next up previous contents
Next: Induced Depression Up: Naturally Occurring Depression Previous: Study characteristics

Differential response latencies

The studies presented a great deal of variability in calculated effect sizes as shown in Appendix B. The greatest variability was observed as a function of study. For example, Matthews and Southall's (1991) study had opposite effect sizes for the different SD conditions for the negative vs. neutral contrast, each nearly twice as large in magnitude as the effect sizes obtained in any other study. This variation could be due to the SD's chosen, the high number of stimuli investigated, or the low n used in Matthews and Southall's (1991) study. Still, homogeneity tests revealed no statistically significant differences in effect sizes averaged within any examined contrast. For negative vs. neutral words, Q=5.16, df=2, p>0.05. For positive vs. neutral words, Q=1.77, df=1, p>0.05. For positive vs. negative words, Q=.43, df=2, p>0.05.

Mean weighted unstandardized effect sizes for the negative vs. neutral condition revealed that depressed people were statistically significantly delayed in responding to negative words by approximately D.=27.44ms, V.= 69.29, D.lb=11.13, D.ub=43.76. This delay is very close to the delay observed by Macleod et al.'s (1987) study which was not included in the meta-analysis (D=29.00) yielding further support for the statistical conclusion validity of this finding. The same delay is reflected when standardized effect sizes are used, d.=.41, v.= 047, d.lb=.00, d.ub=.82. Because the 95$\%$ confidence intervals around the unstandardized mean effect size does not overlap with that for nondepressed people it may be said that depressed people show statistically significantly more of a delay with respect to negative words vs. neutral words than do nondepressed people. As the 95$\%$ confidence intervals do overlap for the standardized effect sizes, the wide variability among standard deviations between studies may cast doubt on this finding.

The relative facilitation with respect to positive vs. neutral words found in nondepressed people is present to a greatly reduced degree in depressed people, D.=-3.79ms, V.= 209.07, D.lb=-32.14, D.ub=24.56. This value is neither statistically significant, nor statistically different from the corresponding contrast for nondepressed people. The small magnitude of the effect with respect to its variance is revealed by the small standardized effect size, d.=.022ms, v.= .055, d.lb=-.44, d.ub=.48. Similarly, depressed people were only very slightly, and not statistically significantly delayed with respect to negative vs. positive words, D.=-11.02ms, V.= 211.87, D.lb=-39.55, D.ub=17.51, d.=-.08, v.=.04, d.lb=-.49, d.ub=.33.


next up previous contents
Next: Induced Depression Up: Naturally Occurring Depression Previous: Study characteristics
Greg Siegle
1999-11-15