 185 children (103 girls, 82
boys) whose preschool delay behavior was observed in a standard selfimposed delay situation as described below and whose parent(s) returned any of the follow-up measures in either of the two waves of
follow-up assessments. Although a minority of subjects had more than
one delay experience, we used only their first exposure to a delay situation to avoid possible reactive effects (as discussed in Mischel et al.,
1988). The mean age of the present sample at the time of experimental
assessment of delay of gratification was 4 years, 4 months. The children were preschoolers in the Bing School of Stanford University, a
preschool for mostly middle-class children of faculty and students
from the Stanford University community. Their mean delay time was
512.8 s, with a standard deviation of 368.7 s. The mean age of those
who responded to the first follow-up (conducted in 1981 -1982) was 1S
years, 9 months, and the mean age at the time of the second follow-up
(conducted in 1984) was 18 years, 3 months. Parents of 67 of these
children (35 girls, 32 boys) returned the CCQ in the first mailing, and
parents of 100 children (58 girls, 42 boys) returned the CCQ in the
second mailing. Parents of 2 children received and returned the CCQ
in both mailings, making the number of subjects for whom the CCQ
was available from either of the mailings a total of 165 (92 girls, 73
boys). Parents of 134 children (78 girls, 56 boys) returned the ACQ
(described below), and parents of 94 children reported SAT scores.

Assessment of Delay of Gratification Behavior

Delay of gratification was assessed in various versions of the basic
self-imposed delay waiting paradigm described in detail elsewhere
(Mischel, 1974; Mischel et al, 1972). Children were escorted individually into an experimental room in the Bing School, played briefly with
some toys with the experimenter, and were told they would play with
them more later. The child was then seated at a table on which there
was a bell, and was shown reward objects determined by pretest to vary
in desirability (e.g., one small marshmallow vs. two, one small stick
pretzel vs. two, one colored plastic poker chip vs. two). The particular
objects in the contingency varied from study to study, but the items in
each pair were all pretested to be of age-appropriate interest and to be
sufficiently close in value to create a conflict for young children between the temptation to stop the delay and the desire to persist for the
preferred outcome when the latter required delay. The experimental
room was deliberately stripped of distractors. After asking which of
the objects in the choice (e.g, one or two marshmallows) the subject
preferred, the experimenter introduced the child to the contingency:
The experimenter indicated that she or he had to go out of the room
then but that "if you wait until I come back by myself then you can
have this one [pointing to the preferred object]. If you dont want to
wait you can ring the bell and bring me back any time you want to. But
if you ring the bell then you can't have this one [pointing to the preferred object], but you can have that one [pointing to the less preferred
object]."
After testing the child's comprehension of the contingency, the experimenter left the room and returned when the subject rang the bell or
reached a predetermined criterion time (usually 15 min, but sometimes
20 min, depending on the particular study). The time until the child

turned by the post office as undeliverable for an additional 32. Thus,

404 subjects probably received our follow-up material (barring unreported mail loss), of which the present sample of 185 represents a return rate of 46%. The 185 subjects in the present sample delayed an
average of 62 s longer than the 219 who did not respond to our mailing
(512.8 s vs. 450.7 s), but the difference was not significant at the .05
level, /(402) = 1.7, p=.09.

rang the bell was measured in seconds. In the present data analysis, to
allow combining data across studies, delay times exceeding 15 min
were truncated at 15 min.

Four Types of Delay Situations

To test the hypotheses described earlier, subjects were divided into
four groups, defined by the two major features of the delay situation:
whether rewards were exposed or obscured and whether or not any
ideation instructions were given suggesting what the child might think
about during the delay period. The number of subjects in each of these
four groups is shown in Table 1.
Subjects in the four groups should not have differed systematically
in sample characteristics observable at the time of the original experiments because they had been assigned randomly into experimental
conditions. In fact, the proportion of the sexes in the four groups was
not significantly different, x2(3, N = 185) = 2.87, p > .40, and a Sex X
Condition analysis of variance (ANOVA) revealed no significant main
effect of sex, F(l, 174) = 0.45, p> .50, condition, F(13,174) = .94, p >
.40, or Sex x Condition, FQ, 174) = 0.47, p > .70, on the age at which
delay time was measured. With delay time as the dependent variable, a
Sex X Condition ANOVA revealed a significant main effect of condition,
FQ, 174) = 4.3, p < .01, as expected, whereas no other effects were
significant. The mean delay times reported in Table 1 confirm that
when no cognitive and attentional strategies were suggested, delay was
more difficult when the rewards were exposed (e.g, Mischel, 1974).
The skewness of the distribution of delay times did not significantly
differ from 0 in any of the four conditions.

The Adolescent Coping Questionnaire

To assess the subjects' cognitive and coping competencies and selfcontrol skills in adolescence, a 14-item questionnaire was sent to all of
the parents who were contacted in the second wave; it was returned for
134 children. The 14 items are listed verbatim in Table 2; they were
given with the following instructions:
In this section we want you to think about your child in comparison to his or her peers, such as classmates and other same-age
friends. We would like to get your impression of how your son or
daughter compares to those peers, each time on a rating scale of
1-9. Record your answers in the space provided by writing in the
best number from the following scale:
1

Not at all

Moderately

9
Extremely

California Child Q-Set

The CCQ is an age-appropriate modification of the California Q-set
consisting of 100 widely ranging, personality-relevant items. It was included in our research because of its extensive previous use in studies
of personality coherence and in the assessment of long-term correlates
of various aspects of the young child's delay of gratification (Funder,
Block, & Block, 1983). In the first mailing, parents of 67 subjects in the
present subsample returned the CCQ (46 subjects were described by
both parents, 3 by father only, and 18 by mother only). In the second
mailing, parents of 100 subjects returned the CCQ (64 subjects were
described by both parents, 9 by father only, and 27 by mother only).
When both father and mother returned the CCQ describing their
child, a composite CCQ description was formed by averaging the father's and mother^ descriptions. With an exception of two cases, there
was no overlap of subjects between the first and the second mailing (in
the second mailing, we did not send the CCQ to those subjects who
returned it in the first mailing). Thus, a total of 165 subjects were

981

DELAY OF GRATIFICATION

Table 1
Composition of the Total Follow-Up Sample of Respondents
Spontaneous ideation
Measure

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Sample size*
Female
Male
Total
Age at delay experiment
(months)

M
SD

Rewards
exposed

Rewards
obscured

Suggested ideation
Rewards
exposed

Rewards
obscured

25
27
52

32
26
58

22
11
33

24
18
42

5L6
6.8

5L9
4.9

50.8

512
4.5

590.4
330.4
-.53

516.7
424.9
-.28

6.2

Delay time (s)

M
SD
Skew

365.2
3729

.51

585.1
320.6
-.32

* Subjects for whom at least one of the follow-up measures was obtained.

described using the CCQ by both parents (110 subjects), by father only
(11 subjects), or by mother only (44 subjects).
In the first mailing, parents were sent 100 cards, each printed with
one of the CCQ items, and sorted the cards into nine equally sized piles
(11 items in each, except for 12 items in the middle pile) according to
the items' descriptiveness for their child. Parents then returned the
piles of cards in separate envelopes marked Pile 1 to Pile 9. Each child
received a score for each of the items according to the pile in which the
item was placed (i.e., if Item 25, uses and responds to reason, was returned in Envelope 6, the child received a score of 6 for CCQ Item 25).
Because this procedure is time-consuming for parents and because we
wanted to increase the likelihood of returns in the second follow-up
mailing, we presented the 100 items as questionnaire items, and the
parents simply rated their children on each item using a scale of 1 to 9,
indicating how descriptive each item was for their child, without forcing a predetermined distribution of scores.
This procedural difference created the possibility that, depending
on how willing they were to give extreme (e.g, 1 or 9) scores, the parents
in the second mailing used more or less of the scale compared with
parents who used a forced distribution in the first mailing. Therefore,
before combining the Q-sort data on the basis of the two mailings,
these data werefirststandardized within any given "profile" of a child.
Specifically, for each CCQ "profile" given by a single rater (i.e, description of a child through ratings given to each of the 100 CCQ
items), we calculated the mean and the standard deviation across
items. (For example, if the forced distribution had been used, the mean
would always be 5.0, and the standard deviation would always be 2.58.)
The raw ratings in a profile were then converted into "within-profile"
standard scores by subtracting the "profile mean" from each item and
dividing the remainder by the "profile standard deviation." Thus, although the shapes of the distributions may have differed across raters,
all standardized profiles had the same mean and standard deviations,
as did the profiles based on the forced distribution. For the CCQ data
obtained in the first follow-up wave, the interrater (mother vs. father)
correlations for each of the 100 items ranged from .01 to .70, with a
median value of .42, yielding a median Spearman-Brown estimated
reliability of .60 for the mother-father composite. For the CCQ data
obtained in the second follow-up wave, the interrater (mother vs. father) correlations for each of the 100 items ranged from .02 to .79, with a
median value of .40, yielding a median Spearman-Brown estimated
reliability of .58 for the mother-father composite.

CCQ Items Judged Relevant

The total of 100 CCQ items included a wide range of personality
descriptions, only some of which were expected to be relevant to children's delay behavior in the self-imposed delay situations on the basis
of the experimental analysis of the delay process. To specify such items
a priori, we were guided by a summary of experimental findings published in Mischel (1974) 7 years before the first wave of the follow-up
mailing. Specifically, in the last two paragraphs, Mischel (1974, pp.
287-288) outlined the cognitive and attentional factors that were
found to play important roles in determining children's delay behavior.
These two paragraphs, slightly edited for readability, were shown in
individual sessions to 7 college-educated subjects who were not psychologists, were not in psychology graduate training programs, and
had no specific knowledge of the findings of the present study. After
reading these paragraphs, they were given the 100 CCQ items printed
on separate cards and were asked (with written instructions) to find 15
cards that characterize a person who is able to use the cognitive and
attentional strategies described in these two paragraphs to delay gratification. Subjects further divided these 15 cards into three groups of 5
cards each according to their degree of relevance for the processes and
strategies described in the paragraphs. We assigned the 5 hems they
chose as most relevant a score of+3 (or - 3 , depending on whether the
expected relationship to delay ability was positive or negative), the next
5 relevant ones a score of+2 (or - 2 ) , the next 5 a score of +1 (or - 1 ) , and
the remaining 85 items a score of 0. The mean interrater agreement
(correlations computed across the 100 CCQ items) was .42, and the
Spearman-Brown estimate, based on this mean correlation, for the
reliability of the composite of 7 raters was .83. The scores given to each
item by the 7 subjects were then averaged, and all items with a mean
relevance score of 1.0 or greater (regardless ofsign) were chosen as items
theoretically expected to be related to the abilities that underlie effective delay of gratification in self-imposed delay situations. There were
11 such items, shown in Table 3 in descending order of their degree of
relevance.

SAT Verbal and Quantitative Scores

On the biographical information sheet included in the mailings,
parents were asked to provide their children's SAT verbal and quantitative scores, when available. Parents of 94 children reported SAT scores.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

982

Y. SHODA, W MISCHEL, AND P. PEAKE

To assess the reliability of the parental reports of SAT scores, we also

contacted the Educational Testing Service (ETS). On the basis of the
names and birth dates, ETS located in its data bases scores for 69 of
these 94 subjects. Because of the confidentiality of the SAT scores,
scores found by ETS could not be released in ways that allowed matching with the identity of individual children. However, by supplying
parent-reported SAT scores to ETS before the data-base search, ETS
could match these scores with the scores in the ETS data bases. The
parent-reported SAT scores and the corresponding scores found in the
data bases were thenreleasedwithout any identifying information and
with the order of subjects randomized (but with the correspondence
between the parent- and the ETS-supplied scores kept intact). The
correlation was .94 between the scores found in the ETS data bases and
the ones reported by the parents. Although these 69 subjects located by
ETS may include some "false matches" (i.e, different children with the
same first and last names and birth dates as our subjects), the high
correlation suggests that the parent-supplied SAT scores were generally accurate for the majority of the subjects. To avoid the possibility of
including false matches, all SAT results reported here are based on the
scores supplied by parents.

Results
Adolescent Coping Questionnaire
The correlations between preschool delay time in the four
types of waiting situations and parental ratings of the children
as adolescents are shown in Table 2. In the exposed-rewardsspontaneous-ideation condition, 12 of the 14 ratings of the children as adolescents were statistically significant. For example,
those who delayed longer in preschool were rated as more likely
to exhibit self-control in frustrating situations, less likely to
yield to temptation, more intelligent, and less distractable
when trying to concentrate. In contrast, in the other conditions,
of the 42 correlations calculated with preschool delay time (14
items X 3 conditions), only 3 reached statistical significance
(p < .05).

California Child Q-Set

SAT Scores
The results also allow us to examine potential links between
preschool delay time in the various conditions and the children's SAT Verbal and Quantitative scores, as shown in Table 4.
Because SAT scores were not available for all subjects, in some
conditions the sample sizes became barely sufficient for a
meaningful computation of correlations. Nevertheless, the observed differences between the conditions are consistent with
the results obtained with the rating measures. As expected, preschool delay time correlated positively with SAT when the rewards were exposed and no strategies were suggested. In contrast, correlations were negative and statistically insignificant
in the other conditions.2
To more closely assess the nature of the competencies in
adolescence predicted by preschool waiting time, and in particular to estimate which of the significant correlations with parental rating measures may reflect such school-related achievements and abilities as assessed by SAT scores, partial correlations between delay time and parental ratings were computed,
controlling for both verbal and quantitative SAT scores. After
the variance attributable to SAT was partialed out, of the 11
CCQ items judged relevant (see Table 3), the following remained significantly correlated with delay time (p < .05, df=
29): uses andresponds to reason (r= .40); is planful, thinksahead
(r = .38); and tends to go to pieces under stress, becomes rattled
anddisorganized(r= -.36). The correlation between preschool
delay time and the item is unable to delay gratification remained
virtually unchanged (r = .33). On the ACQ, 2 items remained
significant ( p < .05, df= 23): how likely is your child to exhibit
self-control in frustrating situations? (r = .47) and when trying to
concentrate, how distractable is your son or daughter? (r = -.54).
Thus, although the association between preschool delay time
and ratings of academically oriented intelligence shares common variance with SAT scores, the correlation with the ratings
of children's ability to cope with social and personal problems
cannot be attributed entirely to their school-related ability or
"intelligence" as assessed by the SAT.
Alternative Analyses

The results reported so far were based on the children's delay

Table 3 presents the correlations between preschool delay
times with the rewards exposed or obscured and with or withtime and the 11 CCQ items that were judged relevant to the
out ideation suggestions. However, beyond these considerbasic cognitive and attentional processes in the self-imposed
ations, the experiments also manipulated a number of other
delay of gratification situation, as described in the Method secfactors, for example, varying the types of strategies suggested
tion.
from those that make delay easier (e.g, mentally transforming
When the rewards were exposed for attention during the dethe rewards into an abstract, nonarousing representation) to
lay period and when ideation was spontaneous, children who
those that make delay more difficult (e.g., thinking about the
delayed longer tended to be rated higher on such relevant CCQ
rewards in a consummatory, arousing way). Although those maitems as is planful, thinks ahead, is attentive and able to concen- nipulations were the focus of the initial experiments, in the
trate, and uses and responds to reason. These same long-waiting
children tended to be rated lower on such relevant items as is
unable to delay gratification and tends to go to pieces under
2
The only notable sex difference in the longitudinal correlates on
stress, becomes rattled and disorganized. In contrast, when the
any of the measures occurred for SAT verbal scores. Although prerewards were obscured and no ideation was suggested (the conschool delay time did not predict these scores among boys in the exdition expected to be unpredictive), none of the 11 relevant
posed-reward-spontaneous-ideation condition (r = .02, N = 17), it was
items correlated significantly with preschool delay time. In conpredictive for girls (r = .74, N - 18), and the difference was statistically
ditions in which ideation was suggested, none of the 11 relevant
significant (z = 2.50, p < .05). However, the difference was smaller and
items were significant.
insignificant for SAT quantitative scores (boys: r = .40; girls: r = .71).

983

DELAY OF GRATIFICATION

Table 2
Correlations Between Preschool Delay Time and the Adolescent Coping Questionnaire
Spontaneous ideation

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Adolescent Coping Questionnaire items

How likely is your child to be sidetracked by
minor setbacks?
How likely is your child to exhibit self-control in
frustrating situations?
How well does your child cope with important
problems?
How capable is your child of doing well
academically when motivated?
How likely is your child to yield to temptation?
Faced with a choice. . . how likely is your child
to settle for the immediate [but less
desirable] one?
How able is your child to pursue his or her goals
when motivated?
How intelligent is your child?
When motivated, how capable is your child of
exhibiting self-control in tempting
situations?
How skilled is your son or daughter at
maintaining friendships and getting along
with peers?
When trying to concentrate, how distractible is
your son or daughter?
How capable is your child of exhibiting selfcontrol when frustrated?
How effectively does your child typically pursue
goals?
How well can your child divert attention from
the frustrativeness of having to postpone
a desired gratification while continuing to
pursue it?
Sample size

V<.05. **p<m.

Suggested ideation

Rewards
exposed

Rewards
obscured

30*

-.01

.19

-.09

58***

-.12

.05

.27

31*

-.10

-.10

-.09

.19
.09

.19
.39

.16
-.09

.11

.23

-.25

.38*
.42**

.11
.15

.03
-.06

.13
.30

.36*

-.32*

-.13

.39*

10

-.16

.14

.00

41**

.08

.09

-.02

40**

.07

-.16

.38*

-.10

-.03

.31

32*

.06

-.08

.37

43

42

21

28

37*
50***
-.32*

21

Rewards
exposed

Rewards
obscured

***p<.ooi.

present analysis of individual differences they essentially constitute random "noise." Accordingly, as an alternative to using
the actual delay times, one could also statistically remove the
experimental "noise" by subtracting from each subject's delay
time the expected delay time for the type of delay situation in
which he or she waited. Such expected delay times can be approximated by averaging the delay times in each distinct type
of delay situation, although the number of subjects for many
types of situations was very low (e.g, less than 5) after eliminating those subjects who had already participated in other types
of delay situations. The results after this procedure were virtually identical to those reported in Tables 2 to 4. The number
of significant correlates among the items of the ACQ in each
condition remained the same, with the exception of one less
significant correlation in the obscured-rewards-spontaneousideation condition. With regard to the SAT scores, the results
were again virtually identical except that the negative (but not
significant) correlations reported in Table 4 became somewhat
weaker. For the 11 CCQ items judged to be relevant, the number of significant correlates remained the same in each condition except that in the obscured-rewards-suggested-ideation
condition an additional item (overreacts to minor frustration)

was negatively and significantly correlated with delay time.

Thus, both the results using actual delay times and those adjusted for experimental main effects were extremely similar.
Discussion

Identifying the Locus ofPredictability

The obtained results help to specify when individual differences in preschool delay of gratification in the present paradigm are predictive of later competence and coping. Expectations for the specific "diagnostic" condition were based on
previous experimental analyses of the role of attention and ideation during delay, as discussed earlier. Consistent with these
expectations, when no strategies were suggested to the child,
ratings of cognitive, self-regulatory, and coping competence, as
well as SAT scores obtained years later, were significantly predicted by preschool delay time when the rewards were exposed
during the delay period, but not when they were obscured.3
3
This pattern of results is also consistent with an unpublished earlier study (Zeiss & Mischel, 1982) showing that parents' estimates of

984

Y. SHODA, W MISCHEL, AND P. PEAKE

Table 3

Correlations Between Preschool Delay Time and the Parent's Q-Sort

Ratings on Items Judged Relevant
Spontaneous
ideation

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

California Child Q-sort items

Is persistent in activities; does not give up easily.
Is planful, thinks ahead.
Is creative in perception, thought, work, or play.
Is attentive and able to concentrate.
Is reflective; thinks and deliberates before
speaking or acting.
Uses and responds to reason.
Is stubborn.
Is unable to delay gratification, cannot wait for
satisfactions.
Tends to go to pieces under stress, becomes rattled
and disorganized.
Overreacts to minor frustrations; is easily irritated
and/or angered.
Becomes anxious when the environment is
unpredictable or poorly structured.

Mean Rewards Rewards Rewards Rewards

levance exposed obscured exposed obscured
2.00
1.57
1.43
1.43

.21
.36*
.03
.39**

-.04
-.01
.03
.06

.29
.19
.14
.05

.18
.09
-.10
.27

1.00
1.00
1.00

.22
.43**
.02

-.15
.03
-.06

-.18
-.05
-.14

.16
.27
-.04

-2.57

-.34*

.21

.09

-.02

-2.00

-.34*

.08

-.14

-.24

-1.43

-.25

.06

.03

-.30

-1.29

.01

-.09

-.15

-.15

48

50

32

35

Sample size

*p<.05.

Suggested
ideation

**p<.01.

In conditions in which various cognitive strategies had been

deliberately suggested to the children for use during the waiting
period, their own self-instructions and cognitive activities were
not entirely spontaneous, and we expected that their delay time
would less clearly reflect their ability to generate such strategies. Indeed, when ideation strategies were suggested, preschool delay time did not strongly or consistently predict later
ratings of self-control, SAT scores, or relevant CCQ items.
Comparisons Across Follow-Up Waves
The present findings provided a more specific analysis than
was possible in the earlier study that used the first, smaller
sample (Mischel et al, 1988). In that study, children's preschool
delay behavior was significantly related to adolescent outcomes, without taking specific conditions into account. If
"diagnosticity" occurs primarily when rewards are exposed and
ideation is spontaneous, as the results of the present study suggest, why was delay time correlated significantly with many
items in the earlier study, which collapsed the conditions?
When the earlier, smaller sample was examined with regard to
conditions, the number of children in each condition became
too small, of course, for confident comparisons. With this in
mind, consider first the 4-item coping measure available in the
first-wave follow-up, a predecessor of the 14-item expanded version used in the present study. On the 4-item measure, the only
significant correlation of preschool delay of gratification behav-

ior (with academic competence, r = .73, N= 13, p < .01) was

obtained in the exposed-rewards-spontaneous-ideation condition. With regard to the CCQ items, there appeared to be strong
relations between delay time and a number of the items in most
conditions. As the sample size increased with the second-wave
follow-up, however, most of these correlations proved to be
insignificant, with the notable exception of the theoretically
predicted condition described in the present study.
It is not clear whether the weakening of these correlations in
the "nondiagnostic" conditions reflects any systematic differences between those who responded in the first mailing and
those who responded only in the second mailing. There are a
number of differences between the first and the second followup: Children were older at the time of the second follow-up, and
therefore the time span from their preschool delay experiment
was greater; the mean preschool delay time of the children of
the second follow-up sample was somewhat shorter (see Note 1);
the 14-item ACQ was introduced for thefirsttime in the second

Table 4
Correlations Between Preschool Delay Time and
Scholastic Aptitude Test (SAT) Scores
Spontaneous ideation
Measure
SAT Verbal

Suggested ideation

Rewards
exposed

Rewards
obscured

Rewards
exposed

Rewards
obscured

.42*

-.12
-.31
33

-.40
-.26
14

-.21
-.23
12

how long their preschool child will wait correlated significantly with
.57**
SAT Quantitative
the child's concurrent delay time in a exposed-rewards condition (r =
Sample size
35
.32, iV= 59, p < .05) but not in a obscured-rewards condition (r = - . 13,
*p<.05. **/?<.001.
Ar=29,/>>.10).

985

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DELAY OF GRATIFICATION

mailing; and, as mentioned in the Method section, the CCQ

was administered without forced distribution in the second
mailing. It is possible that these factors reduced the correlations
between preschool and adolescence in general. However, there
is no reason why any of these factors should be expected to
affect longitudinal predictability differently for children who
happened to have been in different preschool delay conditions.
Thefindingsof theoretical interest, for example, the difference
in adolescent correlates of preschool delay in the rewards-exposed and the rewards-obscured conditions when no ideation
was suggested, thus do not seem interpretable as due to possible
differences in the samples and adolescent measures used in the
two follow-up mailings.

Nature of the Long-Term Links

We must emphasize the need for caution in the interpretation
of the total findings linking preschool delay to adolescent outcomes. This caution applies especially in the interpretation of
the associations between preschool delay in the exposed-rewards-spontaneous-ideation condition and SAT scores. On the
one hand, our faith in the validity of the links between preschool delay behavior and later relevant competencies is
strengthened because the objective test results were consistent
with the parental rating data. On the other hand, even the highest correlations account for only about 25% of the variance. In
addition, given the smallness of the sample, the obtained coefficients could very well exaggerate the magnitude of the true
association. For example, in the diagnostic condition, the 95%
confidence interval for the correlation of preschool delay time
with SAT verbal score ranges from .10 to .66, and with SAT
quantitative score, the confidence interval ranges from .29 to
.76. The value and importance given to SAT scores in our culture make caution essential before generalizing from the present study; at the very least, further replications with other populations, cohorts, and testing conditions seem necessary next
steps.
We also do not wish to overgeneralize or overstate the role of
reward exposure. Although reward exposure seems to create a
more diagnostic condition for assessing the competencies of
interest, these results were obtained from a population of middle-class preschool children not selected for any self-regulatory
difficulties, in a relatively narrow age span (mean age = 4 years,
4 months; SD = 6 months). As children become older, their
delay behavior rapidly becomes less responsive to this manipulation, and it may lose its diagnostic potential early in the
course of development. Many children become aware of basic
rules to facilitate delay, for example, by distracting their attention from the rewards quite early in development (Mischel &
Mischel, 1983), and can purposefully influence their own ideation to overcome the impact of exposed rewards. Consequently, any diagnostic "window" provided by the exposed-reward condition may be fragile and narrow in time. In fact, in
children 6 years of age and older, no differences in delay time
were found when rewards were exposed versus covered (Rodriguez, Mischel, & Shoda, 1989). It is also possible that the effects
of reward exposure interact with the particular characteristics
of the subject population, and such interactions will require
systematic exploration in future work.
A difficult question that remains is the mechanism underlying the associations found between the delay behavior of the

preschool child and the subsequent outcome measures. One

contributing source may be stability in the subjects' family-mediated environments (e.g., Greenberger, Steinberg, & Vaux,
1982; Holahan & Moos, 1986; Lefcourt, Martin, & Saleh, 1984).
For example, stability in parental child-rearing practices and in
the psychosocial environment in the family and the community
may be a common factor underlying both preschool children's
delay of gratification behavior and their cognitive and self-regulatory competence in adolescence. These commonalities may
contribute to the observed long-term correlations.
In our view, the association found between preschool delay
behavior and adolescent competencies may reflect at least in
part the operation of "cognitive construction competencies"
(Mischel, 1973, pp. 265-267). In this view, the qualities that
underlie effective self-imposed delay in preschool may be crucial ingredients of an expanded construct of "intelligent social
behavior" that encompasses social as well as intellectual knowledge, coping, and problem-solving competencies (e.g. Brown &
DeLoache, 1978; Cantor & Kihlstrom, 1987; Flavell, 1982;
Sternberg, 1979). The fact that a wide variety of adolescent
outcomes was predicted by preschool delay behavior is consistent with this interpretation because competencies are likely to
have more diverse, relatively long-term consequences than
other person variables (Mischel, 1968,1973,1990). These competencies, rather than consisting of a global mental entity, may
include relatively specific component skills and processes necessary for effective self-regulation (e.g, Brown & DeLoache, 1978;
Cantor & Kihlstrom, 1987). Some clues about their specific
nature come from related studies. For example, childrens' metacognitive understanding that they must divert attention from
the rewards and generate distracting thoughts or must transform the rewards mentally (Mischel & Mischel, 1983), and their
spontaneous execution of such strategies, was correlated with
their concurrent delay of gratification behavior (Rodriguez,
Mischel, & Shoda, 1989). These are the same components identified as relevant in the experimental studies of the delay process itself (Mischel, 1974,1981).
Although cognitive and attentional strategies and skills play
an important role in the delay situation used in the present
study, there is also much evidence that other factors, such as
motivational and temporal considerations, expectations, and
personality variables are likewise germane for a comprehensive
analysis of delay of gratification (Funder & Block, 1989; Mischel, 1958,1961a, 1961b; Mischel and Gilligan, 1964; Mischel
& Patterson, 1978; Schwartz, Schrager, & Lyons, 1983). For example, when offered choices between less desirable but immediate outcomes and more desirable but delayed outcomes, the
decision to delay or not to delay hinges, in part, on the individual's values and expectations with regard to the specific contingencies (e.g, Koriat & Nisan, 1978; Mahrer, 1956; Mischel,
1966; Mischel & Metzner, 1962; Mischel & Staub, 1965). In a
given situation, therefore, postponing gratification may or may
not be a wise or adaptive choice. However, unless children are
able to sustain delay for desired goals when they want to do so,
their freedom to make that choice risks becoming illusory.

References
Ainsworth, M. D. S. (1979). Infant-mother attachment. American Psychologist, 34, 932-937.
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

986

Y. SHODA, W MISCHEL, AND P. PEAKE

Brown, A. L, & DeLoache, J. S. (1978). Skills, plans, and self-regulaMischel, W (1990). Personality dispositions revisited and revised: A
tion. In R. S. Siegler (Ed.), Children's thinking: What develops? (pp. view after three decades. InL. A. Pervin(Ed), Handbookofpersonal3-36). Hillsdale, NJ: Erlbaum.
ity: Theory and research (pp. 111-134). New York: Guilford Press.
Cantor, N, & Kihlstrom, J. F. (1987). Personality and social intelligence. Mischel, W, & Baker, N. (1975). Cognitive appraisals and transformaEnglewood Cliffs, NJ: Prentice-Hall.
tions in delay behavior. Journal ofPersonality and Social Psychology,
31, 254-261.
Caspi, A., Elder, G. H, & Bern, D. J. (1987). Moving against the world:
Life-course patterns of explosive children. Developmental Psychol- Mischel, W, & Ebbesen, E. B. (1970). Attention in delay of gratification. Journal of Personality and Social Psychology, 16, 329-337.
ogy, 23, 308-313.
Mischel, W, Ebbesen, E. B., & Zeiss, A. R. (1972). Cognitive and attenErickson, H. F, Sroufe, L. A, & Egeland, B. (1985). The relationship
tional mechanisms in delay of gratification. Journal of Personality
between quality of attachment and behavior problems in preschool
and Social Psychology, 21, 204-218.
in a high-risk sample. Monographs of the Society for Research in
Mischel, W, & Gilligan, C. (1964). Delay of gratification, motivation
Child Development, 50,147-166.
for the prohibited gratification, and responses to temptation. JourFlavell, J. H. (1982). Structures, stages, and sequences in cognitive denal ofAbnormal and Social Psychology, 69, 411-417.
velopment. In A. Collins (Ed.), Minnesota Symposia on Child PsyMischel, W, & Metzner, R. (1962). Preference for delayed reward as a
chology (Vol. 15, pp. 1-28). Hillsdale, NJ: Erlbaum.
function of age, intelligence, and length of delay interval. Journal of
Funder, D. C, & Block, J. (1989). The role ofego-control, ego-resiliency,
Abnormal and Social Psychology, 64, 425-431.
and IQ in delay of gratification in adolescence. Journal ofPersonality
Mischel, W, & Mischel, H. N. (1983). Development ofchildren's knowland Social Psychology, 57,1041-1050.
edge of self-control strategies. Child Development, 54, 603-619.
Funder, D. C, Block, J. H, & Block, J. (1983). Delay of gratification:
Mischel, W, & Moore, B. (1980). The role of ideation in voluntary delay
Some longitudinal personality correlates. Journal ofPersonality and
for symbolically presented rewards. Cognitive Therapy andResearch,
Social Psychology, 44,1198-1213.
4,211-221.
Greenberger, E, Steinberg, L. D, & Vaux, A. (1982). Person-environMischel, W, & Patterson, C. J. (1978). Effective plans for self-control in
ment congruence as a predictor of adolescent health and behavioral
problems. American JoumalofCommunity Psychology, 10,511-526. children. In W A. Collins (Ed.), Minnesota Symposia on Child Psychology (Vol. 11, pp. 199-230). Hillsdale, NJ: Erlbaum.
Holahan, C. J, & Moos, R. H. (1986). Personality, coping, and family
Mischel, W, Shoda, Y, & Peake, P. K. (1988). The nature of adolescent
resources in stress resistance: A longitudinal analysis. Journal of
competencies predicted by preschool delay of gratification. Journal
Personality and Social Psychology, 51, 389-395.
Kagan, J, & Moss, H. A. (1962). Birth to maturity: A study in psychologi- of Personality and Social Psychology, 54, 687-696.
Mischel, W, & Staub, E. (1965). The effects of expectancy on waiting
cal development. New York: Wiley.
and working for largerrewards.Journal of Personality and Social
Klineberg, S. L. (1968). Future time perspective and the preference for
delayed reward. Journal ofPersonality and Social Psychology, 8,253- Psychology, 2, 625-633.
Quattrone, G. A., & Tversky, A. (1984). Causal versus diagnostic con257.
tingencies: On self-deception and on the voter's illusion. Journal of
Koriat, A., & Nisan, M. (1978). Delay of gratification as a function of
Personality and Social Psychology, 46, 237-248.
exchange values and appetitive value of the rewards. Motivation and
Rodriguez, M, Mischel, W, & Shoda, Y (1989). Cognitive person variEmotion, 2, 375-390.
ables in the delay of gratification of older children at risk. Joumalof
Lefcourt, H. M, Martin, R. A., & Saleh, W. E. (1984). Locus of control
Personality and Social Psychology, 57, 358-367.
and social support: Interactive moderators of stress. Journal of PerSchack, M. L, & Massari, D. J. (1973). Effects of temporal aids on delay
sonality and Social Psychology, 47, 378-389.
of gratification. Developmental Psychology, 8,168-171.
Mahrer, A. R. (1956). The role of expectancy in delayed reinforcement.
Schwartz, J. C, Schrager, J. B, & Lyons, A. E. (1983). Delay of gratificaJournal of Experimental Psychology, 52,101-105.
tion by preschoolers: Evidence for the validity of the choice paraMischel, W (1958). Preference for delayed reinforcement: An experidigm. Child Development, 54, 620-625.
mental study of a cultural observation. Journal of Abnormal and
Sternberg, R. J. (1979). The nature of mental abilities. American PsySocial Psychology, 56, 57-61.
chologist, 34, 214-230.
Mischel, W. (1961a). Preference for delayed reinforcement and social
Toner, I. J, & Smith, R. A. (1977). Age and overt verbalization in delay
responsibility. Journal of Abnormal and Social Psychology, 62,1-7.
maintenance behavior in children. Journal of Experimental Child
Mischel, W (1961b). Delay of gratification, need for achievement, and
Psychology, 24,123-128.
acquiescence in another culture. Journal of Abnormal and Social
Tversky, A, & Hutchinson, J. W (1986). Nearest neighbor analysis of
Psychology, 62, 543-552.
psychological spaces. Psychological Review, 93, 3-22.
Mischel, W (1966). Theory and research on the antecedents of self-imWalls, R. T, & Smith, T. S. (1970). Development of preference for deposed delay ofreward.In B. A. Maher (Ed.), Progress in experimental
layed reinforcement in disadvantaged children. Journal of Educapersonality research (Vol. 3, pp. 85-132). San Diego, CA: Academic
tional Psychology, 61,118-123.
Press.
Waters, E, & Sroufe, L. A. (1983). Social competence as a developmenMischel, W (1968). Personality and assessment. New York: Wiley.
tal construct. Developmental Review, 3, 79-97.
Mischel, W (1973). Toward a cognitive social learning reconceptualiza- Wright, J. C, & Mischel, W (1987). A conditional analysis of disposition of personality. Psychological Review, 80, 252-283.
tional constructs: The local predictability ofsocial behavior. Journal
Mischel, W (1974). Processes in delay of gratification. In L. Berkowitz
of Personality and Social Psychology, 53,1159-1177.
(Ed.), Advances in experimental social psychology (Vol. 7, pp. 249- Zeiss, A, & Mischel, W (1982). The role of dispositions and conditions in
292). San Diego, CA: Academic Press.
childrens voluntary delay behavior: Bing Consistency Study. UnpubMischel, W (1981). Metacognition and the rules of delay. In J. H. Flalished manuscript, Department of Psychology, Stanford University,
vell & L. Ross (Eds.), Social cognitive development: Frontiers and Stanford, CA.
possible futures (pp. 240-271). New \brk: Cambridge University
Received January 12,1989
Press.
Revision received May 27,1990
Mischel, W (1984). Convergences and challenges in the search for conAccepted May 30,1990
sistency. American Psychologist, 39, 351-364.