Characteristically Short Reaction Times

The relative homogeneity of scores on the lexical decision task may be determined for each valence, for depressed, vs. nondepressed people. The simplest method of doing this would be to count the relatively short reaction times in each condition, for each person, on the lexical decision task. Relatively short, in the present thesis, is operationalized as reaction times occurring below 1.5 standard deviations from the mean of reaction times for the lexical decision task, specifically, 347ms. It is expected that the lexical decision task will reveal that even though the mean reaction time for depressed people is slightly longer for depressed people than nondepressed people with respect to negative v. neutral or positive words, that depressed people will have more unusually short reaction times to negative words than to neutral or positive words. In fact, depressed people displayed on average, 6.1$\%$ more short reaction times to positive and neutral words negative words, while nondepressed people displayed, on average, 4.3$\%$ more short reaction times to negative words than to positve and neutral words, D=10.3$\%$, F(1,74)=6.83, p=0.011. This finding was not robust over the other duration conditions though, as a split-plot multivariate ANOVA using the number of short reaction times, as the dependent variable, and valence, duration, and depression as independent variables did not reveal a statistically significant depression by valence by duration interaction, Hotelling's F(4,71)=2.09, p=.091, t#tex2html_wrap_inline1514#=.105 or depression by valence interaction, Hotelling's F(2,73)=2.33, p=.104, t#tex2html_wrap_inline1516#=.060.

To illustrate that the lack of the expected relationship was not an artifact of the chosen threshold, a trend analysis was performed in the form of a threshold x depression multivariate split plot ANOVA. Fifteen separate levels of threshold were examined, ranging from 150ms to 500ms in 25ms intervals. No statistically significant linear, quadratic, cubic, or quartic polynomial fits were observed for deviation contrasts on depression, at each of the thresholds.