Test-Retest Reliability of a Social Interaction Task

Accurate repeatability of experimental data is the basis of professional scientific research. In this study we analyzed three consecutive experiments: The subjects had to complete a questionnaire three times under similar conditions within a 2-3 week interval to ensure reproducibility of the original data from experiment to experiment, using the method of test-retest reliability. Absolute reliability was assessed by the standard error of measurement (SEM) and smallest real difference (SRD). The relative reliability was estimated by calculating the intraclass correlation coefficients (ICC3,1) (average measures) and the results demonstrate almost perfect agreement. The social interaction model was applied for organization of the experimental study. In this virtual model, the participants had to choose one of three types of reactions (i.e., attacking, avoiding, or friendly) to stimuli-facial expressions (i.e., angry, fearful, sad, neutral, and happy). The results show significant correlation between personal characteristics and social interactions. The results of the influence of such personal characteristics as agreeableness, collectivism, extraversion, neuroticism, and those shown on the Relational-interdependent Self-Construal Scale and State-Trait Anxiety Inventory are highly consistent with other researchers' data and common sense.