2. MARCO TEÓRICO CONCEPTUAL
2.1. MARCO TEÓRICO
2.1.11 MONITOREO Y EVALUACIÓN
2.1.11.3 Sistemas de información
A multiple regression was performed to determine if mental health could be predicted by measures of mental life (daydreaming and life orientation). The direction of this regression is consistent with the first study and recent research (Ben-Zur et al. 2000; Furnham, 2001; Schou et al. 2004). The regression was performed using forward stepwise. This procedure enters one predictor in the regression at a time, thereby ensuring that the regression comprises the smallest number of predictors (Tabachnick & Fidell, 2001). This meant that the regressions identified the most important predictors of mental health by excluding measures of daydreaming and life orientation not providing additional prediction to those already in the regression.
The regression was performed with mental health as the dependent variable. The four measures of daydreaming were entered as independent variables: positive constructive daydreaming, guilt and fear of failure daydreaming, poor attentional control, and quality of daydreaming. The three measures of life orientation were also entered as independent variables: optimism, pessimism, and quality of life orientation. Four demographic characteristics were included as covariates: age, socio-economic status, marital status, and children. Demographic characteristics comprising multiple responses were recoded as dummy variables. Preliminary analysis of the research data had suggested that different regression models might apply to males and females. The regressions were, therefore, performed separately for each sex. The inclusion of independent variables in each regression was set at the significance level of .05.
The results of each regression are presented in two sets. The first set describes significant relationships between measures in the regression. These relationships can have a substantial impact on the final model, particularly if more than one independent variable has correlation coefficients with the dependent variable that are of a similar magnitude. Because little variance separates these independent variables if one enters the regression it is highly unlikely that other variable will also enter. Hence, despite not being in the final model, the effect of this latter variable in the prediction is not inconsequential. The second set of results presents the final regression model. It describes the direction of significant predictions including the contribution each predictor has to the variance in mental health scores.
There were no univariate outliers (using a case-wise plot of outliers outside + 3.0 standard deviations) or multivariate outliers (using Mahalanobis Distance) among
the measures entered in each regression. The number of cases to number of independent variables was above the recommended ratio of 5:1 (Hair et. al., 1995; Tabachnick & Fidell, 1996) for females (ratio = 9:1), but was marginally lower for males (ratio = 4:1). Within-cell scatter-plots (residual values against predicted values) included that interactions between dependent variables were linear and there were no serious indications of collinearity (using collinearity diagnostics). The results of each regression (correlations between variables) are available in Appendix D.
Regression Findings for Males
Lower mental health was significantly associated with lower attentional control,
r (39) = -.50, p ≤ .001, more guilt and fear of failure daydreaming, r (39) = -.41, p <
.001, and greater pessimism, r (39) = -.27, p < .05. Demographic characteristics were
not significantly associated with male mental health (p > .05). Poor attentional control entered the regression first as it had the highest correlation coefficient (r) with mental
health. Lower attentional control significantly predicted lower mental health, F (1, 37)
= 12.32, p ≤ .001, accounting for 25 percent of the variance in mental health scores. Once poor attentional control had entered the regression the independent contribution of guilt and fear of failure daydreaming was no longer significant. The regression confirmed that lower attentional control was significantly associated with more guilt and fear of failure daydreaming, r (39) = .43, p < .001. It is likely that guilt
and fear of failure daydreaming did not predict the mental health of males as much of the variance in this relationship was explained by the inclusion of lower attentional control. Hence, despite not being a direct predictor, more guilt and fear of failure daydreaming remains important to the prediction of lower male mental health.
Regression Findings for Females
The mental health of females was not significantly associated with measures of daydreaming or demographic characteristics (p > .05). Lower mental health was significantly associated with more pessimism, r (83) = -24, p ≤ .01, and lower quality
of life orientation, r (83) = .21, p < .05. Pessimism entered the regression first as it
had the highest correlation coefficient (r) with mental health. More pessimism
significantly predicted lower female mental health, F (1, 81) = 5.04, p < .05,
accounting for six percent of the variance in mental health scores. Quality of life orientation, scored by dividing scores for optimism by those of pessimism, was not a significant predictor of female mental health (once pessimism had entered the model).
Summary of Multivariate Findings
The mental health of males and females was predicted by measures of mental life, but not by demographic characteristics. Patterns of daydreaming, and not life orientation, were associated with the mental health of males. Specifically, lower male mental health was predicted by lower attentional control. More guilt and fear of failure daydreaming, which was associated with lower attentional control, was also associated with lower mental health, although the prediction was not significant. The mental health of males was not associated with positive constructive daydreaming. The mental health of females was associated with life orientation, but not patterns of daydreaming. Specially, lower female mental health was predicted by more pessimism. The mental health of females was not associated with optimism.
These findings suggest that more ‘negative thoughts’ had adverse implications for mental health. However, different negative thoughts were important to the mental health of males and females: more negative daydreams predicted lower mental health of males while more pessimism predicted lower mental health of females. The inability of males to maintain mental control was also associated with poorer mental health. More ‘positive thoughts’ did not improve male or female mental health.