ابراز هیجان و علائم روانی: مقایسه نوشتن و نقاشی

Abstract The psychological effects of writing versus drawing about a stressful experience were compared. Participants were randomly assigned either into a control group (n = 34), a journaling group (n = 29), or a drawing group (n = 29). A measure of psychological symptoms was completed before and after two 15-min sessions. The results indicated that participants in the journaling group had a significantly greater decrease in their psychological symptoms than those in the drawing group or the control group. Journaling was also found to have a more positive effect for participants who initially had higher levels of psychological distress.

Results Power According to the meta-analysis by Symth (1998) concerning the effect size of written emotional expression, the resulting Cohen's effect size index d = .47 was used as an estimate of the possible effect size and was transformed into an estimate of the Cohen's effect size index f, which was about .20 in this case, for a three group analysis of variance/covariance. With the sample size of around 30 for each group, an estimate of the power of the study is .37 ( Cohen, 1988). Hence, because of the small sample size, non-significant findings of this study might be attributed to the small power of the study. Participants’ experiences with journaling and drawing Experience with journaling and drawing In the pre-questionnaire, 80 percent of the participants reported that they never or seldom (less than three to four times a year) journaled about their feelings, whereas 20% journaled at least monthly. With regard to drawing or doing art, 70% of the participants indicated that they seldom or did not engage in these activities in their life, whereas 30% did such activities at least monthly. Cochran's Q procedure was conducted to determine if the proportions of participants engaging in the journaling and drawing are significantly different. The result was not significant, Cochran's Q = 3.24, p > .05. Comfort levels with different expressive methods There was a trend for participants to feel more comfortable (either very comfortable or a bit comfortable) with journaling than with drawing. More specifically, in the pre-questionnaire 61% of participants indicated being comfortable with journaling feelings, 55% with journaling, 55% with expressing feelings, 23% with drawing, and 17% with drawing feelings. Cochran's Q procedure was used to determine if the proportions of participants feeling comfortable among different expressive methods were significantly different. The result was significant, with Q = 81.28, p = .00. For the reason of parsimony, three post hoc pairwise comparisons, using Bonferroni correction of α level of .017, were conducted: between journaling feelings and journaling, between journaling and drawing, and between drawing and drawing feelings. Between journaling feelings and journaling, the result was non-significant, with Q = 0.86, p > .017. Between journaling and drawing, the result was significant, with Q = 24.03, p = .00. Between drawing and drawing feelings, the result was non-significant, with Q = 3.57, p > .017. SCL-90-R Global Severity Index (GSI) distribution The pre-GSI scores ranged from 0.02 to 2.56, and the mean is 0.62. The interquartile range is from 0.22 to 0.84. These data can be compared to the established norms for the GSI index (Derogatis, 1994), where the mean for non-patients is 0.31 and the mean for non-patient adolescents (ages 12–18) is 0.76. Between-group differences Analysis plan In order to determine the treatment effects (between-group differences) among the groups, the post-GSI score was used as a dependent variable, and the independent pre-GSI score was be used as a covariate. When the slopes of the regression lines are homogeneous, analysis of covariance can be performed in order to determine the significance of the resulting treatment effects. However, when the slopes are heterogeneous, calculating the adjusted means for the three groups does not yield a meaningful result because of the differential treatment effects among the groups across the various levels of the covariate. As a result, in this situation, individual pairs of comparisons are needed using the Johnson–Neyman Technique in order to calculate the regions of non-significance across the covariate for each comparison (Huitema, 1980). This enables the researcher to determine for which levels of the covariate are the treatment effect too small to be significant. Consequently, the region of significance, which is the remaining region of the covariate, can then be determined. Assumption of homogeneity-of-slopes As mentioned above, before attempting to employ the analysis of covariance to analyze the significance of between-group differences, a test for the assumption of homogeneity of slopes was conducted. With an α level of .05, the test of homogeneity-of-slope was significant (see Table 1). Hence, there were differential treatment effects across different levels of pre-scores in comparing between group differences. Consequently, calculating the adjusted means for the three groups would not yield a meaningful result, and instead, individual pairs of comparison were needed ( Huitema, 1980). Table 1. Analysis of variance for heterogeneity of slopes Among the three groups d.f. MS F p Heterogeneity of slopes 2 0.31 5.94 .00 Error of residual within-group 86 0.05 Between drawing and journaling Heterogeneity of slopes 1 0.52 12.90 .00 Error of residual within-group 54 0.04 Between journaling and control Heterogeneity of slopes 1 0.37 6.67 .01 Error of residual within-group 59 0.06 Between drawing and control Heterogeneity of slopes 1 0.04 0.73 0.40 Error of residual within-group 59 0.06 Table options Comparison between drawing and journaling For the comparison between the drawing group and the journaling group, a test for the assumption of homogeneity of slopes was performed. The result was significant, with F(1, 54) = 12.90, p = .00 ( Table 1). It indicated that differential treatment effects occurred with this comparison. As a result, the Johnson–Neyman technique ( Huitema, 1980) was used to calculate the region of non-significance, which was [0, 0.88]. Hence, there was no significant difference of the treatment effect between the drawing group and the journaling group within this region. However, as pre-GSI scores increased above 0.88, significant post-GSI score differences between the two groups were found, with the journaling group increasingly having relatively less distress over the drawing group ( Fig. 1). When pre-GSI was 0.88, the treatment effect was 0.14. When pre-GSI was 2.56 (maximum score in the sample), the treatment effect was 0.75. Hence, participants with greater levels of distress (higher pre-GSI scores) derived greater benefit from journaling than those from drawing. Regression lines of post-GSI scores against pre-GSI scores of the journaling ... Fig. 1. Regression lines of post-GSI scores against pre-GSI scores of the journaling group and the drawing group, showing the divide between the region of non-significance and the region of significance. Figure options Comparison between journaling and control For the comparison between the journaling group and the control group, the test of homogeneity of slopes was significant, with F(1, 59) = 6.67, p = .01 ( Table 1). The region of non-significance was calculated, which was [0, 0.95]. As pre-GSI scores increased above 0.95, significant post-GSI score differences were found, with the journaling group having less distress relative to the control group ( Fig. 2). When pre-GSI was 0.95, the treatment effect was 0.17. When pre-GSI was 2.56, the treatment effect was 0.59. This result indicated that participants with greater levels of distress derived greater benefit from journaling than those in control. Regression lines of post-GSI scores against pre-GSI scores of the journaling ... Fig. 2. Regression lines of post-GSI scores against pre-GSI scores of the journaling group and the control group, showing the divide between the region of non-significance and the region of significance. Figure options Comparison between drawing and control For the comparison between drawing and control, the test of homogeneity of slopes was carried out. The result was non-significant, F(1, 59) = 0.73, p > .05 ( Table 1). Therefore, ANCOVA was performed, using the pre-GSI as the covariate. The adjusted mean of the post-GSI of the control group was 0.65, and the adjusted mean of post-GSI of the drawing group was 0.61. The treatment effect, which was the difference between the two adjusted means, was 0.04. However, this result was non-significant, F(1, 60) = 0.35, p > .05 ( Table 2). Table 2. Analysis of covariance between drawing and control Source d.f. MS F p Adjusted treatment 1 0.02 0.35 0.56 Residual within-group 60 0.06 Table options In conclusion, the participants in the journaling group had a significantly greater decrease in their psychological symptoms than those in the drawing group or the control group. In addition, journaling was found to have a more positive effect for participants who initially had higher levels of psychological distress.