Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

The Daily Stress and Coping Process and Alcohol Use

among College Students*


CRYSTAL L. PARK. rii.D.,1 STEPHEN ARMELI, l'll.o.,t AND HOWARD TENNEN, PII.D. 1
Department of Psychology. University of Connecticlll. 406 Babbidge Road. Box U I 020. Storrs. Connecticut 06269-1020

ABSTRACT. Objective: To examine whether students drink more al- had events perceived as relatively more stressful. Students also drank
cohol on higher-stress days than on lower-stress days: whether daily cop- more on days characterized by relatively lower problem-focused cop-
ing Slr'~tegics and affective states mediate the within-person relationship ing and relatively higher positive and negative aiTect. Contrary to pre-
between stress and drinking; and whether the daily stress-drin king pro- dictions. the effect of stress on drinking was not fully mediated by coping
cess varies as a function of previously idcmi fied between-person risk and aflcct. Findings also revea led individual differences in several links
factors (sex. family history, coping motives, sensation seeking. neuroti- in the daily stress coping-drink ing process. Conclusions: The daily stress
cism). Me1hod: College students (N = 137) completed daily surveys for coping-drinking process is complex, and it appears that there arc both
28 consecutive days. Resul!s: With daily coping, afl'ect and weekly trends positive and negative affective pathways underlying daily alcohol con-
in drinking controlled for. students consumed more alcohol on days that sumption. (J. Stud. Alcoho/ 65: 126-135. 2004)

0 F THE DIFFERENT MOTIV ES underlying college


students' alcohol use, coping with stress is believed
to be most closely linked with the development of problem
to the theoretically posited mechanisms underlying the
stress-alcohol link.
According to Conger ( 1956), alcohol reduces negative
drinking and the fai lure to "mature oul" of heavy drinking affective states associated with stress, thus reinforcing con-
following graduation (Baer, 2002). However, much of what sumption and, in turn, increasing the probability of alcohol
is known about stress and alcohol use among college sn•- use when experiencing stress. More recent accounts of the
dents comes from studies that reveal little about the fast- stress-drinking link (e.g., Cooper et al., 1988) have inte-
moving, intraindividual processes outlined in theoretical grated key process variables taken from Lazarus and
models. Folkman's (1984) transactional model of stress, such as the
The present study addresses two key limitations of pre- concept of coping. In this model, people's attempts to cope
vious research: (I) the use of between-person or long-tem1 with stressful encounters directly influence their affective
longitudinal designs that are not sensitive to the ebb and states. Accordingly. coping can be conceptualized as a proxi-
flow of much of the variation in the stress process and mal mediator of the influence of negative experiences on
alcohol use (Tenncn ct al., 2000) and (2) having partici- negative affect and. subsequently, on alcohol use.
pants recall such phenomena weeks or even months after In general, research examining college students and ado-
their occurrence, the validity of which has been questioned lescents supports the notion that Iife stress is an important
in terms of memory decay and retrieval biases (Shiffman, risk for alcohol use in general (e.g., Hutchinson et al., 1998).
2000). We used an Internet-based daily diary design that However. stt•dies explicating the processes linking stressful
reduced recall distortion and allowed us to examine these encounters with alcohol use. namely coping. are limited
relationships at time intervals that might better correspond and conflicting (e.g., Cooper ct al., 1992; Simpson and Ar-
royo, 1998; Windle and Windle, 1996). None of these stud-
ies actually addressed the process question of interest: What
Received: March 14. 2003. Revision: July 28. 2003.
types of coping strategies are proximally related to alcohol
•This study was supported by 1 ationallnstitute on Alcohol Abuse and
Alcoholism grants 5R03AA 12434-02 and P50-AA03510. Crysta l L. Park use?
was supported, in pan. by Na tional Insti tutes of Health training grant A more accurate picture of the role of coping in the
5K1 2 HD01409 to the Center for Interdisciplinary Research in Women's stress-drinking process would require monitoring stress, cop-
Health. University of Connecticut Health Center. ing, affect and alcohol use repeatedly within the person
'Correspondence may be sent to Crystal L. Park at the above address, or
and closer to the time they occur. For example, a series of
via email at: crysdara@aol.com. Stephen Armeli is with the Department of
Psychology. Pace University. New York . lloward Tennen is with the Depart- community-based daily diary studies of adult regular and
ment of Community Medicine and Healthcarc. University of Connecticut heavy drinkers has shown that individuals reponed greater
Health Center, Fam1ington. CT. alcohol consumption on days during which they cncoun-

126
PARK, ARMEL! AND TENNEN 127

tered certain types of negative events (Carney et al., 2000; achieve this goal. and therefore use of avoidance coping
Mohr et al., 2001). Hussong et al. (2001) examined, among might be associated with greater alcohol use. A voidance
college students, the association between affect and alcohol coping also may be linked with drinking on a particular
use across weekday and weekend time intervals. Some sup- day because, as an ineffective strategy, it does not reduce
port was found for the notion that elevated negative affect (and may actually generate) negative affect, which in tum
during one time period anticipated relative increases in might lead to alcohol use.
drinking in subsequent time intervals and that these asso- A secondary goal of this study was to examine how the
ciations varied as a function of social factors. However, daily stress-drinking process varied as a function of previ-
Hussong and colleagues (2001) did not examine the occur- ous! y identi tied between-person risk factors. Specifically.
rence of stressors during these time periods, and none of we examined individual differences in drinking-to-cope mo-
the aforementioned studies measured coping repeatedly tives, neuroticism, sex and family history of alcohol-related
within the person. Frorru11e and Rivet ( 1994) did have col- problems. Based on theory and prior findings. more posi-
lege students record their stress, coping and drinking in tive within-person stress-drinking associations should be
daily diaries over I0 days. However, only between-person found among those high in drinking-to-cope motives (Coo-
associations were examined using the aggregated daily data. per et al.. 1995) and neuroticism (e.g., Carney et al., 2000)
Thus, the primary goal of the present study was to ex- and among individuals with a family history of alcohol abuse
amine the proximal within-person associations among event (Simpson and Arroyo, 1998). Prior research also has found
stress, coping, affect and alcohol use in a college student that gender moderates the stress-drinking association, al-
sample. We chose to examine the daily associations, as though the specific effects have differed across studies
opposed to other time intervals (e.g., Hussong et al., 2001), (Greeley and Oei , 1999).
for several reasons. First, as Hussong and associates (200 I) Finally, we also examined factors relevant to the social-
stated, the time interval chosen to examine the links among enhancement drinking pathway (e.g.. Cooper et al.. 1995:
stress, coping. affect and alcohol should be driven by mecha- Windle and Windle. 1996). We thought it important to con-
nisms under consideration. Prior studies of the stress and trol the effect of positive affective states on daily drinking
coping process in general and studies of stress and alcohol as well as individual differences- such as sensation seek-
use specifically support the validity of examining same- ing and social-enhancement drinking motives -that might
day associations (for a review, see Tennen et al., 2000). be associated with drinking in response to positive affec-
Second, although most drinking occurs on weekends, col- tive states. A lso, in an exploratory fashion, we examined
lege students do report drinking on al l days of the week the role of these variables in the stress coping-drinking
(e.g., Hussong et al., 200 I). Thus, aggregating to larger process.
time intervals (e.g., weekday and/or weekend) might ob-
scure specific instances of stress-induced drinking. Finally, Method
examination of daily drinking allowed us to use all of our
within-person observations, thus maximizing power. Sample and procedure
We examined three coping strategies: avoidance (efforts
to escape focusing on a situation); approach emotion-fo- Participants were recruited from the participant pool at
cused (actively processing and expressing emotions); and the University of Connecticut during the fall semester; stu-
problem-focused (efforts to solve a situation). Recent dents received fu ll research credit for their participation in
conceptualizations of emotion-focused coping distinguish addition to financial compensation. Participants initially ac-
between avoidant and approach behaviors (Teny and Hynes, cessed a secure Internet Web site to complete the baseline
1998). Lazarus ( 1993) and others (e.g., Tennen et al., 2000) measures. Then about a week later, the daily dia1y portion
have argued that any coping strategy may be adaptive or of the study commenced, wherein participants received an
maladaptive in particular circumstances. Yet, there is a great email message each day reminding them to access the di-
deal of support for the notion that problem-focused coping ary pages. To minimize reporting time variation, partici-
is related to less negative affect (Terry and Hynes, 1998), pants could log on to the Web site only between the hours
and avoidance coping is associated with negative emotional of 3:30 PM and 7:00 PM. Each day, they completed mea-
outcomes (Stanton et al., 2000). Approach emotion-focused sures assessing the previous evening's activities (i.e., those
coping appears to operate differently and may be related to occurring after the previous day's daily report), including
better adjustment (e.g., Berghuis and Stanton, 2002). Thus, alcohol use, and the current day's stressor, stressor apprais-
we expected that approach emotion-focused coping efforts als, coping efforts and affect (up to the reporting time).
and problem-focused coping effor1s would lead to less al- Of the 246 participants (73.8% women) who initially
cohol consumption because they help reduce distress. ln enrolled in the study, 13 did not complete the initial assess-
contrast, if the aim of avoidance coping on a particular day ment, 5 did not participate in the daily portion of the study
is to evade dealing with a situation, drinking can help and 38 were dropped from the final analyses because of
128 JOUR AL OF STUDIES 0 ALCOHOL I JANUARY 2004

low compliance (i.e., less than 50% of the study days com- Sensation seeking. At baseline. panicipants completed
pleted) in the daily portion of the study. Of the 190 re- the 8-itcm excitement seeking subscale from the Extrover-
maining participants, 53 did not report alcohol use during sion scale of the revised EO-Persona lity Inventory (NEO-
the daily portion of the study and thus could not be in- PI -R: Costa and McCrae. 1992). The items were rated on a
cluded in the fina l analyses examining drinking. This re- 7-po int scale rangi ng from I (strongly disagree) to 7
sulted in a usable sample size of 137. Of the 3,836 possible (strongly agree). Cronbach 's alpha was .74.
reporting days, students logged in on 3, 132 days (8 1.6% of Daily stress and coping. Each day participants were asked
the total days). to provide a brief description of "today's most negative
The final sample of 137 comprised 101 women (73.7%) experience (no matter how small)" and to rate its degree of
and had a mean (SO) age of 18.72 ( 1.02). Most partici- stressfulness on a 7-point scale from "not at all" to "ex-
pants were white (90.5%): the remainder were Asian (4.4%), tremely." Coping was assessed with 12 items used in Park
black (0. 7%), Latino (0. 7%) and of other racial groups et al. (in press) that were adapted from the Brief COPE
(3.7%). About half (52.6%) of the participants were fresh- (Carver, 1997) and Stanton et al. (2000). Park ct al. (in
men, 37.2% were sophomores, 7.3% were juniors and 2.9% press) found three factors underlying these items: emotion
were seniors or beyond. Most students lived on campus approach coping (e.g., "I've been al lowing myself to ex-
(96.4%), were full-time students (98.5%) and single ( I00%). press my emotions," "I' ve been getting emotional support
Only seven pa11icipants (5. 1%) were members of a frater- from others"); avoidance coping (e.g., " I' ve been giving up
nity or sorority. A comparison of the final sample with trying to deal with it," "I've been saying to myself ' this
individuals not included in the analysis showed no differ- isn't real'"); and problem-focused coping (e.g., "I've been
ences in sex makeup (X2 = .00, I df, p = .99), class year taking action to try to make the situation better," "I've been
(X2 = 4.44. 4 df, p = .35), race (white vs other: x2 = 0.91, I trying to come up with a strategy about what to do"). Par-
df. p = .34) or age (t = 0.91. 242 elf, p = .93). ticipants reported the extent to which they engaged in each
of these strategies in dealing with today's most undesirable
Measures event. Items were rated on a 4-point scale from I ("I haven't
been doing this at all") to 4 (" I've been doing this a lot'').
Drinking motives. At baseline, partiCipants completed We created scale scores by averaging together the appro-
Cooper's ( 1994) adolescent coping, social and enhancement priate items. Alphas on days 2, 14 and 27 ranged from .84 to
reasons for drinking sca les. For each 5-item subscale, re- .94 for emotion approach coping, from .68 to .8 1 for avoid-
spondents reported the relative frequency of drinking for ance coping and from .83 to .87 for problem-focused coping.
such reasons. Items were answered on a 4-point scale rang- Daily affect. Individuals responded to the question "I low
ing from I (almost never/ never) to 4 (almost always/al- much do each of the following words describe your mood
ways). Both in the source article and in the present sample TODAY? Please click on the appropriate response." Draw-
(r = .76, p < .001), a strong positive association was found ing on existing models of afTect and mood (e.g., Larsen
between the social and enhancement subscales; thus they and Diener, 1992; Watson ct al.. 1988), positive affect was
were combined to form an overall social-enhancement mo- assessed with the items happy, cheerful, enthusiastic, ex-
ti ve. Cronbach 's alpha was .89 for the coping motives scale cited, relaxed and content, and negative affect was assessed
and .94 for the social-enhancement motives scale. with the items nervous, jittery, hostile, angry, sad and de-
Family hist01y of alcohol abuse. At baseline, partici- jected. Participants responded using a 5-point scale ranging
pants completed a modified version of the family history from 0 (not at all) to 4 (extremely). We created general
assessment module from Project MATCH (Project MATCH positive and negative affect scores. Cronbach's alphas for
Research Group, 1998). Here, participants reported whether the composite scales across days 2. 14 and 27 ranged from
any immediate fami ly member (mother. father, brother, sis- .75 to .86 for negative afTect and from .85 to .93 for posi-
ter) had problems with alcohol usc. Participants were coded tive afTect.
as 0 if no family member had problems with alcohol use, Dt1ily a/coho/use. Participants reponed their drinking at
or I if at least one family member had problems with alco- home (e.g., "in your apartment, donn room, etc.") and away
hol use. In the present sample, 36 panicipants (26.3%) were from home (e.g., "at a bar, friend's apartment or dorm room,
classified as having a family history of alcohol problems. etc.") for the previous evening. One drink was defined as
Evidence of reliability and validity are presented in the one 12-oz can or bottle of beer, one 4-oz glass of wine,
source article. one 12-oz bottle of wine cooler or I oz of liquor straight or
Neuroticism. At baseline, panicipants completed the 12- in a mixed drink. Response options ranged from 0 to > 12;
item neuroticism (N) subscale from the EO Five-Factor responses of> 12 were recoded to a value of 13 {this value
Inventory (NEO-FFI; Costa and McCrae, 1992). The items represented only 0.5% of the total responses for each cat-
were rated on a 7-point scale ranging from I (strongly dis- egory). In the current analyses, we summed drinking at
agree) to 7 (strongly agree). Cronbach's alpha was .87. home and away to form a total drinking variable.
129

Results motivation (r = .50, p < .0 I) and drinking-to-cope motiva-


tion (r = .36, p < .0 I); no other differences were found.
Descriptive statistics and correlations
Multilevel regression a11a~vses
Because drinking for day 1 was recorded on day 1 + I,
we had daily data for a maximum of 27 days. In addition, The stress-dri11king associatio11. We conducted multilevel
incomplete data for any I day led to two missing data points regression analyses (Raudenbush and Bryk. 2002: Schwartz
for the drinking analyses. This resulted in a Level I sample and Stone, 1998) using Hierarchical Linear Modeling soft-
size of 2.420. or a mean (SO) of 17.66 (5.41) days per ware (vol. 5.04: Raudenbush et al.. 200 I). Because daily
person. Each day of the week was well represented in the drinking is a count variable. we used the Hierarchical Gen-
final sample, ranging from Fridays (290 person days; 12.0% eralized Linear Modeling option, specifying a Poisson sam-
of total days) to Mondays (415 person days; 17 .I% of total pling model. log-link function and overdispersion
days). (Raudenbush and Bryk, 2002). We report population aver-
Panicipants reponed having at least one drink on 21.9% age estimates and robust standard errors for fixed effects.
of the reponing days (range: I to 13 days). The mean (SO)
Tj ,; = bo; + b 1, (Event Stress,;) + e, (I)
number of drinks per study day (based on disaggregated
data) was 1.32 (3.11) drinks. with the most drinking occur- bo = 'Yoo + y01 (Aggregate Event Stress;)+ u0, (2)
ring on Fridays (mean = 3.83 [4.27] drinks) and the least
(3)
on Mondays (mean = 0.20 [ 1.15] drinks). egative daily
experiences fell into the following categories: academic We first evaluated a model predicting the number of
(31.5%), health/fatigue (20.9%), interpersonal (20.6%) and drinks from event stress ratings. Equation I shows the
other (e.g., work-related, financial, traffic/parking, weather) within-person (or Level I) prediction model wherein the
(27.0%). The average event stress level was highest on Mon- log-predicted value (11) for day r and person i is a function
day (mean = 4. 70 [ 1.86]) and lowest on Saturday (mean = of each person's own intercept, b0, and event stress rating
3.89 [ 1.92]). slope, b 1;, and within-person residual, e,;. To obtain unbi-
Correlations are displayed in Table I. Consistent with ased estimates of the pooled within-person associations, all
prior research, men, those high in sensation seeking, drink- predictors were person-mean centered (see Bryk and
ing-to-cope and social-enhancement motives, as well as Raudenbush, 1992, pp. 117- 122). Thus, b01 in Equation I
those with higher average daily negative mood (marginal can be interpreted as individuals' average daily drinking
significance) drank more on average. Consistent with Armeli levels. To reintroduce the variability lost by person center-
ct al. 's (2000) findings. those with higher average daily ing, we included person means (sec Schwartz and Stone,
event stress ratings drank less on average. Although the 1998) in the Level 2 intercept model (Equation 2). Thus,
three aggregate coping strategies were moderately to highly here, the Level I intercepts (or average outcome levels) arc
correlated, Park et al. (in press) found that at the within- a function of the grand mean (y00 ), each person ·s aggregate
person level they were only moderately associated. Finally, event stress and a between-person residual component, u0;.
we examined correlations between drinking status (drink- The Level 2 model (Equation 3) shows the stress-drinking
ers, 11 = 137, coded as I, and nondrinkers, n = 53, coded as slopes as a function of the mean slope (y10) and a between-
0) and a II of the variables. Ori nkers reponed greater aggre- person residual component, u1,. All Level 2 variables were
gate avoidance coping (r = .16, p = .03). social-drinking grand-mean centered.

TAillf I. Correlations and dcscnpuve statistics


Mean (SO) I 2 3 4 5 6 7 8 9 10 II 12
I. Sex
2. Family history -.02
3. euroticism 3.61 ( 1.0 I) . !9• .2 !•
4. Sensation seeking 5.34 (0.96) ·.21° -.16§ -.221
5. Drinking-to-cope motive 2.2 1 (0.95) -.06 .I 0 .46' . 13
6. Social-enhancement motive 3.44 (0.98) -.12 -.03 .w .34 1 .59'
7. Aggregate event stress 4.38 ( 1.05) .41 t .03 .5 1t -.05 .241 . I0
8. Aggregate emotion approach 1.69 (0.49) .09 ·.04 .20° .09 .29' .II .301
9. Aggregate avoidance 1.36 (0.42) ·.02 .0 I .351 -.04 .4 1t .17• .27' .661
I 0. Aggregate problem-focused 2.08 (0.53) .07 .05 .20• .04 .27' . !9• .421 .701 .46'
II . Aggregate negative affect 0.55 (0.48) -.Ot .02 .47' .0 I .4 )-· .25' .291 .551 . 71 I .41 t
12. Aggregate positive affect 1.95 (0.56) .06 -.10 -.24 1 . t8• -. 14t · .08 -.0 I . 14§ .04 .21 ° .01
13. Aggregate drinking 1.43 ( 1.61) -3·
-.) .06 -.03 .27 ' .!9• .35' -.23 1 .02 .I I .01 . 15§ -. I I

Note: N - 132. §p < . 10: •p < .05: 1p < .01.


130 JOUR AL OF STUDIES 0 ALCOHOL I JANUARY 2004

Results from this model revealed a negative within-per- tered, into the Level I model (Equation I) containing the
son association between the daily event stress and the num- event stress variable and the day-of-the-week contrasts. The
ber of drinks (b = -.057, 1 = -2.251. 136 df, p = .024) and coping slopes were modeled as random effects. In addition,
a negative association between aggregate daily event stress the person-level means for the coping variables were en-
and average number of drinks (y - -.2 15, 1 = -2.750, 135 df, tered into the Level 2 model (Equation 2). Results from
p .006). In other words, indi viduals drank less on days thi s model can be seen in Table 2 (Model I). Aggregate
with more stressful events, and individuals who had higher avoidance coping and event stress were unique predictors
average daily stress levels drank less. A significant vari- of aggregate drinking, with those who used avoidance cop-
ance component also was found for the withi11-person event ing more often and those who rated events as less stressful
stress-drinking association (u 1 - .032, x2 = 179.67, 135 df, on average drinking more on average.
p .007), indicating significant individual differences for Of greater interest are the within-person associations (sec
the within-person slopes. The Level I intercepts for all mod- Pooled within-person coefficients). As shown in Table 2.
els demonstrated significant variance components, but for Lhe the unique within-person association between event stress
sake of brevity they arc not reported here. and drinking was now signi flcant, indicating that, on aver-
To control for weekly trends in drinking, we included age, individuals drank more on days characterized by events
six day-of-the-week dummy codes (with Sunday coded 0) perceived as relatively more stressful. Similarly, the within-
in the Level I model as fixed effects (see Anneli et al., person unique association between problem-focused cop-
2000). Results from this model revealed a nonsignificant ing and drinking was significant, but in the negative
positive within-person association between daily event stress direction. Thus, individuals, on average. drank less on days
and drinking (b = .034, 1 = 1.580, 136 df, p = .114). How- when they used relatively more problem-focused coping.
ever, the slope variance component was still significant (u 1 Exponentiation of the partial slopes from Lhis model offers
.0 18, x2 = 174.56, 136 df. p = .0 14), and no change was a straightforward interpretation and an index of effect size:
found in the aggregate daily stress-average drinking asso- for each unit increase in stress, the rate of drinking changes
ciation (y = -.256, 1 = -2.787, 135 df, p = .006). Day-of- by a factor of (exp.(0.067) = 1.069) or increases by 6.9%;
the-week cflccts were as fol lows: Monday. b = -0.240, 2,4 11 for each unit increase in problem-focused coping, the rate
df, p .457; Tuesday, b = -0.067, 2,411 df, p = .870; of drinking changes by a factor of (exp.( -0.154) = 0.857)
Wednesday, b = 0.882, 2,411 dl~ p = .036; Thursday, b = or decreases by 14.3%. Neither emotion approach coping
I.950, 2,4 11 df, p < .00 I; Friday, b =- 2.859, 2,411 df, p < nor avoidance coping showed sign ificant within-person ef-
.00 I; Saturday, b = 2.684, 2,4 11 df, p < .00 I. fects, and only the emotion approach-drinking slopes dis-
Because, on average, there was not a significant within- played a significant variance component (see Random
person positive association between daily event stress and effects).
drinking, a strong case for mediation, which requires the Affect as a mediator of the stress coping-drinking asso-
exogenous predictor to be correlated with the dependent ciation. Daily negative affect was entered, person-centered.
variable (Baron and Kenny, 1986), cannot be made. How- into the Level I model, and aggregate negative mood was
ever, less stringent criteria for mediated effects, not requir- entered into the Level 2 intercept model; the negati ve af-
ing this effect, have been proposed wherein the exogenous fect-drinking slope was modeled as a random effect. Re-
predictor is associated with the mediator, which in tum is sults are shown in Table 2 (Model 2). Aggregate daily
associated with the dependent variable (Kenny et al., 1998). negative affect was positively associated with aggregate
Such effects are especially likely when multiple mediators drinking, and aggregate avoidance coping no longer pre-
have effects on the dependent variable in different direc- dicted aggregate drinking. At the within-person level. daily
tions. In the present case, linear multilevel models (treating negative affect showed a marginally significant unique ef-
all slopes as random effects) showed the following: Event fect. with individuals drinking more on days characterized
stress predicted problem-focused coping (b = 0.124, 136 by relatively higher levels of negative affect. The unique
df, p < .00 I). emotion approach copi ng (b = 0.1 02, 136 df, within-person effects of both daily stress and problem-fo-
p < .00 1) and avoidance coping (b = 0.043, 136 df, p < cused coping were still sign ificant. Significant or margin-
.00 I). Event stress (b = 0.054, 136 df, p < .00 I). emotion ally significant variance components were obtained for
approach coping (b = 0. 177, 136 elf, p < .001) and avoid- emotion approach coping. stress and negative affect slopes;
ance coping (b = 0.143, 136 d f, p < .00 I) were unique nonsignificant variance components were fixed to zero (sec
predictors of daily negative affect; and event stress (b = Snijders and Boskcr, 1999).
-0.107, 136 df, p < .00 I) was a unique predictor of daily Finally, we included positive affect in the model (sec
positive affect. These analyses are based on a subset of the Table 2. Model 3). Aggregate problem-focused coping was
data examined in Park et al. (in press). now a significant positive predictor of average drinking lev-
Coping as a mediator ofthe stress-drinking association. els. and aggregate positive affect was a marginal negative
The three daily coping variables were entered, person-ceo- predictor of average drinking levels. At Lhe within-person
PARK,A RMELIA DTE E 13 I

TABI E 2. Muhilc\el rc~_:rcsston resuhs predicting daily dnnking


umber of drinks
Model I Model 2 Model 3
Between-person coefficients (Level 2)
Aggregate emotion upproach -0.092 (.242) -0. 170 (.269) -0.184 (.263)
Aggregate avoidance 0.619 (.275)• 0.251 (.393) 0.149 {.349)
Aggregate probh:m-focuscd 0.216 (.218) 0.182 (.221 ) 0.375 (.182)•
Aggregate event Mress -0.339 (.094)1 -0.355 (.09W -0.377 (.098)1
Aggregate negative afTect 0.575 (.222)• 0.626 {.210)1
Aggregate posim e affect -0.412 { . 239)~
Pooled\\ ithin-pen.on cocffictents (Level I)
Daily emouon approach 0.()41 (.076) -0.021 {.077) -0.069 {.077)
Daily avoidance 0.052 (.070) -0.025 (.073) -0.013 (.071)
Daily problem-focused -0.154 (.053)' -0.155 {.048)1 -0.163 (.045)1
Daily event stre>' 0.067 (.025)1 0.06 1 (.025 ). 0.086 (.024 )1
Daily negative affect 0.158 (.089)~ 0.288 (.09 1)1
Daily positive affi:ct 0.181 (.046) 1
Random effects
Daily emotion approach 0.38 I' 0.228~
Daily avoidance 0.128
Daily problem-focused 0.1 15
Daily event stress 0.040 0.0421 0.0421
Daily negative afTect 0.302' 0.239
Daily posi tive afTeet 0. 109~

.VoLes: Values for bctwecn-pen.on and within-person fixed efTect~ arc the unstandardi.-ed pantal regression
coefficients; robust standard errors are in parentheses. Both models control for day of \\eek {contra;.ts not
shown). For bctwccn-pen.on effects: 132 df for Model I; 131 df for \llodel 2: 130 df for Model 3. For withm-
person efTeets: 136 df lor all models. except for problem-focused and avoidance coping in Model 2 (2.403 df)
and in Model 3 (2.401 df). Values for random effects arc vanance components for Level I slopes. Tests of
significance for random effects arc x2 with 122 df for Model I and x.2 with 132 df for Models 2 and 3.
*p < .IO:*p < .05: 'p · .01.

leveL the unique within-person effect for negative affect factors. As shown in Table 3, the within-person emotion
was now significant at the .05 alpha level. as was the unique approach coping-drinking slopes varied as a function of
within-person effect for positive affect. Thus. individuals. several person factors. Individuals high in social-enhance-
o n average, drank more on days characterized by relatively ment motives, those low in sensation seeking and drinking-
higher levels of positive affect and relatively higher levels to-cope motives. as well as those with a family history of
of negative affect. Event stress and problem-focused cop- alcohol problems displayed positive associations between
ing remained s ignificant within-person predictors. as did emotion approach coping and drinking. Figure I shows the
the day-of-the-week contrasts (not shown). There was a form of the effect for social-enhancement motives; high
marginally significant variance component for the positive and low social motives correspond to plus/minus one stan-
affect slopes; the variance component for the negative af- dard deviation. {Al l interactions were of similar cross-over
fect slopes was no longer significant. form , with those low in social-enhancement motives, those
Cross-le1•el interactions. ext. we included sex, neuroti- high in sensation seeking and drinking-to-cope motives. as
cism. sensation seeking, drinking-to-cope and social-en- well as those without a family history of alcohol problems
hancement drinking motives and family history of alcohol showing negative associations.)
problems in the Level 2 model. We first entered them as Sex and family history of alcohol problems were sig-
predictors of average drinking (Equation 2). Significant nificant predictors of the problem-focused coping-drinking
unique effects were found for sex (y - -0.98 1, t = -6.641, slopes. The form of the sex interaction was similar to that
124 df, p < .00 I); sensation seeking (y = 0.196, t = 2.909, shown in Figure I, with women showing a negative asso-
124 df, p = .004); social-enhancement motives (y = 0.224, r ciation and men showing a positive association. The form
= 2.829. 124 df, p = .005); and average negative mood (y = of the family history effect revealed that individuals with a
0.376. t = 2.451. 124 df, p = .014), with men and those family history of alcohol problems, compared with others.
high in sensation seeking, social-enhancement motives and demonstrated a stronger negative association between prob-
aggregate negative affect drinking more on average. lem-focused coping and drinking. Sex and sensation seek-
ext. we examined these factors as predictors of the ing were significant predictors of the event stress-drinking
within-person slopes. For exploratory purposes, we exam- slopes, and the fom1 of these effects indicated that men and
ined each Level I slope as a function of both the tension- those low in sensation seeking had a negative association.
reduction and social-enhancement individual-difference risk whereas women and those high in sensation seeking dem-
132 JOURt AL OF STUDIES 0 ALCOIIOL I JA UARY 2004

T\Bit 3. Mululc\·cl regression rC!>ults for cross-Jc,cl mtcrncttons


Between-person predictors
Within-pcn.on
predictor-drinJ.. ing Scnsauon Family DTC SE
slopes Sex Neuroticism scck111g history motives mouvcs
Emotion approach 0.280~ 0. 151 ~ -.206* 0.435' -0.28 1* 0.28 1!
(. 170) (.080) (.082) (. 151) (. 11 3) (.086)
Avoidance -0.274 0.121 0.032 0.076 -0.153* -0.074
(. 181 ) (.08 1) (.082) (. 181) (.090) (.086)
Problem-focused -0.3071 0. 11o~ -0.083 -0.270° 0.051 -0.022
(.I 18) (.066) (.060) (.123) (.073) (.081)
Event ~tress 0.139 1 -0.008 0.084 1 0.006 0.006 0.022
(.0·19) (.025) (.027) (.066) (.037) (.038)
Negauve affect -0.253 -0.212* -0.055 -0.077 -0.086 0.1 II
(. 186) (.090) (.099) (.184) (.097) (. 124)
Positi'e afi"cct 0.199~ 0.041 -0.110~ 0.108 0.048 0.039
(. 11 3) (.060) (.063) (.131) (.089) (.070)

Notes: Va lues arc Level 2 unstandardi7cd panial rcgrcs>ion coefficients: robust standard errors
arc in parentheses. DTC drinking-to-cope; SE • social enhancement.
§p < . 10: *p "'- .05; 1p < .0 1.

onstratcd a positive association. Finally, ncurottctsm pre- scs, usi ng the disaggrcgatcd data, including all of the inter-
dicted the negative affect-drinking slopes. with those low in actions fom1ed from the person-centered and grand-mean-
neuroticism demonstrating a stronger positive association. centered variables. Variance inflation factors (VIFs) revealed
Supplemental analyses were conducted to examine the no problems (i.e., all VIF values< 3.6).
moderating effects of the between-person factors on other
legs of the stress-coping-negative affect process. We rcpon Discussion
only the cross-level interactions significant at the .05 alpha
level; all models used a standard linear setup and treated In general , our find ings underscore the importance of
all slopes as random effects. In predicting negative affect examining the stress-drinking process at multiple levels of
from event stress and coping, neuroticism moderated the analysis. Examination of the simple associations among the
event stress-negative affect association (y = 0.0 13, 1 = 2.00, aggregate daily variables (i.e., the between-person level of
130 df. p = .046), with individuals higher in neuroticism analysis) indicated that those who on average perceived
showing tronger positive associations. In predicting cop- events as more stressful actually drank less, on average.
ing from event stress, only sex moderated the event stress- than those who perceived events as less stressful. !low-
emotion approach association (Y = 0.055. 1 = 3. 13, 130 df, ever. this finding does not inform us as to whether indi-
p = .002), with women showing stronger positive associa- viduals' drinking, on a given day. increases or decreases as
tions. Finally, we examined the final models for problems a function of deviations from their average stress level.

with multicollineariry. We ran standard regression analy- At the daily level of analysis, a complex process was
found wherein variation in alcohol consumption was
4.5 uniquely associated with a variety of factors. including those
Low Soctal Motives that make up the tension-reduction pathway. The effect of
4
daily event stress on alcohol use was apparent only after
3.5 - Htgh Social Motives controlling for day-of-the-week and daily coping, and this
.....,.c 3 ,.. effect remained significant with positive and negative af-
fect in the model. Although it was expected that inclusion
/
...8
.8
0
~

E
2.5

2
./
,.,. / of the day-of-the-week predictors would enhance the stress-
drinking effect by removing drinking variation not related
z" 1.5 ~..., to stress (e.g., social drinking), it was not expected that
...- ...- event stress would rema in significant with the inclusion of
the key process variables (coping and affect) in the model.
o.s One possibiliry is that the remaining unique aspect of stress
appraisal might be its cognitive components of threat or
0 loss. Future research should examine distinct cognitive ap-
Hog)!
Emooon Approach Copng
prai als on a daily basis to funher verify thjs interpretation.
Ft<;~u I. The moderating effect of soctal-cnhancemcnt motives on the Our lack of evidence for a fu lly mediated model
assoctauon between daily emotion approach copmg and drinking wherein the effects of daily stress on alcohol usc take place
PARK. ARMEL! A D TE E 133

solely through coping and mood- docs not rule out the cation, such findings could indicate that. for such individu-
possibility that event stress influences drinking via these als, drinking occasions also arc times at which ongoing
factors. Instead, our findings raise the possibility of a com- problems are discussed with friends, regardless of the day's
plex system of indirect effects. For example, stress apprais- stress and affective valence. Hussong et al. (200 I) presented
als were positively associated with emotion approach and somewhat consistent findings wherein individuals with rela-
avoidance coping, both of which were positively related to tively higher friendship quality demonstrated stronger posi-
negative affect, which in turn was positively associated with ti ve associations between sadness and hos tility and
drinking. In addition, event stress was positively related to subsequent drinking. However, friendship quality was a per-
problem-focused coping, which was negatively related to son-level variable, not an indicator of daily interaction.
daily drinking. These patterns could indicate that different Moreover, Hussong et al. found the opposite effect with
aspects of stressful daily encounters might increase and/or perceived social support. Further research is needed on the
decrease the amount of alcohol consumed. For example, to role of interpersonal interaction as a form of coping in the
the degree that a daily stressor elicits problem-focused cop- stress-drinking process.
ing, drinking might be reduced due to the efforts associated We also found that the within-person problem-focused
with such action. In contrast, to the degree that a daily coping and alcohol use association varied as a function of
stressor evokes negative affect, ind ividuals might be more sex and fami ly history of abuse, and that the associations
likely to engage in drinking. Replication of these findings between daily stress appraisals and drinking varied as a
is needed, and more fine-grained within-day sampling could function of sex and sensation seeking. We are reluctant to
help to tease out the sequential effects posited. offer interpretations of these exploratory findings. How-
The lack of a direct association between daily alcohol ever, we believe that these complex interactive findings
usc and avoidance coping and emotion approach coping might start to shed light on some of the inconsistencies
was inconsistent with our hypotheses. However. consistent (e.g.. concerning sex effects) found in studies examining
with prior research. we did find associations among aggre- the stress-drinking process.
gate avoidance coping and drinking. Thus, one possibility One important caveat to all of the coping findings is
is that, although avoidance coping might not be proximally that individuals reported on their coping up to that point in
related to alcohol use, regular usc of such a strategy exac- the day. Thus. an assumption being made is that such cop-
erbates ongoing problems over time, leading to higher lev- ing extended into the even ing when drinking occurred. Fu-
els of negative mood. which in turn arc associated with ture research using more fine-grained within-day reporting
increased drinking. On the other hand, our lack of effects for is needed to examine concurrent coping and drinking to
avoidance coping might be due to incorrect conceptualization further understand this association.
and measurement of the avoidance construct. Our measure Findings from the present study also provided prelimi-
of avoidance coping consisted of items such as "I've been nary evidence for motivational models positing positive and
giving up trying to deal with it'" and "I've been saying to negative affective pathways underlying alcohol consump-
myself ' this isn't real. "' As a coping strategy. avoidance is tion (e.g.. Cooper ct al., 1995). In addition to the effect of
an intentional effort to cognitively escape potentially pain- negative affect on daily drinking, individuals, on average,
ful circumstances or to refocus one's attention. Whereas tended to drink relatively more on days characterized by
giving up and saying "this isn' t real" appear to be auto- relatively higher levels of positive affect. This effect may
matic reactions to negative events and perhaps even mani- be related to the celebratory and socializing functions of
festations of distress rather than deliberate efforts to deal college drinking and might be mediated via increasing ex-
with such events, a coping item such as " I've given up pectations of such social occasions as the week progresses.
trying to deal with this problem in order to allow me to We removed weekly cycles in alcohol usc, but not in posi-
reevaluate my goals'" might better capture the deliberate tive affect. Trends indicated that positive affect increased
nature of coping strategies (cf.. Tcnncn and Affleck. 1997). as the week progressed, peaking on the weekends. One pos-
Although the pooled association between the emotion sibility is that positive affective states co-occur with in-
approach coping and drinking was not significant, this as- creased anticipation of expected pleasurable outcomes from
sociation varied across persons and was associated with alcohol-related social situations (e.g., going to parties) for
several risk factors. Again, it should be noted that the ef- the coming weekends. Future stud ies assessing expectan-
fect of emotion-focused coping was a partial effect. Thus, cies at the daily level of analysis arc needed to test this
it is the unique aspect of "expressing emotions" and "seek- interpretation.
ing emotional support" that seems to be associated with Findings from the present study also suggest that these
increased alcohol use. on a daily basis, for individuals high two motivational pathways might be intertwined. We found
in social-enhancement motives, those with a family history that person factors (e.g., social-enhancement motives, sen-
of alcohol problems and those low in sensation seeking sation seeking). purportedly related to the social-enhance-
and drinking-to-cope motives. Although in need of repli- ment motivational pathway. were associated with variation
134 JOUR AL OF STUDIES 0 ALCOHOL I JA UARY 2004

in within-person emotion approach coping-drinking asso- BtRGIIUI~. J.P. AM> SrA,'TON. A.L. Adju~unent to a dyadic stressor: A
ciations. Future research is needed to further explicate the longitudinal study of coping and depressive symptoms in infertile
couples over an insemination attempt. J. Cons. ('lin. Psycho!. 70: 433-
roles of such risk factors in the stress coping-drinking pro-
438. 2002.
cess. Finally, it should be noted that the associations be- 8RYK, A.S. A"t> RAtt>t Noesu. S. W. llicrarclucal Linear Models: Applica-
tween the tension reduction variables and drinking became tions and Data Analysis Melhods. Thousand Oaks. CA: Sage. 1992.
stronger aficr controlling for positive mood, highlighting CAR'IEY. M.A .. AR\tl Ll, s.. Tt).'<J "· II .. AHUO.. G. A'D 0' Ill. T.P.
the importance of controlling for the many antecedent fac- Positive and negauve daily event>. percetvcd stress. and alcohol usc:
tors underlying daily drinking behavior. A diary study. J. Cons. Clin. Psycho!. 68: 788-798. 2000.
CARVER. C.S. You want to measure coping but your protocol's too long:
A number of limitations of this study must be noted. Consider the llricfCOPE. Int. J. Bchav. Mcd. 4: 92-100. 1997.
First, our stress and coping measure focused only on the CONGFR. J.J. Reinforcement theory and the dynamics of alcoholism. Q. J.
day's most negative event. which in most cases was a rela- Stud. Alcohol 17: 296·305. 1956.
tively minor problem. This strategy precludes the efTcct of CooPER. M.L. Momations for alcohol usc among adolescent~: De' elop-
multiple negative events. and associated coping, on later- ment and 'ahdauon of a four-factor model. Psycho!. Assess. 6: 117-
128. 1994.
day drinking. The decision to measure only the most nega- Coor~R. M.L., FRoNt, M.R., Russf tt. M. AND Mtt>AR. P. Drinking to regu-
tive event was guided by practicality. I laving participants late positive and negative emotions: A motivational model of alcohol
list all of their negative events and subsequent coping might usc. J. Pcrs. Social Psycho!. 69: 990-1005. 1995.
have caused excess fatigue in completing the daily report. COOPER. M.L .. Rus~ttt, M. AND GtOR(,J, W.ll. Coping. expectancies. and
As is the case in all daily studies, it is a trade-off between alcohol abuse: A test of social learning fom1ulations. J. Abnonn.
Psycho!. 97: 218-230. 1988.
participant burden and coverage of the variables of interest.
COOPER. M.L .. R~~lll, M.. SKI''<~R. J.B.. FRO'-f. M.R. "'() Mt D\R, P.
Second, we focused only on the same-day associations Stress and alcohol usc: Moderating cnccb of gender, coping, and al-
among stress, coping, affect and alcohol use. As stated ear- cohol expectancies. J. Abnorm. PsychoI. I 0 I: 139-152. 1992.
licr, our model is only one way to conceptualize the causal COSTA, P.T. ANt> MtCRAt·, R.R. Revised NEO Personality Inventory ( EO-
association among these constructs (cf., Hussong ct al., Pl-R) and the NEO Five-Factor Inventory (NE.O-FFI): Professional
200 I). Thus, the time interval through which stressful events Manual. Odessa, FL: Psychological Assessment Resources. 1992.
FROMMI·. K. A'It> Rt\1 T, K. Young adults' coping style as a predictor of
innuence drinking (via the proposed mediators) might not lheir alcohol usc and response to daily events. J. Youth Adolcsc. 23:
be fixed. In some situations it might be a few hours (i.e., 85-97. 1994.
same day); in others it might be several days. This interval GREt,LEY. J. AI>D Ott, T. Alcohol and tension reduction. In: LFO,ARil. K.E.
might change as a function of a variety of factors. For A~>D BLA >~t. II.T. (Eds.) Psychological Theories of Drinking and A leo-

example, alcohol might not always be avai lable, especially holism. New York: Guilford Press. 1999. pp. 14-53.
ll ussoNc;. A.M .. lltcKs. R.E.. LEvv, S.A. ,,Nn CURRAN. P.J. Specifying the
among underage students, and other constraints might arise relations between a!Teet and heavy alcohol usc among young adults. J.
that restrict drinking (e.g., nighttime employment). More- Abnom1. Psycho!. 110: 449-461.2001.
over. this interval might covary with drinking level. For II UTC'Iti'-SO'. G.T .. PATOCK-PtCKJtA\1, J.A.. Cmo'<G. J. AW ACo<>~tu. C.T.
example, problem drinkers might exhibit more proximal Irrat ional beliefs and behavioral mi~rcgulauon m the role of alcohol
temporal cfTccts of stress on alcohol usc. Studies using more abuse among college students. J. Ration. Emot. Cog. Beha' . Ther. 16:
61-74, 1998.
detailed event and constraint reporting arc needed to test
K~NNY. D.A .. KA\IIY, D.A. AI'D Bot c.~R. . Data analysis in social psy-
these possibilities. chology. In: Gum·RT, D.T .. f'tSKt', S.T. ANt> LtND7E\', G. (Eds.) The
Finally, the fact that women and whites constituted the Handbook of Social Psychology. Vol. I. 4th Edition, New York:
large majority of the sample was problematic. It limited McGraw-llill. 1998. pp. 233-265.
our ability to examine differences in the effects of the be- LARSE>~. R.J. A't> Ott" R. E. Promise~ and problems with the ctrcumplcx
model of emouon. In: CLARK. M.S. (Ed.) Emotion. Thousand Oaks.
tween-person risk facto rs across gender and race.
CA: Sage. 1992. pp. 25-59.
LAZARUS, R.S. Copmg theory and research: Past. present. and future.
Acknowledgment Psychosom. Med. 55: 234-24 7. 1993.
LA7~\RUS, R.L. AND FolKMAN, S. Stress, Appraisal. and Coping. New York:
The authors greatly appreciate ick Maltby's assi$tancc in conducting Springer. 1984.
this study over the World Wide Web. MOIIR. C.D.. ARMIII, S., TESl\E)o, H.. CAR)ol \, M.A.. Alll f( l(, G. AM)
HR0 \11. A. Daily mtcrpersonal experiences. context and alcohol con-
References sumption: Crying in your beer and toaJottng good times. J. Pers. Soctal
Psycho!. 80: 489-500. 2001.
ARMI u. s.. CAR'I Y, M.A.. Tt '<'IHI, H.. Aflll n... (j, A'D o· Ell. T. Stress PARK, C.L.. ARMttl, S. A'D TE'I"£'· II. Appmtsal-coping goodness of lit:
and alcohol u:.c: A daily process cxarnin;uion of the stressor-vulner- A daily mtcmet study. Pcrs. Social Psycho!. Bull., in pres>.
abi lity model. J. Pcrs. Social Psycho!. 78: 979-994, 2000. PROJE('T MATCII Rl,st.ARCII GROUt'. Mntching alcoholism treatment to cli-
llA~R. J.S. Student litctors: Understanding individual variation in college ent heterogeneity: Project MATCH three-year drinking ou tcomes.
drinking. J. Stud. Alcohol. Supplement o. 14, pp. 40-53. 2002. Alcsm Clin. f:xp. Res. 22: 1300-1311. 1998.
13AR0". R.M. A'D Kt 'w. D.A. The modcmtor·mcdiator variable distinc- RA~o~"BU~II. S.W. \1\D BilYK. A.S. llicrarclucal Linear Models: Applica-
tion in social psychological research: Conceptual. stmtcgic. and ~tatts­ tions and Data Analysis Methods. 2nd Edition. Thousand Oal.lo, CA:
tical considcmuon~. J. Pcrs. Social PsychoI. 5 I: 1173-1182. 1986. Sage. 2002.
PARK. ARMELJ ANDTENNEN 135

RA~;of'<IJl'SH. s.w.. BKYK.


A.S .. CHEONG, Y.F. N~o Co:-.:GOON, R. IILM 5: TE.,NE'I. II. AND 1\FILF.tK, G. Social comparison as a coping process: A
Hierarchical Linear and Nonlinear Modeling. Chicago. IL: Scientific critical rcvicw and application to chronic pain disorders. In: BuusK.
Sofiware lntemational. 200 I. B.P. ,\Nil GllliJONS. R.X. (Eds.) Health. Coping and Well-Being: Per-
Smw,,Rit.. J. AND STONe. A. Strategies for analyzing ecological momen- spectives from Social Comparison Theory. Mahwah. NJ: Lawrence
tary assessment data. lllth Psychol. 17: 6-16. 1998. Erlbaum. 1997. pp. 263-298.
S1nHMA '1. S. Real-time self:rcpon of momentary states in the natural en- TeNNEN. H.. AH·L~('K . G.. ARMELI. S. ANI) CARNI'Y, M.A. A daily process
vironment: Computerized ecological momentary assessment. In: STONE, approach to coping: Linking theory, research, and practice. /\mer.
A.A .. TuRKKAN. J.S .. BAtiiRACII. C.A .. JonE, J.B.. KuRri.MAN. H.S. ,,No Psychol. 55: 626-636, 2000.
CAIN. V.S. (Eds.) The Science of Self-Repon: Implications for Research TFRRY, D.J. AND HvNtiS. G.J. Adjustment to a low-control situation: Reex-
and Practice. Mahwah. NJ: Lawrence Erl baum. 2000. pp. 277-296.
amining the role of coping responses. J. Pers. Social Psychol. 74:
StMrsoN. T. ANt> ARROYO. J.A. Coping patterns associated with alcohol-
I078-l 092. 1998.
related negative consequences among college women. J. Social Clio.
WATSON. D.. Ct.ARK. L.A. A'<D TEtLF.GEN, A. Development and validation
Psychol. 17: 150-166. 1998.
S'uDERS, T.A.£3. A'ID 13osK~R. R.J. Multilevel Analysis: An Introduction to of brief measures of positive and negative affect: The PANAS scales.
Basic and Advanced Multilevel Analysis. Thousand Oaks. CA: Sage. J. Pcrs. Soci:tl Psycho!. 54: 1063-1070. 1988.
1999. WtNDLE. M. ,\ND Wl'lt>LE. R.C. Coping strategies. drinking motives. and
STAI\'TON. A.L.. KIKK. S.B.. CAMERo:-~. C.L. AND DAI\OH-13URG. S. Coping stressful life events among adolescents: Associations with emotional
through emotional approach: Scale constntction and validation. J. Pers. and behavioral problems. and academic functioning. J. Abnorm.
Social Psychol. 78: I 150- 1169. 2000. Psycho!. IOS: 551-560. 1996.

You might also like