A variety of statistical methods are employed in analyzing data collected in research on adult development—chi-squares, correlation coefficients, much more proficient at it than I. An odor, a sound, or a familiar object could resurrect a whole host of detailed memories of things that occurred in his childhood and youth. He also found that recall could be facilitate by isolating himself from other people and things, so the only voices he was aware of were internal ones. As it was with Proust, most of our memories are still available to us if, in solitude, we simply relax and focus on some object or person associated with a particular experience. In this way we can recapture, or shall I say "reconstruct," the past.
Once we have remembered, we can reexamine particular events and even experiment with them: What might have been the result if this had happened instead of that, if I had responded in this way rather than that? Granted, these are only hypothetical experiments, or "thought experiments" as Einstein labeled them, and the outcomes may not actually have been as we imagine. Still, like Einstein, we can exercise our critical faculties in testing the hypothetical outcomes for plausibility and obtain the opinions of other people. In any case, by speculating on other possible outcomes we become free of the fatalism of "whatever will be will be" and that nothing we could have done would have prevented what actually happened.
Life is full of mystery and uncertainty, but this is just what makes it so incredibly interesting. Obtaining insight into the mystery of what we are and why we turned out this way requires attending not only to how other people react to us but also to what our memories and inner voices tell us about ourselves. "Know thyself," advised the Oracle at Delphi; the road to self-knowledge lies within you. We need not bewail with the poet that "For all sad words of tongue or pen, the saddest are these 'It might have been."' True, preoccupation with the past and things we cannot change may be depressing. However, recollection and self-examination can be beneficial if they teach us how to make our personal present and future better and motivate us to help others become aware of alternative ways of looking at things and the multiple options in their own lives.
analyses of variance, multiple regression analyses, and so on. Both chi-square and correlation coefficients are measures of relationship between variables, whereas analysis of variance is used to determine the significance of differences between mean scores on a dependent variable obtained at various levels of one or more independent variables. Analysis of variance actually consists of a family of designs and procedures by which a series of F ratios can be computed to determine (1) between-subject and within-subject effects, (2) main effects and interactions, (3) univariate (one dependent variable) or multivariate (two or more dependent variables) effects of treatment (independent variable) conditions, as well as controlling for one or more concomitant variables (covariates).
Because most developmental studies involve several variables, multivari-ate procedures such as multivariate analysis of variance, multiple regression
Abstracts of Illustrative Studies Using Correlational and Multiple Regression Analyses
Wilbur, J., Dan, A., Hedricks, C., & Holm, K. (1990). The relationship among menopausal status, menopausal symptoms, and physical activity in midlife women. Family and Community Health, 13 (3), 67-78.
375 women (aged 34-62 yrs.) completed questionnaires on demographics, health, dietary calcium intake, and physical activity. Ss represented 4 menopausal status groups: premenopausal, perimenopausal, postmenopausal, and hysterectomy. Vasomotor and general symptoms were significantly related to menopausal status. Significant negative correlations were found between leisure activity and nervous and general symptoms, and between aerobic fitness and nervous and general symptoms. There was a significant positive correlation between occupational activity and general health symptoms. (Reprinted with permission of the American Psychological Association, publisher of Psychological Abstracts and the PsycLIT database. All rights reserved.)
Julian, T., McKenry, P. C., &McKelvey, M. W. (1992). Components ofmen's well-being at mid-life. Issues in Mental Health Nursing, 13 (4), 285-299.
Examined correlates of psychological well-being for 75 middle-aged professional men. Three sets of predictor variables (interpersonal family factors, role adjustment, and extrafamilial interpersonal factors) were hierarchically entered into a multiple regression equation. Well-being was influenced by interpersonal family factors. Role adjustment and extrafamilial interpersonal factors did not account for a significant increase in variance. The best univariate predictors of men's well-being at midlife were perceived closeness to child, perceived closeness to wife, adjustment to the husband role, and number of close friends. (Reprinted with permission of the American Psychological Association, publisher of Psychological Abstracts and the PsycLIT database. All rights reserved.)
analysis, discriminant analysis, and canonical correlation are often applied. The first abstract in Report 1-4 summarizes a study employing correlational analysis, and the second abstract is of a study using multiple regression analysis. A multiple regression analysis assigns different numerical weights to several independent variables in predicting a single dependent variable. In the multiple regression analysis described in Report 1-4, psychological well-being is the dependent variable predicted by the independent variables of interpersonal family factors, role adjustment, and extrafamilial interpersonal factors. Discriminant analysis is similar to multiple regression except that, rather that predicting scores on a dependent variable from scores on several independent variables, discriminant analysis differentiates between various criterion (dependent variable) groups on the basis of their scores on weighted combinations of independent variables. Finally, canonical correlation analysis is an extension of multiple regression in which there are several dependent variables as well as several independent variables. All of the statistical procedures listed here can be carried out by the Statistical Package for the Social Sciences (SPSS/PC+) or other packages of computer programs in statistical methods.
Was this article helpful?