Premium
Time Sampling: A Methodological Critique
Author(s) -
Mann Janet,
Have Thomas,
Plunkett James W.,
Meisels Samuel J.
Publication year - 1991
Publication title -
child development
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.103
H-Index - 257
eISSN - 1467-8624
pISSN - 0009-3920
DOI - 10.1111/j.1467-8624.1991.tb01527.x
Subject(s) - sampling (signal processing) , statistics , psychology , sampling error , observational study , stratified sampling , observational error , mathematics , computer science , filter (signal processing) , computer vision
Numerous developmental studies of social behavior employ an observational technique called time sampling. This technique is used estimate the frequencies and proportions of time that actual behaviors occur and to measure group or individual difference. By contrasting time sampling to continuous sampling, we demonstrate that (a) time sampling is inaccurate for estimating actual durations or frequencies of behavior, and (b) resulting individual and/or group differences can change depending on whether time sampling or continuous sampling is used. Error rates are particularly high when the total bout lengths of the actual behaviors are short (as in mother‐infant interactions) and/or when interval length is long. Because the majority of mother and infant behaviors are short in bout length (less than 60 sec), when percent error is the greatest, we suggest that alternative sampling methods that have proven validity be used in future studies (e.g., scan, point, or continuous sampling).