The relationship between daily events and mood: The mood measure ...

2 downloads 139 Views 807KB Size Report
Eileen Kennedy-Moore; Melanie A. Greenberg; Michelle G. Newman; Arthur A. Stone. Article. DOI : 10.1007/BF00995516. Cite this article as: Kennedy-Moore, E.
Motivation and Emotion, Vol. 16, No. 2, 1992

The Relationship Between Daily Events and Mood: The Mood Measure May Matter 1 Eileen Kennedy-Moore, Melanie A. Greenberg, Michelle G. Newman, and Arthur A. Stone 2 State University of New York at Stony Brook

Measures of daily mood have been used as immediate indicators of the effects of the psychosocial environment, the latter concept often measured by daily events. We examined the prediction of two measures of daily mood, the Positive and Negative Affect Schedule (PANAS) and Nowlis Mood Adjective Checklist (MACL), by daily desirable and undesirable events and by day of the week. Unlike prior studies, the event assessment and type of subjects studied (community residents) were the same in both studiea, allowing an attribution of differential associations between the two mood scales and either daily events or day of the week to the different mood measures. The mood measures had similar and expected associations with daily events, although the MACL scale generally had stronger associations with events. Surprisingly, the pattern of day of the week effects for positive, yet not negative, mood were different for the two mood measures. Consistent with the weekend's increase in desirable daily events, MACL positive affect increased on weekends, relative to weekdays, whereas PANAS positive mood decreased on weekends. Also, for both positive and negative affect scales, the MACL scales had stronger associations with day of week than the PANAS scales. These results suggest caution in the choice of mood measure to use in studies of daily events and pose questions about the meaning of these mood measures.

As a means of evaluating the immediate effects of the psychosocial environment, several investigations have predicted end-of-the-day mood from 1This work was supported, in part, by grant MH39234 from NIMH. The authors thank two anonymous reviewers for their helpful comments. 2Address all correspondence, including requests for reprints, to Arthur A. Stone, Ph.D., Department of Psychiatry and Behavioral Science, Putnam Hall, SUNY at Stony Brook, Stony Brook, New York 11794-8790. 143 0146-7239/92/060043143506.50/0© 1992PlenumPublishingCorporation

144

Kennedy-Moore, Greenberg, Newman, and Stone

same-day measurements of events or hassles/upli• (for example, Bolger, DeLongis, Kessler, & Schilling, 1989; Clark & Watson, 1988; Lewinsohn & Libet, 1972; Rehm, 1978; Stone, 1981; Stone & Neale, 1984). This is a reasonable strategy for assessing the impact of events for at least two reasons: First, unlike other outcome measures such as specific symptoms or disease states, which occur infrequently in daily studies of healthy samples, mood is usually assessed as a continuous measure that yields considerable daily variability. Thus, it may be a more sensitive indicator of functioning than more distal outcomes such as illness. Second, increases in dysphoric mood, and possibly positive mood as well (Clark & Watson, 1988), have far-reaching social significance, potentially affecting suicide rates, work productivity, illness and symptom presentation, alcohol abuse, and marital discord. One intriguing possibility is that negative mood is a mediator between the effects of the environment and processes relating to health outcomes (e.g., immune functioning, hormonal levels). Until the late 1980s, mood in daily diary studies was assessed either by a single-point bipolar scale or by one of several mood adjective checklists. The checklists include the Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1971) and the Nowlis Mood Adjective Checklist (MACL; Nowlis, 1965); these questionnaires are very similar to one another in structure and in their relationships with daily events. Recently, Watson and colleagues published a new checklist measure of mood called the Positive and Negative Affect Schedule (PANAS; Watson, Clark & Tellegen, 1988). Unlike the previous adjective checklists in which positive and negative mood subscales tended to correlate negatively, the PANAS was developed specifically to produce positive affect (PA) and negative affect (NA) scales independent of each other. Because the PANAS's psychometric properties are impressive (Watson et al., 1988), it is likely that the PANAS will be widely used in research on daily mood. Investigations of daily life events' ability to predict PANAS scales have shown patterns of association similar to those found in previous research (e.g., Bradburn, 1969; Stone, 1981; Warr, Barter, & Brownbridge, 1983); namely, higher levels of PANAS positive affect are associated with more desirable events and higher levels of PANAS negative affect are associated with more undesirable events (Clark & Watson, 1986; Watson, 1988). There is, however, evidence that the PANAS may show a different relationship with day of the week (DOW) than that found with other selfreported mood scales. Studies with the MACL checklist have found more positive mood and less negative mood on weekends relative to other days in college women (Rossi & Rossi, 1977) and community adults (Stone, Hedges, Neale, & Satin, 1985). In contrast, research with the PANAS has shown a different pattern: PA remained constant across all days of the

Daily Events and Mood Measures

145

week, whereas NA was lower on Sunday than other days (Clark & Watson, 1988). It should be noted that DOW is a strong predictor of desirable (pleasant) and undesirable (unpleasant) daily event frequencies (Stone, 1987). As such, mood might also be expected to vary by D O W because weekends, for instance, have fewer undesirable events and more desirable events than do other days of the week. While this is not a statistical necessity, since the correlations between DOW and events are only modest, it is nevertheless reasonable to expect some DOW effect on mood. It is not clear whether the divergent results with the PANAS vs. other mood checklists reflect methodological differences in the studies (e.g., college student vs. community adults as subjects) or are attributable to the distinctive properties of the different mood measures. We had the opportunity to clarify this issue by comparing the pattern of relationships between mood, daily events, and DOW in two studies that employed very similar study designs and comparable subject samples. In these two studies community residents completed the same daily events inventory (based on Stone and Neale's (1982) Assessment of Daily Experience) and a mood inventory for up to 80 consecutive days. In the first study, the MACL was used; in the second study, the PANAS was used. Thus, we could explore the relationship between DOW and mood as well as between daily events and mood with both mood assessments, holding event assessment and subjects constant. We hoped to shed light on the potentially different relationship that the two mood measures might have with daily events and DOW.

METHOD

Subjects" The two samples will be referred to as MACL subjects and PANAS subjects, depending on which mood measure was used. Both MACL (n = 77) and PANAS (n = 94) subjects were married men, most with children, recruited through mail solicitation and paid for their participation. Subjects participated for an average of 91 days in the MACL study and for an average of 75 days in the PANAS study. Average age was 43.2 (SD = 10.1) for MACL subjects and 42.6 (SD = 6.8) for PANAS subjects. Average educational level was 14.0 (SD = 2.3) for PANAS subjects and college graduate for MACL subjects.

146

Kennedy-Moore, Greenberg, Newman, and Stone

Materials Mood Measures. MACL subjects completed a revised 12-item version of the Nowlis Mood Adjective Checklist (Nowlis, 1965). Due to low factor loadings, one item was dropped (item number 6), leaving the negative engagement factor with five items and the positive engagement factor with six items) Each MACL item was rated on a 3-point scale, with 1 = definitely applies, 2 = slightly applies, 3 = does not apply. In order to make the mood measures comparable, the MACL items were reversed by subtracting them from 4, so that large numbers mean "more" on both mood scales. These reverse-scored items were summarized into positive and negative mood scales (see Stone, 1981). PANAS subjects completed the Positive and Negative Affect Schedule (Watson, Clark & Tellegen, 1988). This is a self-report measure consisting of 20 mood descriptors selected on the basis of previous factor analytic studies as relatively pure markers of either high negative affect or high positive affect (Watson et al., 1988). Subjects were asked to rate the extent to which they were experiencing each mood on a 5-point scale (1 = very slightly or not at all, 5 = very much). The 10 PANAS items assessing PA (interested, excited, strong, enthusiastic, proud, alert, inspired, determined, attentive, active) and the 10 items assessing NA (distressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, jittery, afraid) were summed to yield separate PA and NA scales., Event Measures. Both samples completed the Daily Life Event Assessment (DLE), and 80-item checklist of daily events that is a modification of a 66-item instrument, the Assessment of Daily Experience (ADE), developed by Stone and Neale (1982). Subjects checked which events had occurred that day and rated them on the dimension of desirability-undesirability (6-point scale: 1 = extremely desirable, 6 = extremely undesirable). An important procedural aspect of the DLE is that husbands and wives discuss the husbands' daily events and "agree" on a final set of events that occurred during the day. Prior research had shown that there were many event reporting errors when only one person completed the event checklist; the conjoint procedure was developed to increase confidence that reports reported were, indeed, objective occurrences (Stone & Neale, 1982). Desirability ratings were summarized into a dichotomous desirable/undesirable indicators. The numbers of desirable and undesirable events were summed separately for each study day. 3Negative engagement includes original Nowlis mood adjectives of angry, clutched up, concentrating, skeptical, and sad, while positive engagement includes playful, elated, energetic, kindly, self-centered, and leisurely (Stone, 1981).

Daily Events and Mood Measures

147

Procedure For both samples, demographic information was collected at a training interview. At the end of each study day, both groups of subjects rated their mood and also completed the DLE (Stone & Neale, 1982). They checked which events had occurred that day, and rated the desirability of each of these. The completed forms were mailed back to the experimenters the next day.

RESULTS First, the average levels of mood and events were examined to determine whether these were comparable across the two samples. Means and standard deviations for the two samples on the mood and daily events measures are presented in Table I. PANAS subjects reported a greater number of desirable, t(169) = -31.43, p < .01, and of undesirable, t(169) = -12.11, p < .01 events, compared to MACL subjects. To determine whether the relationships between positive and negative affect were similar across the two scales, average within-person correlations were computed between these two variables for each sample by converting the within-subject correlation coefficients to Fisher Zs, averaging them, then converting the average Z back to a correlation coefficient (el. Michela, 1990). These correlations were 0.01 (SD = 0.24) and -0.36 (SD = 0.25) for the PANAS and MACL, respectively. Thus, as has been found in previous studies, PA and NA were virtually uncorrelated on the PANAS and negatively associated on the MACL. Next, relationships between daily events and daily positive and negative mood were examined, using the within-subject correlational method described above. As can be seen in Table II, the magnitude and direction of relationships between affect and events were somewhat different across the samples, although all of the transformed, average correlations were significantly different from zero. With the exception of the relationship between undesirable events and NA, the magnitude of MACL associations with events was significantly stronger than those with the PANAS (see Table II for t values). There was no significant difference between the samples in average association between undesirable events and NA. The effect of day of the week predicting mood was examined using pooled within-subjects regressions with between-subject differences in level removed (cf. Bolger et al., 1989; Repetti, 1989). Day of the week was a significant predictor of NA in the MACL sample, F(6, 8416) = 65.11, p < .01, and in the PANAS sample, F(6, 7012) = 6.83, p < .01). For both

148

Kennedy-Moore, Greenberg, Newman, and Stone

Table I. Daily Mood Score and Daily Event Frequency Means and Standard Deviations for MACL and PANAS Samplesa

MACL: PA MACL: NA PANAS:PA PANAS: NA Number of desirable events Number of undesirable events

MACL sampteb

PANAS samplec

M

(SD)

M

(SD)

10.20 7.44

(2.36) (1.96) 24.96 12.65 4.77 1.41

(7.37) (3.67) (2.86) (1.57)

4.1t 1.18

(2.51) (1.35)

aMACL = Nowlis Mood Adjective Checklist; PANAS = Positive and Negative Affect Schedule; PA = positive affect; NA = negative affect. bar = 8,515 person-days or slightly less due to missing data. CN = 7,013 person-days.

Table II. Average Within-Subject Transformed Correlations Between Daily Events and Daily Mood a MACL sampleb, PANAS sample c, Difference between r(t) r(t) samples, t Number of desirable events With PA With NA Number of undesirable events With PA With NA

.36

(13.32) a

-.21

(-9.44) d

-.22 (-12.53) a .30 (16.89) d

.20 (8.84)** -.14 (-9.37) `/ -.05 .34

(-2.73) e (17.82) d

3.54d 2.63 6.91 a 1.12

aNote: T-tests in parentheses indicate test of difference from zero of cell mean. The difference between sample t-tests used a standard error term weighted for the different N in the two samples. MACL = Nowlis Mood Adjective Checklist; PANAS = Positive and Negative Affect Schedules; PA = positive affect; NA = negative affect, bN = 77. C-N= 94. dp < .01.

'p < .05.

samples, NA was found to be lower on the weekend compared to on weekdays (see Table III and Fig. 1). Similarly, day of the week significantly predicted PA in both samples [MACL: F(6, 8416) = 65.87, p < .01; PANAS: F(6, 7012) = 7.47, p < .01]. However, the two samples showed different patterns of relationships between PA and day of the week. For the MACL sample, PA was higher on weekends compared to weekdays, whereas for the PANAS sample, PA was lower on Sundays compared to on weekdays.

Daily Events and Mood Measures

149

Table III. Regression Coefficients and Variance Explained by Day of Week Predicting Mood a

PA Mon.

Tues. Wed Thurs. Fri. Sat. Sun. %R2tot

%DR2wkday NA Mon,

Tues. Wed. Thurs. Fri. Sat.

Sun %RZtot %DR2kd~y

MACL

PANAS

sampleb

samplec

-1.08 -1.13 -1.11 -1.05 -0.89 -0.21 0.00

0.62 1.04 0.98 1.t7 1.24 1.17 0,00

36.4% 3.5%

53.1% 0.3%

0,83 0.87 0.82 0.87 0.75 0,23 0.00 35,4% 2,8%

0.51 0.46 0.78 0.63 0.61 0.27 0.00 31.6% 0.4%

aMACL = Nowtis Mood Adjective Checklist; PANAS = Positive and Negative Affect Schedule; PA = positive affect; N A = negative affect. b~V = 8,417 for PA; 8,429 for NA. CN = 7,013.

Beyond the pattern of day-of-week effects on mood, another consideration is the strength of the relationships. Although the overall percent of variance explained (R 2) by the model for PA is higher for the PANAS (53%) than for the MACL (36%), the change in R 2 attributable to the day of week effect is much larger for the MACL sample (3.5%) than for the PANAS (0.3%). This observation also applies to the NA scales: 2.8% percent of the variance is attributable to day of week for the MACL sample and only 0.4% for the PANAS sample. To determine whether a minority of adjectives was creating the differences between the two scales, PA adjectives were examined individually by weekday using within-subject correlations. With the MACL, all but one of the adjectives followed the same pattern such that there were higher levels of those positive moods on weekends compared to on weekdays. The exception was "self-centered," where means were similar on weekends compared to on weekdays. With the PANAS, all but three adjectives followed

150

Kennedy-Moore, Greenberg, Newman, and Stone

2.5

2.0



.....

. •

...

.........



..........



.

.

1.5 I-0 W LL. L~

1,0

W

0.5


_