A two-part study contrasted the utility of free-response and checklist methodologies for ascertaining ethnic and gender stereotypes. Descriptions of data collection, organization, and cluster and entropy analyses are provided. Results indicate that important differences emerge between data resulting from free-response methodology and those obtained with traditionally employed adjective checklists. These differences include the generation of a large percentage of physical descriptors and within-ethnic-group gender differences in stereotype content. A major finding is the generation of a large number of distinct responses coupled with low-frequency use of any particular response. Study 2 specifically examined whether free-response data are more schematic than checklist data. Results indicate that free-response data have a greater dependency and may thus be indicative of schematic response. This schematic response may, in turn, indicate more automatic processing than is evident with data from checklist methodologies.

Entering seventh grade, most children believe that smoking endangers their long-term health. Nevertheless, three sources of social pressure were found to often have an overriding influence t o begin smoking: peers, models of smoking parents, and media. Interventions were developed t o train children t o cope with these pressures. Also stressed were immediate rather than long-term physiological effects of smoking. A dependent measure was utilized that generates increased validity of self-reports of smoking. A 10-week investigation was completed with 750 male and female students entering the seventh grade. Rates of onset of smoking in treatment groups were significantly lower than in a pretest-single posttest control group.

In the cross-lagged panel correlation technique for choosing among causal hypotheses, four hypotheses are in competition rather than the two previously assumed. A significant inequality chooses only one confounded pair of hypotheses over the other pair. Where the synchronous cross-correlations are zero, or where appropriate reliabilities are available for all four measures, the confounded hypotheses may be separably examinable.

No abstract

No abstract