Multiple Regression it Is Hypothesized That Self-Efficacy,

Multiple Regression

It is hypothesized that self-efficacy, appraisal, challenge and resources will negatively predict perceived stress whilst avoidant strategies will increase levels of perceived stress.

Sample Description

The sample consisted of 100 participants, 27 males and 73 females. Forty percent of the participants were not students, 30 were studying as part time students and 29% were studying as full time students. Fifty-five percent of the sample was employed full time, 25% were employed part-time and 20% reported being unemployed. Participants in the sample ranged from 16 to 70 years of age, with an average age of 35.6.

The assumption of normality was met for each individual variable. All of the variables included in the regression analysis displayed relatively normal distributions. The distribution for Self-Efficacy has a slight negative skew, but not one that should influence the interpretation of the results. An examination of scatter plots also revealed no problems with respect to linearity.

The correlation matrix shows that many of the predictor variables are correlated with one another, indicating that their unique contribution to predicting the outcome variable of perceived stress might be limited, however no multicolinearty is evident.

A small number of univariate outliers were found on individual measures, but no cases were identified through Mahalanobis distance as multivariate outliers with p < .001. The regression analysis was run with and without the univariate outliers; the results remained robust, and thus the results with outliers included are presented below.

Regression Analysis

A multiple regression analysis was conducted to evaluate how well 5 different variables predict perceived stress. The predictors were individual self-efficacy (SELF-EFFICACY), how an individual appraises stressful situations (APPRAISAL), whether or not an individual views a situation as a challenge (CHALLENGE), an individual's use of resources (help seeking) (RESOURCES), and whether or not the individual avoids the situation (AVOIDANT). The results of the multiple regression analysis are shown in Figure 1. The regression equation with all 5 predictor variables was significantly related to perceived stress, R2 = .279, adjusted R2 = .240, F (5, 94) = 7.263, p = .000. The Bs, as labeled on the output, are the weights associated with the regression equation. According to the B. weights, the regression equation is as follows:

Predicted Perceived Stress = -.24 Self-Efficacy - .34 Appraisal -.30 Challenge +.62 Avoidant + .51 Resources + 45.13

More useful for comparing the relative importance of predictors are the standardized weights, labeled as Beta on the output. The prediction equation for the standardized variables is as follows:

ZPredicted Perceived Stress = -.15 ZSelf-Efficacy -.19 ZAppraisal -.09 ZChallenge + .28 ZAvoidant + .18 ZResources

The sample multiple correlation coefficient was .53, indicating that approximately 28% of the variance in perceived stress in this sample can be accounted for by the linear combination of the 5 predictor variables. Thus, 72% of the variance in perceived stress is not accounted for by any of the predictor variables in this particular model for this particular sample. The relative strength of individual predictors can be examined through their zero-order and partial correlations with the predicted variable of perceived stress. As expected, self-efficacy, appraisal and challenge were all negatively correlated with perceived stress, and avoidance was positively correlated. Unexpectedly, resources was also positively correlated with perceived stress. Only the partial correlation between perceived stress and avoidance was significant. Avoidance on its own accounted for 6% of the variance in perceived stress, with the other variables contributing an additional 22%. Despite this, the existence of mild to moderate correlations between the predictor variables makes it difficult to make solid judgments about the relative importance of each individual predictor. The correlations among the predictor variables ranged from .16 to .48. Notably, resources, the variable that did not fit with the hypothesis, had the fewest number of significant correlations with the other predictor variables.