A Large Group Effect Doesn’t Mean They All Got Better: Discrepancies Between Statistical and Individual Analyses of Change in PSI Scores Among Parents
Poster Number
13
Format
Poster Presentation
Faculty Mentor Name
Scott Jensen
Faculty Mentor Department
Psychology
Abstract/Artist Statement
A 10 week behavioral parent training (BPT) course was conducted in which the Parenting Stress Index (PSI) was administered at weeks one, three, five, and 10. Data include consecutive referrals, though participants who dropped out of training prior to completion, had fewer than three weeks of PSI data, and/or had total PSI scores that were below the 75th percentile were not included in the analysis. For the qualifying participants (N = 17) group statistical analyses were compared to individual visual analysis. PSI total scores for each qualifying participant were plotted on individual graphs with the clinical cutoff (90) and sample mean (69) lines identified. Results from the group statistical analysis suggest a large statistically significant decrease from initial to final PSI scores, F(1,16) = 17.79, p = .001, .2 = .527. However, when the PSI scores were plotted individually for each of these qualifying participants, individual analysis suggests only 10 of the 17 (59%) showed a meaningful clinical change from initial to final measures of stress (defined as a decrease of 18 points pre to post, which is equivalent to a mean .5 decrease per item). The importance of performing individual analyses of data even in the face of large group differences is highlighted and discussed as the large group differences do not adequately capture the reality that over 40% of the participants did not make meaningful improvement. Trajectory and rate of change in PSI scores over a 10 week-BPT session are also discussed.
Location
DeRosa University Center, Ballroom
Start Date
20-4-2013 10:00 AM
End Date
20-4-2013 12:00 PM
A Large Group Effect Doesn’t Mean They All Got Better: Discrepancies Between Statistical and Individual Analyses of Change in PSI Scores Among Parents
DeRosa University Center, Ballroom
A 10 week behavioral parent training (BPT) course was conducted in which the Parenting Stress Index (PSI) was administered at weeks one, three, five, and 10. Data include consecutive referrals, though participants who dropped out of training prior to completion, had fewer than three weeks of PSI data, and/or had total PSI scores that were below the 75th percentile were not included in the analysis. For the qualifying participants (N = 17) group statistical analyses were compared to individual visual analysis. PSI total scores for each qualifying participant were plotted on individual graphs with the clinical cutoff (90) and sample mean (69) lines identified. Results from the group statistical analysis suggest a large statistically significant decrease from initial to final PSI scores, F(1,16) = 17.79, p = .001, .2 = .527. However, when the PSI scores were plotted individually for each of these qualifying participants, individual analysis suggests only 10 of the 17 (59%) showed a meaningful clinical change from initial to final measures of stress (defined as a decrease of 18 points pre to post, which is equivalent to a mean .5 decrease per item). The importance of performing individual analyses of data even in the face of large group differences is highlighted and discussed as the large group differences do not adequately capture the reality that over 40% of the participants did not make meaningful improvement. Trajectory and rate of change in PSI scores over a 10 week-BPT session are also discussed.