Momentary time sampling with time series data

The accuracy of momentary time sampling was investigated as a function of (a) the emitted duration of instances of behaviour, and (b) the rate (probability) of emission of behaviour. A computer simulation randomly generated pseudo‐behaviours of durations 1, 10 and 20 notional seconds (constant within each condition) at probabilities of onset of 1/6 and 1/60. Twenty runs, each of one notional hour, were generated for each combination of duration and rate. Simulated momentary time sampling of each run was carried out at intervals of 15, 30 and 60 notional seconds. The results indicated that the accuracy of momentary time sampling was not simply a function of the interval size, but that the emitted duration and rate of behaviour must be considered when evaluating the technique. The conclusion of Brulle & Repp (1984) that momentary time sampling at intervals of 10, 20 and 30s accurately estimates duration of behaviour was modified to take account of these results. A more cautious use of the technique is advocated and it is suggested that, in general, the use of momentary time sampling at intervals of 30s or less is justifiable only when the record indicates that behaviour occurred during at least 25 per cent of the observational period

Biology · Cognition · Duration (music) · Function (biology) · Interval (graph theory) · Notional amount · Physics · Sampling (signal processing) · Statistics · Telecommunications · Time perception · Behavioral and Psychological Studies · Computer Science · Mathematics · Psychology