Journal of Personality and Social Psychology 1975, Vol. 31, No. 3, 473-478

An Attributional Approach to Moods and the Menstrual Cycle Gary F. Koeske University of Pittsburgh

Randi K. Koeske Carnegie-Mellon University

Acceptance of an attribution pattern linking negative moods (depression, irritability) to the approach of menstruation and the likelihood of internal and external attributions were examined in a questionnaire study in which cycle phase (pre- versus postmenstrual), mood (positive versus negative), and environment (pleasant versus unpleasant) were varied. Subjects' ratings indicated that (a) biology was judged important for explaining negative moods occurring premenstrually; (b) inconsistency between mood and environment produced more internal (personality) attributions, while consistency enhanced external attributions; and (c) emotionally expressive behavior was thought to reflect underlying personality dispositions despite extenuating situational factors (assumed personal causation). The theoretical relevance of the findings to a new conceptualization of premenstrual emotionality and to an attributional chain relating female self-concept and premenstrual tension is discussed.

Despite the well-established assumption that hormonal fluctuations underlie cyclic mood changes in women, conclusive evidence of the specific physiological bases of the phenomenon is lacking. While many promising biological leads exist, recent evidence also supports the role of perceptual and social conditioning factors in menstrual cycle mood swings. Paige (1971), for example, found premenstrual and menstrual anxiety were tied more closely to the number of days of menstrual flow and the observance of the sex taboo than to hormonal variations. Presumably socially learned connotations mediated the anxiety. Subjects' self-reports (Koeske, Note 1) suggest that a fairly coherent set of beliefs about menstruation and associated mood changes exists. These beliefs form an attribution pattern linking negative mood swings (depression, irritability) to the approach of menstruation. This attribution pattern may form part of an attributional chain involving physical and emotional symptoms and selfattributions. A conceptualization of the sources of cyclic mood change which builds on this chain of attributions could prove more fruitful than an approach based on either biology or social factors alone. The present study was designed to isolate the attribution pattern linking negative moods to the approach of menstruation and assess its influence on attributions about mood and 473

personality. Stories in which mood, situation, and phase of the menstrual cycle were varied independently provided a test of two hypotheses. Hypothesis 1 predicted that biology would be important in explaining negative (but not positive) moods occurring premenstrually but unimportant for postmenstrual moods. Following attribution theory (Jones & Davis, 196S; Kelley, 1967), it was also predicted (Hypothesis 2) that stronger internal attributions (to the target person's personality) would result when mood was inappropriate to the situation. METHOD Subjects and Procedure Forty-nine male and S3 female introductory social psychology students at the University of Pittsburgh served as subjects. The experiment, administered on the first day of classes by the second author, was presented as a questionnaire study of "how people use information to make decisions." Student reactions to an "excerpt from a normal 1 college student's interview with a (clinical) counselor" were requested. Two levels of each of the three independent variables, menstrual cycle phase, CP, (preThis research was supported by National Institute of Mental Health Grant MH06718. Requests for reprints should be sent to Randi K. Koeske, Department of Psychology, Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213. 1 Instructions stressed that the person was not in need of psychological help but that her history had been presented in a clinical format merely as a means of getting at decision-making styles.

474

RANDI K. KOESKE AND GARY KOESKE

versus postmenslrual), environment, E, (pleasant versus unpleasant), and mood, M, (positive versus negative) were included in the factorial design. Information about a hypothetical female student (Miss A) was contained in five categories. All subjects received identical information about Miss A's background (sex, age, student status), current responsibilities (full-time student, part-time job), and personality (which they were to infer from the content of the interview). Menstrual cycle phase information ("menstrual period expected in 2 days" versus "3 weeks") was provided as medical information, and a summary of the events mentioned by Miss A in the interview was included under current environment. Depending on condition (pleasant versus unpleasant), one of two lists of events appeared: (a) (positive) results of examination received, news of father's recovery, news of brother's absence, apologetic boyfriend; or (b) (upcoming) course examination, news of father's illness, news of brother's return, exasperating boyfriend. The information concerning the brother was always inconsistent with the tone of the other information. Subjects also received an interview excerpt containing Miss A's account of a typical day. For example, the negative mood-pleasant environment condition was created as follows: Well, today was just an average day, I guess. I felt tense although I wasn't rushed for once and kind of tired even though last night was my night off from work. My spirits weren't even lifted after I got the results of my examination. I did pretty well even though I missed some items I really should have known. I should have been relieved about my father too. I got a letter from my mother today and she said my father was doing well after a recent hospitalization. I had been kind of worried about him. But even that news didn't raise my spirits. My mother also said my older brother will not be coming home for the holidays. I'm really sorry that I won't be able to see him. It's been quite a while since we saw each other. I guess I was pretty tired and had a lot on my mind by the time I saw my boyfriend at lunch. He was apologetic about our last quarrel and I really should have felt close to him. I suppose it was silly but instead I just blew up right out in public. By the time the lunch was over, I was on the verge of tears.

and I thought about all that had happened that day.

Dependent Measures Three sets of dependent measures were collected on the pages following the interview excerpt. Assessment of mood. Two separate measures of mood were included. Subjects first rated Miss A's mood by writing a number from 1 (not at all) to 7 (very) on the line next to each of 13 adjectives. The list included such adjectives as happy, angry, and unpredictable. After rereading the final paragraph, subjects were asked to decide what Miss A had meant when she described herself as being in "that kind of a mood" by rating the likelihood (1 = not at all likely; 7 = very likely) with which each of eight moods characterized her feelings. Sample moods were quiet, depressed, and peaceful. Assessment of attribution patterns. Patterns of causal attribution elicited by the interview were evaluated in two measures. Subjects were first required to divide 100% among the five categories of information (background, current responsibilities, medical information, current environment, and personality) according to their contribution to Miss A's mood. A more detailed index was provided by ratings (from 1 to 7) of the importance of each of eight items in determining her mood. The list included sex (female), menstrual cycle phase, the four separate elements in the story which constituted current environment, part-time job, and personality. Characterizations of personality. Subjects were asked to rate Miss A's personality by writing a number from 1 (not at all) to 7 (very) on the line next to each of 17 personality characterizations. The list included descriptions such as argumentative, sensitive, and warm. Following completion of the questionnaire booklets, subjects were debriefed.

RESULTS

The excerpts were written so as to minimize the use of mood or personality terms and were parallel except that in half of the stories Miss A's mood was positive and in half it was negative. Thus her mood was either consistent with the overall tone of the day's events or it was not. All stories concluded with the following paragraph:

Effectiveness of the Mood Manipulation Separate 2 X 2 X 2 multivariate analyses of variance 2 were performed on the two measures of mood. These analyses were undertaken primarily to provide confirmation of the mood (M) manipulation but they secondarily evaluated subjects' use of the cues of cycle phase (CP) and environment (E) for mood perception. The latter findings are reported below. Analyses revealed that mood was successfully manipulated and resulted in significant main effects of M for both measures, F (13, 81) = 19.266, p< .001; F (8, 86) = 25.412,

I spent the evening quietly. I didn't feel like being energetic or going out or anything. So I just sat around and listened to records. I did a lot of thinking—I was in that kind of a mood—

2 Analyses were based on Dean J. Clyde's multivariate analysis of variance program which uses Wilks's Lambda Criterion for multivariate tests of significance.

ATTRIBUTIONAL APPROACH TO MOODS AND THE MENSTRUAL CYCLE TABLE 1 MEANS, STANDARD DEVIATIONS, AND UNIVARJATE t' VALUES FOR THE Two SETS OP MOOD SCALES : EFFECT OF MOOD MANIPULATION Positive mood

Negative mood Univariate F (d/= 1,93)

Mood scales

Happy

Angry Depressed Upset Emotional Excited Anxious Peaceful Joyful Sensitive Energetic Unpredictable Exhilarated Quiet Depressed Peaceful Lazy Unsettled Happy Thoughtful Contented

M

SD

M

SD

5.284 1.252 2.255 1.896 4.712 4.059 2.819 4.665 4.822 4.506 3.308 3.954

1.309 .501 1.368 1.237 1.620 1.499 1.587 1.885 1.642 1.467 1.759 1.979

2.2-55 3.026 5.237 4.823 5.019 2.867 3.865 2.440 1.629 5.235 2.031 3.538

1.052 1.665 1.388 1.395 1.674 1.298 1.819 1.481 .751 1.652 1.392 1.922

4.221 5.625 2.228 5.673 3.325 2.145 4.508 6.025 5.183

1.730 1.483 1.457 1.188 1.945 1.349 1.316 1.267 1.455

1.975 5.515 5.338 3.263 2.640 5.048 1.988 5.230 2.538

1.491 46.125*** .191 1.453 1.426 109.617*** 2.000 54.019*** 3.686 1.900 68.412*** 1.963 .938 117.232*** 5.438* 1.858 1.781 65.193***

150.128***

52.129*** 103.965*** 120.123*** 1.182 15.469*** 10.113*** 43.182*** 139.576*** 5.972* 16.449*** 1.125

Note. Higher mean scores indicate higher ratings on the mood listed. * p < .05. **/> < .01. *** p < .001.

p < .001. Of the 13 mood scales, only emotional and unpredictable failed to reach significance. Among the eight likelihood-of-mood scales six attained significance, one (lazy) was marginally significant, and one (quiet) was nonsignificant. Table 1 presents the means, standard deviations, and univariate F values for both sets of scales. Evaluation of Hypotheses In accordance with established techniques for handling attributional data (e.g., Frieze & Weiner, 1971), separate analyses of variance were performed on each item of the two attribution measures, percentage of contribution,3 and importance. Four of the 13 analyses of variance performed were judged critical to the hypotheses. Hypothesis 1 stipulated significant Cycle Phase X Mood interactions for the categories, medical information (per8 For the percentage measure, scores which did not total 100% were adjusted to preserve the relationships between the categories.

475

centage measure) and menstrual period (importance measure), with an association between negative moods and the premenstruum expected. The second hypothesis that moods inappropriate to the environment would be more internally attributed (to the personality) should be reflected in significant Mood X Environment interactions for the two measures of attribution to personality. The results of the four critical analyses of variance revealed that both hypotheses were confirmed. Significant CP X M interactions were obtained for both menstrual period measures, F (1,94) = 10.472, p < .005 and F (1,93) =4.367, p < .05, for medical information and menstrual period, respectively, and the M X E interactions were significant, F (1, 94) = 5.981, p < .05 and F (1, 92) = 11.2001, p < .005, for percentage contributed and importance of personality, respectively. Inspection of the cell means relevant to the two sets of interactions (Tables 2 and 3) indicated that the effects were in the predicted direction in three out of four cases. For the menstrual period measure, both positive and negative moods enhanced attributions to the approach of the period, with the attribution stronger for negative moods. For medical information, there was again an association between negative moods and the premenstruum, with a lesser association between positive moods and the other cycle phase. Of the two personality measures, only the importance rating showed the predicted effect clearly. The percentage measure indicated that positive moods were more internally attributed TABLE 2 MEAN CONTRIBUTION RATINGS AS A FUNCTION OF MOOD AND MENSTRUAL CYCLE PHASE : MENSTRUAL PERIOD MEASURES Mood

Medical information" 2 days 3 weeks Menstrual periodb 2 days 3 weeks

Positive

Negative

10.00 14.02

21.40 7.80

3.00 2.52

4.43 2.27

ft These numbers represent percentage of contribution. b These numbers represent importance of contribution on a 7-point scale.

RANDI K. KOESKE AND GARY KOESKE

476

TABLE 3 MEAN CONTRIBUTION RATINGS AS A FUNCTION or MOOD AND PLEASANTNESS or ENVIRONMENT: PERSONALITY MEASURES Mood

Personality" Pleasant Unpleasant Personality1* Pleasant Unpleasant

Positive

Negative

15.80

29.38

14.35 12.44

4.40 5.12

4.78 3.23

a l

These numbers represent percentage of contribution. > These numbers represent importance of contribution on a 7-point scale.

regardless of environment, with personality judged most responsible for the inappropriate positive mood. Because the percentage measure did not permit total independence in the category ratings, however, the increased rating given to medical information for negative moods occurring premenstrually may have been responsible for the relatively lower personality rating for negative moods. As can be seen from Table 4, the predicted effect was obtained for the statistically independent importance measure. Other findings based on the total set of analyses of variance, although not specifically predicted, provide further confirmation of the second hypothesis. Subjects readily used the combined M and E cues (M X E interaction) to evaluate the effects of background, F ( I , 94) = 9.55, p < .005, current environment, F (1,94) = 28.347, p < .001, and three aspects of the environment, father, F (1, 94) = 46.453, p< .001, boyfriend, F ( l , 9 4 ) = 16.274, p < .001, and examination, F (1, 94) = 66.738, p < .001. As with the personality measures, background was judged a more potent contributor when mood and environment were inconsistent. Current environment and the three significant elements of it, father, boyfriend, and examination were judged more important when mood was appropriate to the situation. Since the environment involves external factors and background, as personality, can be seen to involve internal factors, these results lend support to the attribution-theoretic expectation that consistencies would be

attributed externally and inconsistencies internally. Other Findings Perception of mood. Analyses indicated that assessments of mood were affected by the independent variable manipulations. Multivariate F values indicated a significant effect of E for the 13-scale set, F (13,81) = 2.126, p < .021, and of CP for the eight-scale set, F (8,86) = 2.114, p < .043. These results reflect the subjects' tendency to judge mood as appropriate to the situation and to give more negative mood ratings when the target person was premenstrual. Univariate F values for only three scales in both sets reached significance, however (anxious, for the 13 mood scales; depressed and happy for the eight likelihood scales) although several were marginally significant. On both measures, the M X E interaction reached significance, F (13, 81) = 4.004, p < .001; F (8, 86) = 3.260, p < .003. Univariate Fs for the following scales were significant: happy, depressed, excited, joyful, sensitive, and unpredictable for the 13-scale set and happy for the eight-scale set. Generally, subjects judged positive moods appropriate to the situation more positive and inappropriate negative moods more negative. Both M and E cues were used to assess mood, with the former usually weighted more heavily. Patterns of attribution. Because of the large number of tests performed for this set of measures, unpredicted effects were evaluated more stringently, using a critical p = .01. The following effects were obtained: 1. Positive moods resulted in more internal (personality) attributions than did negative moods, as evidenced by the significant M effect obtained for both the percentage, F (1, 94) = 8.307, p < .01, and importance, F (1, 92) = 7.193, p < .01, measures of personality. 2. Unpleasant environments produced more internal (personality) attributions, especially when the menstrual period was not expected for 3 weeks—CP X E effect for the percentage measure of personality, F (1, 94) = 50.243, p < .001. Inspection of the cell means indicated that this pattern resulted from the inordinately high ratings obtained for positive

ATTRIBUTIONAL APPROACH TO MOODS AND THE MENSTRUAL CYCLE moods occurring in unpleasant environments and the tendency to rate personality as more influential in producing postmenstrual moods. 3. Premenstrually, the menstrual period, F (1,93) = 10.970, p < .005, was judged to have a greater influence on behavior than postmenstrually. No triple interaction was significant. Characterization of personality. As with the mood scales, a 2 X 2 X 2 multivariate analysis of variance was performed on the 17 personality scales. Only two of the multivariate ^"s were significant. Table 4 presents the means, standard deviations, and univariate Fs for the larger effect, that of mood on personality characterization. The significant M effect, F (17, 78) = 6.063, p < .001, indicated that moodiness was overassimilated to the personality. The display of negative moods led subjects to give more negative personality characterizations (e.g., argumentative, hostile) while the display of positive moods enhanced positive descriptions (e.g., warm, happy-go-lucky). This effect is reminiscent of subjects' well-documented tendency to assume personal causation (e.g., Jones & Harris, 1967) when making attributions. The barely significant M X E interaction, F (17, 78) = 1.784, p < .045, reflected the higher ratings given to characterizations like sensitive, competitive, and assertive when mood was appropriate to the situation. Subjects apparently used both M and E cues to make assessments of personality, usually weighting the former more heavily. DISCUSSION The data reported above generally confirm the two hypotheses under investigation. Subjects display a clear-cut attributional pattern linking negative moods (depression, irritability) to the premenstruum. They generally make more internal (personality) attributions when moods are inappropriate to the situation and emphasize aspects of the situation more when mood is appropriate. Subjects also appear to assume personal causation regardless of extenuating situational factors: the display of emotionally expressive behavior (moodiness) is thought to reflect underlying personality dispositions.

477

TABLE 4 MEANS, STANDARD DEVIATIONS, AND UNIVARIATE F VALUES FOR RATINGS OF PERSONALITY: EFFECT OF MOOD MANIPULATION Positive mood Negative mood Personality scales Argumentative Depressive Sensitive Happy-go-lucky Friendly Competitive Emotional Independent Assertive Unreasonable Moody Impulsive Warm Hostile Pensive Introspective Enthusiastic

Univariate F (df= 1,93) M

SD

M

SD

2.063 2.693 4.955 4.448 5.360 4.447 5.239 4.946 3.848 1.803 4.059 4.184 5.320 1.513 3.896 4.542 4.733

1.270 1.545 1.061 1.773 1.285 1.773 1.320 1.341 1.391 1.095 1.819 1.631 1.225 .927 1.518 1.545 1.468

3.068 4.423 5.639 2.328 4.148 3.614 5.474 4.428 4.168 2.323 4.618 3.654 4.210 2.403 4.174 4.433 3.023

1.495 1.667 1.195 1.487 1.698 1.480 1.262 1.536 1.523 1.278 1.722 1.931 1.523 1.307 1.638 1.659 1.373

12.864*** 28.562*** 10.150** 43.107*** 15.794*** 1.879

An attributional approach to moods and the menstrual cycle.

Acceptance of an attribution pattern linking negative moods (depression, irritability) to the approach of menstruation and the likelihood of internal ...
512KB Sizes 0 Downloads 0 Views