Abstract:
Suen and Ary's (1986a) statistical model to specify time sampling intervals that should produce accurate estimates of response duration and frequency was tested. Intervals with low probability of error (appropriate) were compared to those with high probability of error (inappropriate), as determined by the Suen and Ary method. Discrepancy (from real-time data) indices were calculated for partial-interval, whole-interval, and momentary time-sampling methods using data from three social behaviors of two preschool children in 30 min observation sessions. In general, for both duration and frequency measures (for both children), all three time sampling methods produced statistically reliable error in comparison to real-time recording, However, reliably greater error was found with inappropriate than with appropriate intervals, in accord with Suen and Ary's model.
