Systematic error in behavioural measurement: Comparing results from interview and time budget studies