34

– This study explores the dynamics of departure time for the afternoon commute, which has received little attention both in theoretical and empirical work. Reported monthly and weekly departure time data, obtained from a survey of Brussels commuters, are used to study the impact of socio-economic, transportation and workrelated variables on the propensity of afternoon departure time changes. Binary and ordered probit models are used to estimate the impacts of these factors on two measures of departure time change propensity. The first one is the reported frequency of departure time changes in a month and the second measure is the time (in minutes) that a respondent left earlier or later than their normal time during the last five working days. In the first measure, the respondent selects his or her own threshold for reporting departure time change, whereas in the second measure the respondent merely provides the deviation from his or her normal departure time. The two measures together suggest three types of respondent groups: A group that perceives a wide-window (greater than or equal to 30 minutes) of regular departure times as normal, a group that perceives a narrow window of departure time as normal and change departure time occasionally and a third group that does not change their normal departure time. Statistical evidence indicates that work-related factors, particularly tolerant policy of the employer toward leaving work earlier, flextime and occupation type (scientific and executive professions) are associated with higher propensity of departure time changes. The implications of the findings are discussed.