Abstract: This paper aims to evaluate which is more accurate among three commonly used time measurement methods in databases for household survey: “general time use reporting”, “yesterday time use reporting”, and “one day’s time diary”. Husband and wife in the same family are invited to write time diary in seven consecutive days and simultaneously answer two other sets of questionnaires concerning family time allocation and the couple’s time use. Each respondent’s housework time or leisure time measured by above three methods are compared against the “benchmark time” which is the respondent’s average “housework time” or “leisure time” of each day on five working days. It is found that “general time use reporting” has largest estimation bias, and “one-day time diary” has least bias among the three methods. We further investigate the husband’s and wife’s perceptions of time use by themselves and their partners, and found that each individual’s time use estimation of his or her partner is not mainly based on “facts” or accurate “observation” and instead on partner’s perception of his/her own time use.

Key words: time measurement, housework time allocation, time diary, estimation bias