Experimental Effects and Person Effects in Delay of ...

Adults. The American Journal of Clinical Hypnosis, 21, 134-147. Samko, M. R., & Schoenfeld, L. S. (1975)9 Hypnotic susceptibility and the Lamaze childbirth experience. American Journal of Obstetrics and Gynecology, 121, 631-636. Samuelly, I. (1972). Lamaze method of child. birth, conditioning or hypnosis. American Journal of Clinical Hypnosis, 15, 136-139. Spiegel, H., & Bridger, A. A. (1970). Manual for the Hypnotic Induction Profile. New York: Soni Medica. Venn, J. (1986). Self-hypnosis and Lamaze method: An exploratory study Manuscript submitted for publication. Wideman, M. V., & Singer, J. E. (1984). The role of psychologicalmechanisms in preparation for childbirth. American Psychologist. 39, 1357-1371. Correspondence concerning this comment should be addressed to Jonathan Venn, 10840 Green Mountain Circle, Columbia,MD 21044.

Experimental Effects and Person Effects in Delay of Gratification David C. Funder and Monica J. Harris Harvard University The behavior of delay of gratification has been a prominent topic of study in psychology in recent years. The research has examined both situational and personality influences on this behavior, and comparisons between these influences have been made. Some years ago Mischel (1968) summarized his own research on delay of gratification in the following way: although often statistically significant, the relations between various individual difference measures and delay preferences generallywere too low to account for more than a tiny fraction of the variance. . . . These weak associations, accounting for a trivial portion of the variance, become understandable when the enormous variance due to situationally specific variables 9. . is recognized. (pp. 82-83) More recently, Ross (1977) wrote: Mischel and his associates have . . . demonstrated that in at least one paradigm of general interest--the "delay of gratification paradigm"--relatively subtle situationalfactors (i.e., the experimenter's suggestion concerning cognitive strategies) overwhelmany individual differences that might be anticipated. (p. 187) Such comparisons of personal and situational effects have important implications, and make Mischel's latest (1984) research on delay particularly interesting. In this most recent presentation, Mischel (1984) summarized evidence that various experimental manipulations of children's ideation (e.g., instructions on what to think about) during a waiting period can indeed have strong effects on their 476

abilities to delay gratification. "A child's momentary mental representation of the outcomes in the delay paradigm influences his or her waiting time and allows us to predict it from knowledge of the psychological conditions and how they operate to influence behavior" (p. 354). However, Mischel also demonstrated the importance of person factors to this behavior, presenting an impressive array of correlations between children's delay performance measured in an experiment when they were aged 4 and Q-sort personality assessments provided by their parents 12 years later (1984, p. 355; comparable patterns were reported by Funder, Block, & Block, 1983). The reported correlations have a median of.29 and range as high as .49. Mischel gave theoretical interpretation to these correlations, implying they are just the ones he would have predicted: "The attributes suggested by the adolescent [personality] ratings are congruent with the cognitive competencies essential for delay revealed by our experimental research" (1984, p. 355). Although Mischel noted that "these correlations seem impressive," he went on to comment that "the magnitude of these relations is modest, leaving most of the variance unexplained" (1984, p. 355). Setting aside the suitability of"percentage of variance" as a measure of the importance of a relationship (cf. Ozer, 1985; Rosenthal & Rubin, 1979), it can be noted that no information pertaining to the "variance explained" by the various experimental studies, nor any other measure of experimental effect, was included in Mischel's article. As it stands, therefore, Mischel's statement could be taken by some readers to imply that experimental effect sizes are uniformly much larger, and even as tending to support earlier summaries of his research such as were quoted above9 We would like to provide the missing information. Table 1 presents a metaanalysis (Glass, McGaw, & Smith, 1981) of the various experimental studies of delay of gratification summarized in Mischel's article. For each study we have calculated the size of the experimental effect in terms of an r that is analogous to the reported correlations involvingpersonality variables. For nonindependent contrasts within single studies, average effects were calculated. Each study offered a host of different possible contrasts, some significant, some not, and some more relevant to the various theoretical hypotheses than others. Because selection was necessary, we chose to calculate effect sizes from only those contrasts most relevant to the effects specifically

mentioned in Mischel's (1984) article. Furthermore, as standard meta-analytic practice, we based our effect sizes only on Fs with a single degree of freedom (Rosenthal, 1984). For example, Mischel (1984) summarized Mischel and Baker (1975) this way: "If the preschoolers ideate about the rewards for which they are waiting in consummatory or 'hot' ways, they cannot delay long" (p. 354). This summary is operationalized by the single degree of freedom contrasts between consummatory relevant images and control, and nonconsummatory relevant images and control (Mischel & Baker, 1975, p. 258), which we averaged for our meta-analysis. In several other cases our procedure involved combining two or three experimental conditions for contrast with another. Of course, if we had simply calculated the size of each overall "effect of experimental condition" by including all reported contrasts, the rs would have been much smaller,just as the average personality effect would have been smaller had we included all correlations, theoretically relevant and irrelevant, significant and insignificant. It can be seen that the average experimental effect sizes vary widely, from a low of .247 to an astonishing .954. No obvious differences in experimental procedure seem to differentiate the studies with larger and smaller effect sizes, but it is noticeable that larger effect sizes seem to be associated with smaller subject sample sizes (r = -.50). We are led to conclude that the highest and lowest figures are best regarded as "outliers" in the distribution of estimates of the effect of ideation on delay. The simple mean of these estimates is .554, the median is .459, and the average weighted-by-sample size (a standard metaanalytic, summary statistic; Mosteller & Bush, 1954) is .448. We do not know whether most readers would consider the magnitude of these relations to be "modest," but like the personality effects they do "leave most of the variance unexplained." If one chooses to use such terminology, an effect size of.448 serves to "explain" only 20% of the variance. The experimental effect sizes are somewhat larger than most of the relationships between delay behavior and personality as assessed 12 years later. It is difficult to evaluate precisely this difference between the effects of immediate experimental manipulations and of assessments of personality as filtered through parents' judgments a dozen years after the behavioral fact. It is also less than obvious how representative these samples of situation and person effects might be of the total universe of possible effect size calculations. Perhaps it is sufficient to make a simple April 1986 9 A m e r i c a n Psychologist

Table 1

Meta-Analysis of Experimental Studies of Delay of Gratification Effect size

Study

df

Mischel & Baker (1975) Average of nonindependent contrasts for consume relevant ideation versus control and transform relevant ideation versus control.

22

.268

28

.504

22

.326

45

.701

Mischel & Ebbesen (1970) Main study: Reward present conditions (combined) versus reward not present. Replication: Reward present versus reward not present Mischel, Ebbesen, & Zeiss (1972) Experiment 1 Play with toy and think fun (combined) versus no distraction Experiment 2 Think fun versus think sad and think rewards (combined)

(r)

23

.566

Experiment 3 Think fun and no ideation (combined) versus think rewards

14

.954

Mischel & Moore (1973) Average of nonindependent contrasts between "relevant image" and "no slide" within the working and waiting conditions

107

.247

Mischel & Moore (1980) Average of five nonindependent contrasts used to "test for differences between groups of particular interest" (p. 218)

81

.305

Moore, Mischel, & Zeiss (1976) Average of the three nonindependent contrasts "of particular interest" (p. 422)

42

.459

observation that requires neither precision nor representativeness: At the very least, the two sets of effect sizes lie on overlapping distributions. To the extent that personality's influence on delay can be disparaged on the grounds of small effect size and "variance explained," at least some experimental effects must come in for equivalent disparagement. It might be wiser to disparage neither these person nor situation effects on the grounds of their effect sizes. A growing body of evidence both empirical (Funder & Ozer, 1983) and statistical (Ozer, 1985; Rosenthal & Rubin, 1979) suggests that effect sizes within the broad range considered here (about .20 to .50) are larger and more important than psychologists have traditionally tended to think. Therefore, although we have tried to provide some important information not included

April 1986 9 A m e r i c a n Psychologist

within the Mischel (1984) article, we offer it in the same overall spirit of showing how both personality factors and experimental manipulations of cognition c~ln be important influences on delay behavior.

Mischel, W., & Baker, N. (1975). Cognitiveappraisals and transformations in delay behavior. Journal of Personality and Social Psychology, 31, 254-261. Mischel, W., & Ebbesen,E. B. (1970). Attention in delayof gratification.Journal ofPersonality and Social Psychology, 16, 329-337. Mischel, W., Ebbesen, E. B., & Zeiss, A. R. (1972). Cognitive and attentional mechanisms in delay of gratification. Journal of Personality and Social Psychology, 21, 204218. Moore, B., Mischel, W., & Zeiss, A. (1976). Comparative effects of the reward stimulus and its cognitiverepresentation in voluntary delay. Journal of Personality and Social Psychology, 34, 419-424. Mischel, W., & Moore, B. (1973). Effects of attention to symbolicallypresented rewards on selfcontrol. Journal of Personality and Social Psychology, 28, 172-179. Mischel, W., & Moore, B. (1980). The role of ideation in voluntary delay for symbolically presented rewards. Cognitive Therapy and Research, 4, 211-221. Mosteller, E M., & Bush, R. R. (1954). Selected quantitativetechniques. In G. Lindzey(Ed.), Handbook of social psychology (Vol. 1, pp. 289-334). Cambridge,MA: Addison-Wesley. Ozer, D. J. (1985). Correlation and the coefficient of determination. Psychological Bulletin, 97, 307-315. Rosenthal, R. (1984). Meta-analyticprocedures for social research. BeverlyHills, CA: Sage. Rosenthal, R., & Rubin, D. B. (1979). A note on percent variance explained as a measure of the importance of effects. Journal of Applied Social Psychology, 9, 395-396. Ross, L. (1977). The intuitive psychologistand his shortcomings.In L. Berkowitz (Ed.), Advancesin experimentalsocialpsychology (Vol. 10, pp. 173-220). New York:AcademicPress. MonicaHarris is supported by a fellowshipfrom the National Science Foundation. Correspondence concerningthis comment should be addressed to David C. Funder, Department of Psychology and Social Relations, Harvard University, 33 Kirkland St., Cambridge, MA 02138. After July 1, 1986, correspondence should be addressed to David C. Funder, Department of Psychology, University of Illinois, 603 East Daniel St., Champaign, IL 61820.

REFERENCES Funder, D. C., Block, J. H., & Block, J. (1983). Delay of gratification: Some longitudinal personalitycorrelates. Journal of Personality and Social Psychology, 44, 1198-1213. Funder, D. C., & Ozer, D. J. (1983). Behavior as a function of the situation. Journal of Personality and Social Psychology, 44, 107-112. Glass, G. V., McGaw, B., & Smith, M. (1981). Meta-analysis in social research. Beverly Hills, CA: Sage. Mischel, W. (1968). Personalityand assessment. New York: Wiley. Mischel, W. (1984). Convergences and challengesin the search for consistency.American Psychologist, 39, 351-364.

M o r e on D e t e r m i n a n t s o f D e l a y o f Gratification John F. Kihlstrom

University of Wisconsin Funder and Harris (this issue, pp. 475476) have raised anew the question of the comparative strength of dispositional and situational determinants of behavior. They show, through a reanalysis of data cited by Mischel (1984), that neither "situational" nor dispositionalvariables account

477