Misuse of Linear Models: Understanding Community Linkages in Family Adaptation to Unexpected Death

In research, the use of linear additive methods is expedient when prediction is the goal; however, when understanding is the goal, an exploration of non-linear multiplicative procedures is more appropriate. To compare the results of linear and non-linear models, data obtained from a survey of 35 bereaved mothers were compared using scattergrams, correlational analyses, analysis of variance, and factorial plots. Independent variables in the study included length of time since the death, the mother's level of education, her perception of family religiosity, and family stress. The dependent variable was maternal perception of family adaptation to the unexpected death of a child. Results showed that analysis based only upon correlation coefficents was misleading since this approach did not account for the disordinal interactions indicated by the Anova. These findings suggest that research and theory on family adaptation to crisis, based only on additive linear information, could misdirect family practitioners and counselors. Researchers must be careful to define their goals, conduct exploratory data analyses, and use non-linear models when theory is not strong, in order to derive the most beneficial and "true" results from their data. (Author/BL)