Method
Participants
In total, 78 female students aged 18-28 years old (M = 21.7, SD = 4.3) from a university in Tehran participated in the study, which was announced as a student project. They were recruited via social media and on-campus sites. All participants were solely Iranian and were at the undergraduate (84%) and postgraduate (16%) levels. Furthermore, participants received feedback about their emotion regulation as an incentive, and those who completed at least 80% of the episodes were rewarded. The sample size and design of the study are sufficient to identify correlations with a minimum effect size of r = 0.22, with a desired power of 0.80. This suggests that the study is capable of detecting relatively small, yet still meaningful, effects within the moderate range, making it suitable for exploratory research purposes.
Procedure
In December 2023, participants provided written informed consent and completed a survey that gathered demographic information (age, gender, and education level) and self-reported distress tolerance capacity during the screening process. Following the initial three-week measurement period, individuals with poor distress tolerance were invited to a one-hour training session, during which the ethical considerations and guidelines for completing daily measurements of emotion regulation goals and strategies were explained. The participants could complete the questionnaire for that day from 6-11 pm through a secure website. A reminder message was sent to the participants by email every evening at 7 pm. The average time to answer the questions was 86 seconds (SD = 12). The aforementioned time was 43 seconds (SD = 9) in episodes without emotion regulation. The participants who completed less than 50% of the episodes (n = 7) were excluded. The average response for each participant was 9.47 (SD = 2.83). Finally, a total of 904 reported episodes were collected, but 126 episodes were not included in the analysis because of no recall of negative emotion or emotion regulation or being completed in less than 30 seconds. The researcher was available via email to address participants' questions or issues throughout the study.
Measures
At the screening stage, the participants' capacity for distress tolerance was measured using the tolerance subscale of Distress Tolerance Scale (Gaher & Simmons, 2005), which consists of three items scored on a 5-point Likert scale. Participants who scored less than 9 were included in the study. In each episode, if participants reported experiencing negative emotions and attempting to regulate them, they were directed to items related to regulatory goals and strategies. They were asked to recall the event with the most negative emotions and rate the intensity and valence of that emotion to improve their clear recall of the emotion and event. Regulatory motives included pro-hedonic, pro-social, and impression management goals, and the examined strategies were cognitive reappraisal, expressive suppression, experiential avoidance, and rumination, similar to the first study. The degree to which each of the goals and strategies was employed in regulatory efforts was measured on a 7-point scale (1: Not at all to 7: Considerably). Each of the goals (e.g., pro-hedonic goals: I aimed to increase my positive emotions or reduce my negative emotions) and strategies (cognitive reappraisal: I tried to alter my thoughts or look at the situation from a different perspective) was measured with an item adapted from the questionnaires used in Study 1. To control the duration of the daily reports, additional questions related to the events of the same day (e.g., I spent most of the day on my assignments) were asked in cases where participants reported no experience of negative emotion or no attempt to regulate such emotion.
Data analysis
Multilevel modeling was employed using the lme4 package in R software to account for the nested structure of the experience sampling data, which consisted of multiple measurements from each participant. This approach enabled the accurate parsing of variance at both the between- and within-person levels (Singer & Willett, 2003). The daily responses were nested within participants, with Level 1 representing the day level and Level 2 representing the participant level. Consequently, Level 1 predictors included time and within-person goals, while Level 2 contained between-person goals. To separate within- and between-person levels of variables, each goal was grand-mean centered, and the grand-mean centered goals were person-centered. Prior values of each endogenous variable were controlled for to account for changes that occurred before the experience. To address minor deviations from normality (skewness or kurtosis > 1.00) in certain variables, a square root transformation was applied to all variables. Maximum likelihood estimation was used to account for missing data. To examine the relationships between goals and strategies, separate two-level models with within- and between-person scores were designed for each goal, predicting each strategy simultaneously. Random slopes were calculated to account for within-person effects, and no convergence problems were encountered. An alpha level of 0.05 was chosen as the significance threshold to detect significant effects while minimizing the likelihood of chance findings. Following the recommendations for multilevel modeling, semi-partial R² (Rβ²) values were calculated as an index of effect size (Edwards et al., 2008). The magnitude of small, medium, and large effects was set at 0.02, 0.13, and 0.26, respectively. Table 4 presents the intercepts (representing mean strategy use), unstandardized coefficients, standard errors, Rβ² values, and 95% confidence intervals.
Results
The sample's distress tolerance scores ranged from 3 to 8, with a mean of 6.13 (SD = 1.94). On average, the participants attempted to regulate their negative emotions in 73% (SD = 0.18) of episodes where they experienced negative emotions. Impression management (intercept = 4.02) was considered the most prominent regulatory goal, followed by pro-social (intercept = 3.57) and pro-hedonic (intercept = 3.16) goals, respectively. The strategies employed most frequently were experiential avoidance, expressive suppression, cognitive reappraisal, and rumination, respectively (see Table 2). The intraclass correlation coefficient (ICC) describes the proportions of between- and within-person variance (see Table 3), which can be interpreted as how similarly emotion regulation goals and strategies vary over time. A low ICC (e.g., cognitive reappraisal = 0.35) indicates that the observed variance in a variable is largely determined by the event and day, and the pattern of the variable varies significantly within individuals. In contrast, a high ICC (e.g., impression management = 0.64) suggests that the variance is primarily determined by the individual, and the pattern of the variable is similar across different events and days within one individual, but differs among individuals (state versus trait; Wilms et al., 2020). The model's goodness of fit was good, as indicated by the robust root mean square error of approximation (0.04) and the robust standardized root mean square residual (0.02), both of which were below the 0.05 threshold (Kline, 2016).
Table 3 Mean, SD, and ICCs for Emotion Regulation Goals and Strategies
IM
|
PS
|
PH
|
RU
|
EA
|
ES
|
CR
|
|
4.66(1.84)
|
3.85(1.10)
|
3.07(1.41)
|
3.94(1.26)
|
5.98(1.19)
|
4.65(1.05)
|
3.34(1.16)
|
Mean (SD)
|
.64
|
.47
|
.42
|
.38
|
.72
|
.66
|
.35
|
ICC
|
Note. ICC: intraclass coefficient, CR: cognitive reappraisal, ES: expressive suppression, EA: experiential avoidance, RU: rumination, PH: pro-hedonic, PS: pro-social, IM: impression management. *p < .05, **p < .01.
Table 4 MLM lagged analyses for daily emotion regulation goals predicting daily strategy use
|
Cognitive reappraisal
|
Expressive suppression
|
Experiential avoidance
|
Rumination
|
Predictor variables
|
B(SE)
|
Rb2
|
95%CI
|
B(SE)
|
Rb2
|
95%CI
|
B(SE)
|
Rb2
|
95%CI
|
B(SE)
|
Rb2
|
95%CI
|
Intercept
|
|
3.12
|
|
|
3.45
|
|
|
3.77
|
|
|
2.87
|
|
Pro-hedonic
|
|
|
|
|
|
|
|
|
|
|
|
|
Within-person
|
.48**(.02)
|
0.01
|
[.28, .61]
|
-.37**(.01)
|
.02
|
[-.47, -.29]
|
-.05(.07)
|
.03
|
[-.16, .11]
|
-.10(.02)
|
.01
|
[-.17, -.02]
|
Between-person
|
.41**(.04)
|
0.01
|
[.17, .56]
|
-.25*(.06)
|
.01
|
[-.38, -.13]
|
-.01(.06)
|
.00
|
[-.13,.16]
|
.02(.04)
|
.01
|
[-.03, .08]
|
Pro-social
|
|
|
|
|
|
|
|
|
|
|
|
|
Within-person
|
.31*(.04)
|
0.01
|
[.19, .41]
|
.36**(.02)
|
.00
|
[.22, .48]
|
.26*(.04)
|
.00
|
[.18, .31]
|
-.08(.03)
|
.03
|
[-.12, -.03]
|
Between-person
|
.21(.05)
|
0.00
|
[.15, .27]
|
.24*(.04)
|
.00
|
[.09, .41]
|
.28*(.05)
|
.00
|
[.17, .39]
|
.01(.02)
|
.00
|
[-.01, .04]
|
Impression management
|
|
|
|
|
|
|
|
|
|
|
|
|
Within-person
|
.04(.04)
|
0.00
|
[-.02, 0.7]
|
.37**(.02)
|
.02
|
[.25, .48]
|
.56**(.03)
|
.02
|
[.42, .67]
|
.29*(.06)
|
.01
|
[.17, .43]
|
Between-person
|
-.11(.03)
|
0.00
|
[-.02, .21]
|
.29*(.05)
|
.01
|
[.20, .39]
|
.46**(.05)
|
.01
|
[.33, .58]
|
.19(.08)
|
.01
|
[.08, .29]
|
REs
|
B(SE)
|
Z
|
95%CI
|
B(SE)
|
Z
|
95%CI
|
B(SE)
|
Z
|
95%CI
|
B(SE)
|
Z
|
95%CI
|
Level 2 residual
|
1.32*(.13)
|
10.18
|
[1.27, 1.46]
|
1.37* (.07)
|
10.67
|
[1.29, 1.46]
|
1.52*(.13)
|
10.81
|
[1.38, 1.66]
|
1.30*(.01)
|
9.72
|
[1.29, 1.32]
|
Level 1 residual
|
.84*(.07)
|
25.33
|
[.75, .93]
|
1.18*(.08)
|
27.61
|
[1.07, 1.27]
|
1.74*(.04)
|
28.35
|
[1.69, 1.78]
|
.93*(.03)
|
25.48
|
[.88, .97]
|
Autocorrelation
|
.15(.02)
|
9.24
|
[.11, .18]
|
.19(.02)
|
5.83
|
[.17, .22]
|
.26(.03)
|
9.72
|
[.21, .30]
|
.07(.02)
|
3.38
|
[.04, .11]
|
Note. B(SE): unstandardized fixed effect estimates standard errors in parentheses, 95%CI: 95% confidence intervals, Rβ2: semi-partial effect size, Z: wald’s Z; Intercept: mean pursuit of each goal across days, RE: Random effect. Random slopes were included unless models showed convergence issues. Between-person effects were grand-mean centered. *p < .05, **p < .01.
Both within-person and between-person effects were calculated to determine the impact of goals on strategies. The within-person results indicate the relationship between pursuing a goal (low or high) and the use of strategies, whereas the between-person results represent the degree of difference between individuals in using strategies based on the average regulatory goals reported. At the within-person level, individuals with higher pro-hedonic goals utilized more cognitive reappraisal (p =.004) and less expressive suppression (p =.009). Pro-social goals were positively related to experiential avoidance (p =.035), cognitive reappraisal (p =.017), and expressive suppression (p =.008) strategies. Furthermore, a positive relationship was observed between impression management goals and experiential avoidance (p <.001), expressive suppression (p =.004), and rumination (p =.038). The aforementioned relationships were significant after controlling for other goals and time. In certain instances, the within-person results exceeded the between-person effects, but they were mostly comparable. For example, individuals with stronger impression management goals tended to use more experiential avoidance (p =.002) and expressive suppression (p =.018) on average. Additionally, individuals with a greater focus on hedonic goals tended to utilize more cognitive reappraisal (p =.005) and less expressive suppression (p =.020).
Discussion
The participants were screened for having poor tolerance for aversive states, which indicates that they habitually perceive distress and negative emotions as unbearable and uncontrollable. They reported stronger social goals during emotion regulation than pro-hedonic goals, and, as expected, they used experiential avoidance more than other strategies. Furthermore, the results show that in the daily emotion regulation attempts of individuals with low distress tolerance, goals play a critical role in selecting strategies. The direct relationships between goals and strategies were consistent with the first study, and each goal predicted the use of different strategies. However, impression management was related to more strategies and showed stronger relationships overall. In terms of variability, based on ICC scores, pro-hedonic and pro-social goals were used with moderate within-person variability. However, impression management goals exhibited greater stability across the reported episodes (trait). Among the strategies, cognitive reappraisal and rumination showed greater within-person variability (state), while expressive suppression and experiential avoidance exhibited greater between-person variability (trait). Nonetheless, it is important to interpret the variability cautiously as an indicator of flexibility in emotion regulation goals and strategies. Since ICC scores can be influenced by various factors, such as the number of episodes measured throughout the day, the number of items in each assessment episode, and the extent of actual changes in reported emotional situations.