COWEN & BEIER investigated the validity of perceptual defense using a design that attempted to rule out the potential artifacts in the McGinnies study. C&B constructed booklets, each containing a word (the same ones used by McGinnies). Words were very blurry and unreadable on the first few pages, but then became more legible with each page turn. Experimenters (E) recorded how much time and how many page turns were required for correct recognition.

In study 1, subjects took more page turns and a longer time to identify the “critical” words

In study 2. Subjects were alerted to the presence of threat. Prior to the page turning task Es and participants read aloud a list of 50 words, 10 of which were the stimuli used later. This (hopefully) made the words of similar familiarity to all subjects (adjusting for word frequency problems) and made it all right to say the words out loud (adjusting for response suppression). The results of study 2 were not significant by t-test; but were significant by other statistics that will be explained in class.

A comparison of studies 1 and 2 showed that when subjects were alerted to expect threat the threshold differences between recognition of threat and non-threat decreased significantly. The data (Mean and variance of total number of page turns) were:

Threat--alerted, M=30.8, Var=3.5

Neutral--alerted, M=27.3, Var=1.98

Threat--not alerted, M=36.5, Var=.56

Neutral--not alerted, M=24.9, Var=1.8

The variance of the Threat--alerted (3.5) was significantly larger than the variance of Threat--not alerted (.56), leading the authors to conclude that alerting subjects to threat made them super vigilant and perhaps identifying “critical” threatening words on the first page. This super vigilance presumably increased the variance in the Threat--alerted condition. However, there is the possibility of a problem in the data. And as detectives, it is you job to discover a reason why the means and variances above look “funny”.

Hints: Each mean was calculated from the total number of page turns in recognizing 5 words in booklets that contained 8 pages. Hence possible scores could range from 5 (recognizing all words on the first page) to 40 (recognizing all words on the last page). The “funny” group is the Threat--not alerted group. What could be going on? Happy sleuthing!!!