Women's Pursuit of Personal Goals in Daily Life With ...

3 downloads 0 Views 2MB Size Report
Women's Pursuit of Personal Goals in Daily Life With Fibromyalgia: .... fatigue increased across the day, study participants were more likely to report at day's end that ...... challenging situations, they can make the best of a bad situation, they are ...
Women's Pursuit of Personal Goals in Daily Life With Fibromyalgia: A Value-Expectancy Analysis

Glenn Affleck

and Howard Tennen

University of Connecticut Health Center

Paul Karoly

Susan Urrows and Micha Abeles

A n / o n ; . S^ilc L ' n i v t r ;

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.

University of Connecticut Health Center

Eighty-nine women with fibromyalgia coir.oiciixl il-.; i iff O:'?rt.'.;vr

i t - , - , H u m i f i e d health and social

goals, and answered questions from the Goal Systems Assessment Battery (P. Karoly & L. Ruehlrnan, 1995) about iheir valuation of, and self-efficiency

and pain- and fatigue-related

goal barriers.

in attaining, each goal. Foe 30 days, they responded

Goal barriers increased and goal efforts

and progress

decreased on days with greater pain and fatigue; goals valued more highly were pursued more effortfully and successfully: more optimistic individuals were less likely to perceive goal barriers and, on days that were more fatiguing than usual, were less likely to reduce their effort and to retreat from progress in

were less painful than usual.

The pursuit i>i p e r s n n u l j.;nflk 1'if.ire^ j : r u i n h e i i ; l > in p ^ d i i M e e ical adaptation Karoly

&

to

life

with

Ruehlrnan,

chronic

1996).

pain (Affleck

Personal

goals

et

are

al., 1998;

"motivational

goals. The perception of pain-related barriers to goal proprcs-s can thus be as critical a factor in adapting to life with chronic pain as is the pain itself (Affleck et al., 1998).

units" that are less global than broad cognitive representations or

In our study, we targeted the pursuit of medium range health and

personality characteristics, but more integrative than isolated be-

social goals in daily life with chronic pain. These goals concern the

haviors or attitudes (Karoly, 1991). Whether they are called life

improvement or maintenance of health and fitness (e.g., eating a

tasks

low fat diet, following a regular exercise routine) and the enhance-

(Cantor e(

personal projects 1991),

these

which Price

an

al.,

1991),

(Little,

constructs

current

concerns

(Klinger,

1975),

1989), or personal strivings (Emraons, refer

to

an

imagined

end

state toward

individual aspires and which drives voluntary action.

and

Harkins

(1992)

noted

that

although

chronic

pain

is

initially experienced nonreflectively as an intrusion, it may eventually be processed more elaborately in terms of its implications for

achieving personal goals.

similarly

that chronic pain

Karoly and Jensen (1987) argued

patients can develop

"self-defealing

schemas" of pain in relation to the accomplishment of cherished

ment of social relationships (e.g., spending more time with one's children,

being

more

patient with

becoming a better person) and is more likely to influence psychological well-being than is carrying out Ihe more mundane lasks of everyday life (e.g., to get to the supermarket after work; Karoly, 1999). Health and social goals may also be construed

of Connecticut

Health

Center; Alex

Zautra

and

Paul

differently.

For example, individuals have been shown to attach greater value their

social

goals

than

themselves more often Glenn Affleck and Howard Tennen, Department of Community Medi-

movement

than is the fulfillment of more global and distant aspirations (e.g.,

to

cine, University

coworkers). The

toward and away from these goals is generally easier to monitor

to

their

health

goals,

but

to reward

when they make progress in Iheir health

goals than when they move toward their social goals (Karoly & Ruehlrnan, 1995).

Karoly, Department of Psychology, Arizona State University; Susan UrThe study participants were women with fibromyalgia. a synrows and Micha Abeles, Department of Medicine. University of Connectdrome

icut Health Center.

characterized

thresholds

to

was funded by a grant from the Fetzer Foundation. We are grateful for

nied in many by del

Pamela

and

Higgins's assistance

with

daia

collection

and

are indebted

to

widespread

soft-tissue

stimulation (Wolfe

et

pain al.,

and

low

1990). Far

than in men, fibromyalgia is accompa-

etal, and Skin Diseases Grant No. AR-20621, and preparation of this article

Jeffrey Siegel of National Technology Services for his help in program-

by

nociceptive

psychological

i

ng fatigue, as well as sleep disturbance is

{Boissevain

& McCain,

1991). Al-

though its pathogem

. controversial, fibromyalgia is believed

by many investigate

be a complex biopsychosocial disorder

ming the electronic diaries described in this article and to Saul Shiffman for his

generous assistance

in helping us design

our electronic

interview irising from diverse, per) laps multiple, underlying causes (Brad-

protocol. Correspondence concerning this article should be addressed to Glenn Affleck, Department of Community Medicine, MC-6235. University of Connecticut

Health Center, Farmmgton, Connecticut 06030. Electronic

ley,

1998; Chrousos, 1998; Wolfe, 1994).

Not only pain, but fatigue, can undermine the pursuit of personal goals

among

individuals with fibromyalgia. Henriksson, Gund-

mark, Bengtsson, and Ek (1992) found that most individuals with

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.

588

AFFLECK ET AL.

fibromyalgia experience pain and fatigue each day and claim that both states affect their ability to accomplish daily goals. A qualitative study (Henriksson, 1995a, 1995b) revealed several common stressors of living with fibromyalgia, including the burden it imposes on achieving personal goals and the hardship of abandoning their most cherished life plans. Efforts to cope with these burdens included changing daily routines, reorganizing daily priorities, and pursuing new goals and interests. Mounting levels of pain and fatigue associated with fibromyalgia may thus constitute a severe challenge to the fulfillment of health and social goals. In a preliminary test of this hypothesis, Affleck et al. (1998) analyzed daily reports of goal-directed efforts, barriers, and progress toward a health and a social goal in a sample of women with fibromyalgia. On days when their pain and fatigue increased across the day, study participants were more likely to report at day's end that their pain and fatigue hindered their progress toward both goals. On days when their pain increased across the day, they reported less progress in reaching their social goals. These initial findings, however, ignored sources of individual differences in goal effort, barriers, and progress and in the extent to which these goal processes track the vicissitudes of pain and fatigue. The present study looks to value-expectancy models of motivation for constructs that may explain how and when women with fibromyalgia remain engaged in the pursuit of their goals, particularly when they face elevated levels of daily pain and fatigue. Value-expectancy models of motivation share the assumptions that (a) behavior is organized around valued goals, and (b) the pursuit of such goals is fueled by the expectation that they can be accomplished (Carver & Scheier, 1998, 1999; Rotter, 1954). These models predict that when individuals perceive a goal to be both meaningful and attainable, they will persist in their efforts to reach the goal, even in the face of obstacles. Although there is little disagreement among goal theorists that the meaningfulness of a goal should predict the initiation and persistence of goal-directed effort (e.g., Carver & Scheier, 1998; Karoly, 1999), there has been debate over the relative importance of constructs related to goal outcome expectancies (cf. Carver et al., 2000; Tennen & Affleck, 2000; Snyder et al., 1991). In addition to evaluating the importance of goal valuation in the pursuit of daily goals in the face of pain and fatigue, we examined two constructs that have been theorized to influence goal outcome expectancies: self-efficacy appraisals and dispositional optimism. In Karoly's taxonomy of goal construal, self-efficacy is a critical appraisal directing the choice to actively pursue a goal (Karoly & Ruehlman, 1995). Self-efficacy theory (Bandura, 1977, 1997) belongs to a family of conceptual models that stress the explanatory power of personal control over desired outcomes. Self-efficacy is typically an appraisal that one has the capacity to do what it requires to accomplish a specific goal. Believing that one possesses the knowledge, skills, or abilities required to achieve that goal is thought to instill confidence in goal attainment and spur goal-directed efforts. Over the past two decades, Bandura and many others have documented the importance of self-efficacy judgments for accomplishing such diverse goals as overcoming fear (Bandura, Reese, & Adams, 1982), quitting smoking (Owen & Brown, 1991; Haaga & Stewart, 1992), and coping with stressful events and chronic illness (Bandura, 1997).

A sense of self-efficacy in overcoming the challenges of living with chronic pain has repeatedly been shown to predict psychological and physical well-being (e.g., Lorig, Chastain, Ung, Shoor, & Holman, 1989; Keefe et al., 1997; Smarr et al., 1997). A recent study by Levfebvre et al. (1999) examined the ability of selfefficacy appraisals to predict the daily experiences of individuals with rheumatoid arthritis. Self-efficacy appraisals regarding one's ability to manage pain, fatigue, and physical functioning were associated with less daily pain, more positive daily mood, and greater confidence in the effectiveness of one's daily coping strategies. Whereas individuals in a national study of chronic pain sufferers reported lower levels of self-efficacy for the accomplishment of vocational goals than did a nonpain comparison group, those in chronic pain sample who expressed higher goal selfefficacy also reported less anxiety than those with lower goal self-efficacy (Karoly & Ruehlman, 1996). Carver and Scheier (Carver et al., 2000; Carver & Scheier, 1998) contend that personal control appraisals such as self-efficacy expectancies are not the only perceptions that increase people's confidence that they can meet their goals. They draw attention to the expectation itself that the goal will be achieved, an appraisal that can depend on circumstances that have little to do with a sense of personal control or self-efficacy. The critical element in Carver and Scheier's model of goal attainment is thus "whether the desired outcome seems likely to occur, not how it is to occur" (Carver et al., 2000, p. 141). People's estimates of the probability that a goal will be achieved can turn on their personal history of having met similar goals. They can also turn on their personality: namely, their disposition to hold generalized positive outcome expectancies (optimism) or negative outcome expectancies (pessimism) about the future (Carver & Scheier, 1998). Dispositional optimists display superior psychosocial adaptation to a host of health-related stressors, including coronary artery bypass surgery (Fitzgerald, Tennen, Affleck, & Pransky, 1993; Scheier et al., 1989), failed in-vitro fertilization (Litt, Tennen, Affleck, & Klock, 1992), bone marrow transplantation (Curbow, Somerfield, Baker, Wingard, & Legro, 1993), HIV-positive status (Taylor et al., 1992), pregnancy termination (Cozzarelli, 1993) and cancer (Carver et al., 1994). A prospective study of individuals with rheumatoid arthritis or fibromyalgia has also documented some of the benefits of optimism for adapting to daily life with chronic pain (Affeck, Tennen, & Apter, 2000; Tennen, Affleck, Urrows, Higgins, & Mendola, 1992). Optimists did not report less daily pain, but they did cite significantly higher levels of positive daily mood, more frequent positive daily events, fewer painrelated activity limitation days, and greater confidence in the effectiveness of their daily pain-coping strategies. Scheier and Carver (1992) have summarized effects of optimism on physical well-being, effects that span the appearance of physical symptoms in healthy individuals to milestones of recovery from life-saving surgery. Most pertinent to the present investigation are studies that show that optimism may motivate the changes in behavior, which help people achieve their health-related goals. This includes evidence of optimism's ability to predict health-promoting behaviors (Robbins, Spence, & Clark, 1991), heart patients' adherence to cardiac rehabilitation regimens (Shepperd, Maroto, & Pbert, 1996), and the limitation of sexual partners by individuals at high risk for HIV (Taylor et al., 1992).

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.

PERSONAL GOALS WITH FIBROMYALGIA Evidence is also mounting in support of a two-factor model of generalized outcome expectancies. Optimism and pessimism appear to be better conceived as related but independent constructs rather than as opposite poles of an optimism-pessimism continuum (Chang, D'Zurilla, & Maydeu-Olivares, 1994; RobinsonWhelen, Kim, MacCallum, & Kiecolt-Glaser, 1997). This conceptualization is strengthened by findings that optimism and pessimism may not have equivalent relations with other personality and adaptational outcome measures (e.g., Marshall, Wortman, Kusulas, Hervig, & Vickers, 1992; Mroczek, Spiro, Aldwin, Ozer, & Bosse, 1993). In daily diary studies of individuals with rheumatoid arthritis or fibromyalgia, Affleck, Tennen, and Apter (2000) showed that whereas optimism was a unique predictor of positive daily experiences such as pleasurable mood and desirable events, pessimism was a unique predictor of negative daily experiences. This evidence of a two-factor model of generalized outcome expectancies led us to isolate these constructs as independent predictors in the present study. Our principal research questions concerned how the valueexpectancy constructs of goal valuation, goal self-efficacy, and dispositional optimism and pessimism predict the ongoing pursuit of a health and a social goal by women with fibromyalgia. In addition to monitoring on a daily basis their progress in meeting these goals, our study participants rated the effort they expended to reach each goal, the extent to which their pain and fatigue were impeding goal progress, and how much pain and fatigue they were experiencing. Our mixed nomothetic-idiographic design (Epstein, 1983; Tennen & Affleck, 1996), coupled with multilevel data analytic procedures (Affleck, Zautra, Tennen, & Armeli, 1999; Schwartz & Stone, 1998), offered us a novel opportunity to uncover both commonalities and differences in goal-related activity with the ebb and flow of pain and fatigue. This methodology allowed us to examine the temporal dynamics of goal activity within a single individual across time, to determine the generalizability of these individual dynamics across persons, and, finally, to locate the sources of individual differences in these dynamics. First, we identified the within-person relations over time between changes in daily goal processes and rising and falling levels of daily pain and fatigue. Second, we examined the extent to which value-expectancy constructs predict individual differences in the ongoing chronicles of goal effort, goal progress, and pain- and fatigue-related barriers to goal attainment. Third, we determined the ability of these constructs to predict daily goal processes on those days when the goal-seeking barriers of pain and fatigue are more or less severe than usual.

Method Sampling Procedures Study participants were 89 women who met American College of Rheumatology criteria for primary fibromyalgia syndrome (PFS: Wolfe et al., 1990): pain in all body quadrants of at least 3 months duration, pain in 11 of 18 tender point sites on moderate digital pressure exerted by a clinical examiner, and the absence of other musculoskeletal pain disorders that could cause secondary fibromyalgia (e.g., rheumatoid arthritis). These women were recruited from a rheumatology faculty practice (patient group) and a pool of community volunteers with widespread pain (nonpatient group). Health-care seekers with PFS differ psychologically from those who do not seek care for their pain (Aaron et al., 1996); accordingly,

589

nonpatients were included in the study to extend the range of study variables and make the findings more generalizable to individuals with PFS. The 60 patients in the sample represented 70% of the faculty practice's female patients with a history of PFS who were invited to participate and for whom evidence of active PFS was confirmed by a rheumatology nurse practitioner. The 29 nonpatients in the study were recruited from an initial pool of 117 individuals who responded to newspaper advertisements soliciting calls from "women who have frequent muscle aches and pains and are not receiving medical care for their pain." A rheumatology nurse practitioner screened these individuals by phone and determined that 85 met criteria for widespread pain and absence of physician visits for pain treatment within the last 2 years. Medical histories and tender point examinations were done in person and identified 46 individuals meeting inclusion criteria. All 29 nonpatients randomly selected from this group accepted the invitation to join the study cohort.

Daily Variables and Measures Data from both end-of-day summaries of daily goal processes (using paper-and-pencil diaries) and within-day "momentary" interviews of pain and fatigue (using hand-held computers) were supplied by each participant for 30 consecutive days. As a modest incentive to increase adherence to the data collection protocol, participants were paid $.50 for each completed electronic interview and $2.50 for each completed diary. Daily goals record (DGR). Before beginning the self-monitoring phase of the study, participants were interviewed in person about their current midrange personal goals. Specifically, they were asked to identify two of the personal goals you're currently trying to accomplish .. . one in the area of your health and fitness, and a second in the area of your relationships with others.... We are interested in learning about your medium-range goals, ones that you conceivably could achieve during a period of between several weeks and a year. These goals may be extremely important to you, but they do not have to be. We will soon be asking you questions about these goals and to evaluate your progress in reaching these goals at the end of each day you will be participating in this study. Thus, these goals should be ones that you expect will be on your mind during the one-month course of this study. All participants were able to identify at least one of these goals; 86 of the 89 participants identified a health goal and 79 a social goal. Examples of chosen health/fitness goals were losing weight, staying on a low-fat diet, and maintaining an exercise routine. Social goals included such aims as being more patient with coworkers, spending more time with my family, and being less critical of my husband. Each night before bedtime, participants answered three questions about each of the personal goals they had identified in the initial interview. Specifically, participants used 0 to 10 scales (where 0 = no and 10 = extreme) to rate (a) how much effort they made to try to make progress toward this goal today (i.e., goal effort); (b) how much their fibromyalgia pain and/or fatigue interfered with their progress toward this goal today, that is, pain- and fatigue-related goal barriers; and (c) how much progress they made toward achieving this goal today, that is, goal progress. These ratings along with other diary questionnaires were mailed to the investigators the following morning. Of a total number of 2,670 DGR's requested, only 5 (0.3%) were recorded as missing data. Electronic interviews. Participants carried palm-top computers programmed as an electronic interviewer (ELI), which requested information about their pain and fatigue three times a day. This device was a programmable battery-powered Psion Organizer II (Psion, Concord, MA) weighing 8.8 oz (249 g). The Psion Organizer has amply demonstrated its feasibility and reliability as a data collection instrument in several other daily self-monitoring studies (e.g., Affleck, Apter, et al., 2000; Carney,

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.

590

AFFLECK ET AL.

Tennen, Affleck, DelBoca, & Kranzler, 1998; Shiffman et al., 1994; Stone et al., 1994). Some procedures for the ELI protocol parallel those designed by Shiffman and colleagues (e.g., Shiffman et al., 1994) for their electronic diary studies of cigarette smokers. The data entry procedure for each ELI request proceeded from the termination with a keystroke of an audible beep to the choice to answer the interview either then, 5 min later, or 15 min later. The auditory signal lasted 60 s; if not answered within this time, it was repeated 5 min later, and if not answered again, another 5 min later. Failure to answer this sequence of three requests for data produced a missing entry for that time period. Interview questions were presented one at a time in a fixed order on a liquid crystal display. Participants replied to each question by scrolling across fixed response options with backward and forward arrows and then pressing an "enter" button to save the response and its time stamp on a data storage device, which could not be erased without exposing it deliberately to 30 min of ultraviolet light. The response option appearing first on the screen with each new question was randomized to minimize response set. Halfway through the study and at its completion, a research associate visited participants in their homes and uploaded their data to a laptop computer, where it was entered automatically in a spreadsheet file for data analysis. Pain and fatigue assessments were scheduled three times each day: at times randomly selected by ELI within response windows opened between 9:45 am and 11:15 am (morning interview), between 2:45 pm and 4:15 pm (afternoon interview), and between 6:45 pm and 9:15 pm (evening interview). Participants rated their current pain intensity on 0-6 scales (anchored verbally at 0 = none, 2 = mild, 4 = moderate, 6 = severe) in each of 14 areas of the body: neck, shoulders, chest, buttocks, upper (lower) back, left (right) upper leg, left (right) lower leg, left (right) lower arm, and left (right) upper arm, reflecting the widespread pain criteria required for PFS diagnosis. Current pain for that interview was scored as the sum across body regions. Fatigue was measured by the two adjectives tired and drowsy; each was rated on a scale of 0-6 (anchored verbally at 0 = none, 2 = slightly, 4 = moderately: 6 = extremely) and then summed for a measure of current fatigue. Of a total of 8,010 pain/fatigue interviews requested of the 89 participants, only 135 (1.7%) were missing.

Other Variables and Measures One week before beginning the self-monitoring phase of the study, participants completed several questionnaires. These assessed their goal valuation, goal-specific self-efficacy appraisals, and dispositional optimism/pessimism. Goal valuation and self-efficacy. Participants completed the directive functions questionnaire of the Goal Systems Assessment Battery (Karoly & Ruehlman, 1995), once for their health goal and again for their social goal. Four items measure goal valuation (e.g., "This goal is worthwhile") and four items measure goal self-efficacy (e.g., "I have what it takes to reach this goal"). The response format for each item is a scale on which 0 = a not at all accurate, 1 = a slightly accurate, 2 = a quite accurate, and 3 = an extremely accurate description of the way respondents think about this goal. The reliability and validity of this questionnaire have been established for ratings of both health and social goals. This includes evidence of test-retest reliability, internal consistency, and associations with reports of depressive symptoms. Cronbach alphas in the present sample ranged from .82 of health goal self-efficacy to .94 for social goal valuation. Dispositional optimism and pessimism. The Life Orientation Test (LOT; Scheier & Carver, 1985) is a 12-item questionnaire (eight items and four distractors) assessing the general tendency to be optimistic or pessimistic with respect to outcome expectancies. Substantial evidence exists for this scale's concurrent, construct, and discriminant validity, internal consistency, and test-retest reliability.1 The LOT appears to measure two related but distinct constructs: optimism and pessimism (Chang et al.,

1994; Marshall et al., 1992; Robinson-Whelen et al., 1997). Hence, we retained separate optimism and pessimism scores for analysis. The four optimism items had an internal consistency of .82 in the current sample, and the four items tapping pessimism had an internal consistency of .84.

Results Demographic Correlates of Study Variables Participants' age, education, time from symptom onset, and healthcare-seeking status (patient vs. nonpatient) were examined as correlates of optimism, pessimism, health and social goal valuation and self-efficacy scores, as well as of average daily levels of pain, fatigue, and each of the three DGR scores. Seven of those 44 correlations were statistically significant. Specifically, patients, compared with nonpatients, reported greater daily pain, r(87) = .46, p < .001; daily fatigue, r(87) = .32, p < .01; and greater pain- and fatigue-related barriers to achieving their health, r(84) = .39,p < .001; and social, rill) = .26,p < .05, goals. More educated individuals also reported less daily pain, r(87) = -.26, p < .05, and younger individuals, r(87) = -.21, p < .05, and those with more recent symptom onset, r(87) = —.26, p < .05, were more pessimistic. This pattern of correlations nominated only one potential demographic confound for the planned analyses—the possibility that healthcare-seeking status might confound the relations of pain and fatigue with perceived pain- and fatigue-related barriers to goal attainment.

Descriptive Findings Table 1 presents descriptive statistics for the health and social goal valuation and self-efficacy scores and each aggregate (mean) DGR score. The average participant stated that it was between a "quite accurate" and an "extremely accurate" statement that her chosen goals were very important to her and a "quite accurate" statement that she was able to do what was required to attain each of these goals. There were modest-to-high correlations in scores across the two goal domains, indicating that participants' appraisals of their health goals and their aggregate efforts, perceived painand fatigue-related barriers, and progress for health goals paralleled those of their social goals. At the outset of the study, participants valued their health and social goals similarly and felt equally confident about their abilities to accomplish both goals. Yet, health goals appeared to be more difficult to accomplish than social goals: Participants reported less progress in their health goals and greater pain- and fatigue-related barriers to health goal attainment, which perhaps was reflected in their comparatively diminished effort to accomplish them.

Approach to Multilevel Data Analyses The principal analyses for this study derive from a class of statistical procedures called multilevel modeling (Kreft & de Leeuw, 1998). These procedures partition the two sources of variance in our person-day data set—differences between persons in their average levels of the variables and differences within

1 The revised version of the LOT (Scheier, Carver, & Bridges, 1994) was unavailable at the time our study was initiated.

PERSONAL GOALS WITH FIBROMYALGIA

591

Table 1 Descriptive Statistics for Goal Appraisals and Mean Daily Goals Record Items Health goal (n = 86)

Social goal (n = 79)

Variable

M

SD

M

SD

r(75)

t(16)

Goal valuation Goal self-efficacy Mean daily goal effort" Mean daily barrier to goal progress" Mean daily goal progress"

10.27 8.27 4.06 3.51 3.36

2.14 2.35 1.78 2.32

10.16 8.17 4.97 2.86 4.15

2.05 2.54 2.25 2.24 2.14

52*** 37*** .56*** 70*** .57***

0.50 0.33

1.80

-3.78*** 2.55** -3.35***

a

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.

On scales of 0-10. **/?< -001.

persons in their daily reports over time. In the vernacular of multilevel modeling, Level I units refer to the discrete daily reports of goal processes, pain, and fatigue; the Level 2 units are persons, who are described in this study by their optimism, pessimism, goal valuation, goal self-efficacy, and average levels of pain and fatigue across 30 days. Our multilevel analyses addressed (a) the withinperson relations between Level 1 variables: for example, the relation between daily fatigue and that day's perceived pain- and fatigue-related goal barriers; (b) the effects of Level 2 variables on Level 1 intercepts: for example, the relation between optimism and a person's average daily goal effort; and (c) the cross-level interactions between persons and observations: for example, the relation between self-efficacy and daily goal progress on more or less painful days. The SAS Proc Mixed procedure for multilevel data analysis furnished parameters in the form of unstandardized maximum likelihood estimates (SAS Institute, 1996). For all analyses, we allowed intercepts and slopes to vary randomly; this enabled us to generalize the findings to the population of persons from which the sample was drawn, to the population of observations from which their daily reports were sampled, and to the population of within-person relations that these samples are intended to represent (Affleck et al., 1999). Because autocorrelated residuals are a common consequence of equally spaced (daily) observations, and can bias standard errors and significance levels, these analyses also fit the error terms to an AR(1) model.

Level 1 Analyses of DRG Relations with Pain and Fatigue

(1)

Table 2 presents the results. On days with greater pain or fatigue, participants reported less progress toward both of their goals and reported greater pain- and fatigue-related barriers to attaining these goals. In addition, days with greater fatigue were characterized by diminished efforts to accomplish health goals.2 Level 2 Effects

DGR,, = B 0/ + e,,.

(2)

The intercept (B0l) was predicted by the average intercept (mean DGR score across persons: B10) and by person fs optimism (OPT,), pessimism (PESS,), goal valuation (GVALU,), goal selfefficacy (GSEFF,), average pain (AVGPAIN,) and average fatigue (AVGFAT,), the respective regression coefficients for these effects (B12 to B 16 ), and a random component (u,): B oi = B 10 + B,,(OPT,.) + B I2 (PESS,) + B 13 (GVALU,)

Each DGR score on each day was regressed on that day's level of pain and fatigue (each averaged across the three within-day observations). Specifically, the following model was tested, predicting each DGR score for person i on day ;. DGR,, = BO, + B,,(Pain) + B 2 ,(Fatigue) + eit.

levels of pain and fatigue. Their intercorrelations are presented in Table 3. Participants experiencing more pain and fatigue did not differ in their levels of optimism or pessimism, nor did they appraise their goals any differently. Those higher in optimism were more self-efficacious concerning their social goal, whereas those lower in pessimism were more self-efficacious concerning their health goal. Within each goal domain, those who attached greater value to their goal also felt more self-efficacious in attaining it. The low-to-moderate correlations appearing in Table 3 mitigate concern about multicollinearity in the multivariate analyses of Level 2 predictor sets for each goal domain. To assess the relations of Level 2 variables with individual differences in each DGR score intercept (mean), the following general model was tested. Each DGR score on day / for person i (DGR/r) can be partitioned into that person's average goal progress (B0[) across all diary days and that day's deviation from the average (e,,):

on Level 1 Intercepts

The Level 2 predictors of daily goal processes include optimism, pessimism, goal valuation, goal self-efficacy, and mean daily

+ B 14 (GSEFF,.) + B 15 ( AVGPAIN,.) + B 16 ( AVGFAT,) + «,,

(3)

Parameter estimates from Equation 3 indicate that individuals who experienced more severe pain reported greater pain- and fatigue-related barriers to their health goal (B15 = .066), F(l, 79) = 23.33, p < .001, and their social goal (B15 = .043), F(l, 2 Another family of Level 1 analyses examined the ability of today's DGR scores to predict tomorrow's pain and fatigue (controlling for today's pain and fatigue) and the ability of today's pain or fatigue to predict tomorrow's DGR scores (controlling for today's DGR scores). None of these lagged relations was significant; hence, there were no cross-day carryover effects characterizing the relations of pain and fatigue with daily goal processes.

592

AFFLECK ET AL.

Table 2 Multivariate Level I (Within-Person) Relations of Pain and Fatigue With Daily Goals Record Scores Pain Variable

B

F(l, 2287)

B

F(l, 2490)

Social goal effort Social goal barrier Social goal progress Health goal effort Health goal barrier Health goal progress

-.009

1.63 47.43*** 7.23** 0.83 108.79*** 4.16*

-.071 .243 -.104 -.078

3.56 42.16*** 6.00* 3.95* 97.80*** 8.37**

*p