The Person By Situation Interaction Myth: Implications For the Definition of Situations

Joseph A.Cote, Washington State University
ABSTRACT - This paper examines the ability of common versus individualized situational effects to predict behavior. The results show a greater generalizability of situational responses than previously thought. Some implications for the definition of situations are discussed.
[ to cite ]:
Joseph A.Cote (1986) ,"The Person By Situation Interaction Myth: Implications For the Definition of Situations", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 37-41.

Advances in Consumer Research Volume 13, 1986      Pages 37-41

THE PERSON BY SITUATION INTERACTION MYTH: IMPLICATIONS FOR THE DEFINITION OF SITUATIONS

Joseph A.Cote, Washington State University

[The comments of James McCullough, Wendy Bryce, and an anonymous ACR reviewer are gratefully acknowledged.]

ABSTRACT -

This paper examines the ability of common versus individualized situational effects to predict behavior. The results show a greater generalizability of situational responses than previously thought. Some implications for the definition of situations are discussed.

INTRODUCTION

Past research on situational variables has assumed that situational effects depend upon the individual rather than being similar across individuals. This assumption has never been well tested in the consumer behavior area, yet has been used as a bases for defining and evaluating situational effects. The purpose of this paper is to discuss the generalizability of situational effects. Given the results, the current definitions of situations are reexamined

Much of the work in situation research has argued that people react differently to a given situation, and it is this interaction between the person and the situation that explains most of the variance in behavior (Bowers 1973; Endler 1975). The basis for this claim comes from ANOVA studies of situations. In these studies, the person by situation interaction is often found to explain behavior better than either the person main effect or the situation main effect (Bowers 1973; Ekehammer 1974; Magnusson 1975; Sarason et al. 1975). These results have been used to claim that situational effects depend upon the individual. The belief is that the individualized effects of situations are more important (explain more behavior) than the common effects of situations

The importance of the individualized effects of situations has been used as support for a subjective definition of situations (Lutz and Kakkar 1975, p. 440).

In restricting the definition of situations to 'observable aggregate effects,' Belk differs with most situational theorists. For example, Cottrell (1950) stated (rather categorically): 'We are quite certain that individuals and groups react to their own definitions of situations...' (p. 711), thus suggesting a subjective, rather than an objective definition.

Are Situational Effects Always Subject Specific

The conclusion that situational effects depend upon the subject may be misleading. The common effects of situations can be defined as the extent to which subjects behave similarly in a given situation. If a situation has a common effect on subjects, then all subjects should behave the same in that situation. The opposite position maintains that situational effects are individualized, in other words, subjects behave differently in a given situation.

When consumer researchers began to examine situation effects they simply replicated the ANOVA procedure used in the psychology literature. The only difference was that consumer researchers examined a set of product choices rather than examining a single trait. This change, however, affects how the ANOVA results are interpreted (Bonner 1984). The situation main effect no longer measures common situational effects. Instead, the situation by response interaction indicates the common effect for situations (Bonner 1984). A similar problem exists when examining individualized situational effects. When trait categories are used the person by situation interaction indicates individualized situational effects. However, when product categories are used, the person by situation by product interaction indicates individualized situational effects.

Unfortunately, very few ANOVA studies in consumer behavior have reported the person by situation by response interaction. A reexamination of two of these studies (Belk 1974; Sandell 1968) indicates that common situational effects are a major predictor of behavior. In the Belk study, common situational effects accounted for 18.7% of the variance in behavior while individual effects accounted for 3.48. In the Sandell study common situational effects accounted for 39.88 of the variance while individual effects accounted for 27.8%. These results indicate that while individualized situational effects exist, there are also significant common situational effects. In fact, these common effects may be as important (predict behavior as well or better) as the individualized effects.

Other evidence indicates that situations have common effects across individuals. Srivastava (1980) found that situational factors were stable across individuals and that individuals had similar response patterns to situations. In the ANOVA studies, situation scenarios are presented and subjects are asked to imagine their response to that situation. Therefore the situation by response interaction is the appropriate measure of common situational effects. In the regression studies, however, the occurrence or expected occurrence of a situation is measured rather than imagined. Some subjects experience the situation; others do not. In this case, the appropriate measure of common situational effects is the ability of occurrence to explain behavior. In other words, if a situation occurs, the regression coefficient indicates the effect of the situation for all subjects. Regression studies have also shown that situational effects can be similar across subjects (Wicker 1971; Bearden and Woodside 1976, 1978; Greer 1977; Cote, McCullough and Reilly 1985). These studies only included situation main effects situation by person interactions were not included or were not found to be significant.

HYPOTHESES

The results discussed above raise two important questions. Do individuals have similar perceptions about the occurrence of a given situation and do they react similarly to that situation. An exploratory study was conducted to determine the extent to which consumers perceive and react similarly to situations. Both individualized and common situational variables were used to predict behavior.

Past research indicates that consumer reactions may be similar in a given situation. But if a situation is to have a common effect on behavior, then consumers must agree that the situation has occurred. For example, if all subjects have a major paper due before a social event, they should all indicate a high likelihood of the situation occurring. Therefore

H1: If a situation is certain to occur, subjects should all perceive that the situation will occur.

Under normal circumstances, individualized situational effects would be expected to predict behavior better than common effects. This is true because individual effects allow for a subject's idiosyncratic behavior. However, this is only important when each subject has a completely different view of the world. In many situations, segments of people may have fairly similar views of the world. Within these market segments, common situational effects would exist.

As stated earlier, past research supports the idea that situations may have common effects on consumers. However, there has never been a direct comparison of individualized versus common situational effects. Hypothesis two is a explicit attempt to compare the predictive ability of common situational effects to individualized situational effects.

H2: Common situational effects should predict behavior as well as individualized situational effects.

If subjects respond similarly to a particular situation, then adding individualized situational effects should not improve our ability to predict behavior. Simply knowing that the subject experienced the situation would provide enough information to predict behavior. If subjects each have a different reaction to a particular situation, then behavior can not be predicted unless the individualized effect is included.

METHODOLOGY

Subjects

Data were collected on the intentions and actual attendance of 61 students to a social function (class party held by the instructor). Measures of intentions to attend the social function, the likelihood of various situations occurring, and the expected influence of the situations were collected at three different times over a two month period. A paper and pencil questionnaire was administered six weeks before the social function (long term expectations), repeated three weeks before the social function (medium term expectations), and one day before the social function (short term expectations. Attendance at the social function and the actual occurrence of the situations was measured three days after the social function.

Intentions and Behavior

Intentions were measured using the following likelihood scale: "Given the information provided, how likely is it that you will attend the function?"

FIGURE

Intention values ranged from 0 (not come) to 100 (will come) depending on the distance the subject's mark was from the endpoint of the scale. Actual attendance was measured using a yes (1)/no (0) question three days after the social function.

Situations

The list of relevant situations was generated using focus group interviews. Students were asked to describe any situations they thought would affect their attendance to a social function. The researcher then reduced this list by combining similar situations and dropping variables that did not fit Belk's (1974) objective definition of situations. The following list of situations was used in the study:

1) Function is scheduled just before a big examination or class paper.

2) Function is scheduled just after a big examination or class paper.

3) Other functions or activities scheduled (not scheduled) at the same time as this one.

4) Student forgets about the function.

5) Directions are poor.

6) There are not many people at the function.

7) Ability to get transportation to the function.

8) Student gets involved in other things.

9) Bad weather (Social function occurred in the winter).

10) Friends or spouse want to do something else that night.

11) Any other situations.

To assess the individual effect of situations, subjects were asked to report the expected effect of a situation if it occurred. The questions on the expected effects followed the following format, "The function is scheduled just before a big examination or class paper. This would:"

FIGURE

Scores ranged from 0 (Insure not come) to 100 (insure would come) depending on the distance the subject's mark was from the endpoint of the scale. The influence of "other situations" was measured by asking the subject to indicate any other situations they expected to influence their behavior and the extent of that influence.

In addition, the expectation of the situation occurring was also measured. This question was worded as follows, "What is the likelihood that you will have a big exam or class paper just before the function?"

FIGURE

Scores ranged from 0 to 1008 chance of occurrence depending on the distance the subject's mark was from the endpoint of the scale. The situational questions were repeated for each of the situations studied. The actual occurrence of the situations was measured using a simple yes (1)/no (0) format.

It was assumed that changes in intentions are attributable to unexpected situational effects (Cote, McCullough, and Reilly, 1985). Hence only changes in the expected effects or likelihood of occurrence were used in the analysis. Unexpected situational occurrences were defined as the difference between the expected probability of the situation occurring in the previous time period subtracted from the expected probability of the situation occurring in the current time period. Figure 1 outlines the model used in the study.

FIGURE 1

SITUATIONAL MODEL

The first step in the analysis was to develop common and individualistic measures of unexpected situations. Common situational effects assume that all subjects will react the same in a given unexpected situation. The only information needed to determine common situational effects is whether the subject experienced the situation unexpectedly. If two subjects both unexpectedly experienced the situation, then the situational effect would be the same (assuming common effects). Since not all situational effects were controlled or known, self-reported unexpected situations (as defined above) were used as the measure of situational occurrence. Regression or discriminant analysis was then used to estimate the common situational effects given the occurrence of unexpected situations. As long as subjects had the same perceptions about the situational occurrences, the analysis assumed the effect of the situation on behavior was the same for all subjects. Therefore, common situational effects were estimated using the following equation.

BIt = a + bt-1(BI1) + 3i (bi * US)

or

B = a + b1(BI) + 3i (bi * US)

where:

B=Behavior (attend/not attend)

BI=Intention

US=Unexpected Situational Occurrences

t=time period 2 and time period 3

i=Situation 1 to 11

bi=Common Effect on Behavior (BI) for a given Unexpected Situational Occurrence i

Individualized situational effects were assessed by including the reported effect of the situation on behavior separately for each subject . This measure includes the subject's individual reaction to the situation. For example, two subjects could report that a given unexpected situation had occurred, but one might report that the situation had no effect on behavior while the other might report a large effect on behavior. The individualized measure needs to include both the unexpected occurrence of the situation and the effect of that situation on the subject. Therefore individualized situational effects were assessed by multiplying the expected effect of the situation by the self-reported occurrence of an unexpected situation. Individualized situational effects were modeled using the following equation.

BIt = a + b1(BIt-1) + 3i (bi * US *SE)

or

B = a + b1(BI) + 3i (bi * US * SE)

where:

B=Behavior (attend/not attend)

BI=Intention

US=Unexpected Situational Occurrences

SE=Self-Reported Situational Effect

t=time period 2 and time period 3

i=Situation 1 to 11

bi=Effect on Behavior (BI) for the Subject's Reported Effect of a Situation i

ANALYSIS

Several situational occurrences were controlled or known. The date for the function was set in period two. The function was scheduled right after a major paper was due in the class. Therefore, subjects should have realized that a major paper was due in the class just before the function. In addition, the weather the night of the function was clear and dry. If subjects had similar and accurate perceptions of reality, then they should have reported a high probability of occurrence for, wa major test or paper due before the function", and a low occurrence of bad weather. The first hypothesis will be supported if the expected occurrences of the major test due before the function is close to 100% and if the occurrence of bad weather is reported to be close to 0%.

The second hypothesis will be tested by predicting "current" intentions and behavior using previous intentions and unexpected situations (each situation included as a separate variable). Stepwise regression will be used to model intentions at each time period. Stepwise discriminant analysis will be used to model actual behavior. Models will first be fit using the common situational effects equation described above. A second set of models will be fit using the individualized situational effects equation described above.

The second hypothesis will be supported if the two sets of models (common effects vs individual effects) predict intentions and behavior equally well.

Limitations

Several limitations should be kept in mind when interpreting the results. First, the relative importance of common versus individual effects can not truly be assessed unless all relevant situations are included in the analysis. Therefore the estimated situational effect in this study will be biased since all relevant situations are not included in the study. Secondly, terms like "just after", and "good weather" may mean different things to different people. Therefore, comparing subject's perceptions to the researcher's perceptions may be inaccurate. Finally, the research assumes that subjects will be able to accurately report the effect a situation may have on them. This may not be the case, especially if the subject does not expect the situation to occur.

RESULTS

Hypothesis one was not well supported by the data. Subjects may not have accurately perceived the known situational occurrences. In period three and four, the expected probability of occurrence should have been close to 100% for, "have a paper due just before the function." The reported expected probability was 28.7% for period three and 24.6 % for period four. This finding may have been caused by the use of the term "just before." The class paper was due the day before the function. Therefore, subjects may not have felt it was "just before" the function. The reported occurrence of bad weather was as expected. No subjects should have stated there was bad weather, and only 0.2% of the subjects did report bad weather.

The data did support hypothesis two. Common situational effects predicted behavior as well as individualized effects. The results of the analysis are presented in Table 1. When examining medium term intentions common situational effects explained 64.7% of the variance while individual situational effects explained 38.7%. The findings for short term intentions were similar with common situational effects explaining 75.3% and individual effects explaining 64.3% of the variance. Finally, common and individual effects explained about the same amount of behavior (attendance) 91.8% and 86.9% respectively.

These results agree with previous studies which indicate that individualized situational effects did not predict behavior better than common situational effects. This supports the idea that there might exist certain market segments that would react similarly to a given situation

DISCUSSION AND SUMMARY

A reanalysis of past research has shown that situations can have common effects across consumers. The results of this exploratory study supports this position. Individualized situational effects do not appear to be needed to accurately predict behavior. However, some subjectivity may be required to model situational effects since subjects may not accurately perceive all situations.

TABLE 1

ABILITY OF PREVIOUS INTENTIONS AND CHANGES IN SITUATIONAL VARIABLES TO EXPLAIN PRESENT INTENTIONS AND BEHAVIOR

Implications for the Definition of Situations

As stated earlier, the use of the subjective definition of situations is based on the possibly false claim that situational effects depend on the person. Given that a basic premise of the subjective definition of situations may be erroneous, it would be wise to reconsider its use

There are several additional problems with the subjective definition of situations. The subjective definition is concerned with whether and how the subject perceives, processes, and responds to the situation. This definition, in fact, contains two separate issues, what is a situation, and what is the subject's response to the situation. Taken to the extreme, the subjective definition of situations is actually a type of behavior because it is concerned with response. It makes sense to separate these issues. We need to first determine what a situation is and then examine the effect of its occurrence

The subjective definition also does not allow for lawlike generalizations. As stated by Mausner (1963, p. 107), !If one specifies the stimulus in terms of the nature of the receiver, lawfulness becomes impossible." Belk (1975, p.429) also notes that the subjective definition may lead to a tautology, "(the subjective definition) tells us no more than the comparably hollow individual differences explanation that this is the sort of behavior engaged in by the sort of person who engages in this sort of behavior."

Given the problems associated with the subjective definition of situations, it would appear reasonable to consider only an objective definition of situations. The objective definition not only has numerous advantages over a subjective definition, but contributes to definitional parsimony by excluding several additional concepts from the definition of situations.

Subjective versus Objective Orientation?

The real question being raised by the subjective versus objective debate is not really how a situation should be defined, but what orientation situation research should take. Lutz and Kakkar (1975) argue that, "there appears to be a need to adopt a subjective view of the situation in order to understand its effects on human behavior." The ability to explain variations in behavior should not be the focus of situation research. Instead, research should focus on why and how situations affect behavior. This clearly has merit, but may lead to two important misconceptions. First, that the definition of situations should include other constructs such as subject responses. Second, that the use of situations to predict behavior is not interesting in itself.

The selection of the appropriate orientation toward situations will depend to some extent on the topic of interest to the researcher. If the researcher is interested in understanding consumer behavior, then the subjective orientation is preferable. If the researcher is interested in how to manipulate situations or what differentiates situations from other types of environmental stimuli, then an objective orientation is more appropriate. When it comes to predicting behavior, the researcher can use either orientation.

REFERENCES

Bearden, William O. and A.G. Woodside (1976), "Interactions of Consumption Situations and Brand Attitudes," Journal of Applied Psychology, 61:6, 764-769.

Bearden, William O. and A. G. Woodside (1978), "Situational and Extended Attitude Models as Predictors of Marijuana Intentions and Reported Behavior," Journal of Social Psychology, 106, 57-67.

Belk, Russell W. (1974), "An Exploratory Assessment of Situational Effects in Buyer Behavior," Journal of Marketing Research, 11, 156-163.

Belk, Russell W. (1975), "The Objective Situation as a Determinant of Consumer Behavior" in M. J. Schlinger (ed), Advances in Consumer Research, Vol 2. Chicago: Association for Consumer Research.

Bonner, P. Greg (1984), "Considerations for Situational Research," Advances in Consumer Research, XII, 368-373.

Bowers, K.S. (1973), "Situationism in Psychology: An Analysis and Critique," Psychological Review, 80, 307-336.

Cote, Joseph, James McCullough, and Michael Reilly (1985), "Effects of Unexpected Situations on Behavior-Intention Differences: A Garbology Analysis," forthcoming, Journal of Consumer Research, (September, 1985).

Cottrell, L.S., Jr. (1950), "Some Neglected Problems in Social Psychology," American Sociological Review, 15, 705-712.

Ekehammer, B. (1974), "Interactionism ln Personality from a Historical Perspective," Psychological Bulletin, 81, 1026-1048.

Endler, Norman (1975), "The Case for Person Situation Interactions," Canadian Psychological Review, 16(1), 12-21.

Greer, Candyce (1977), "The Contribution of Attitudinal and Situational Variables in the Prediction of Family Planning Behavior," Dissertation Abstracts International, 77-14285 (July, 1977, vol. 38(B) #416).

Lutz, Richard J., and Pradeep S. Kakkar (1975), "The Psychological Situation as a Determinant of Consumer Behavior," in M.J. Schlinger (ed.), Advances in Consumer Research, Vol 2. Chicago: Association for Consumer Research.

Magnusson, D. (1975), "The Individual and the Situation in Personality Research," Reports from the Department of Psychology, University of Stockholm, Suppl 30.

Mausner, B. M. (1963), "The Specification of the Stimulus Situation in a Social Interaction," in S. B. Sells (ed), Stimulus Determinants of Behavior. New York: Ronald Press.

Sandell, R.G. (1968), "Effects of Attitudinal and Situational Factors on Reported Choice Behavior," Journal of Marketing Research, 5, 405-408.

Sarason, I. G., R. E. Smith, and E. Diener (1975), "Personality Research: Components of Variance Attributable to the person and the Situation," Journal of Personality and Social Psychology, 32(2), 199-204.

Srivastava, R.K. (1980), "Usage-Situation Influences on Perceptions of Product Markets: Response Homogeneity and its Implications for Consumer Research," in J.C. Olsen (ed.), Advances in Consumer Research, Vol. 7. Ann Arbor, Michigan: Association for Consumer Research, 644-649.

Wicker, Allan W. (1971), "An Examination of the 'Other Variables' Explanation of the Attitude-Behavior Inconsistency," Journal of Personality and Social Psychology, 19, 18-30.

----------------------------------------