Findings Web Probing:
There was no difference in non-response between question versions, with exactly one respondent not answering in either format. When examining both the responses to the survey question and the following probing question, there were no implausible values (in either question version). In the open-ended numeric format, 8% (
n = 61) of the respondents answered with 0 minutes of commuting, indicating that these respondents worked from home. A similar distinction between respondents with a short versus no commute could not be made based on the predefined response options in the closed question. The responses to the open-ended numeric question showed that respondents tended to provide rounded responses (i.e., 10, 15, 20 or 30 minutes) rather than “exact” times (i.e., 24 minutes).
The response distributions between the two question formats differed significantly when the open-ended numeric responses were recoded to exactly depict the predefined closed responses (see Table 27; χ2(7,790) = 19.143, p = .008).
However, response distributions were nearly identical when the open-ended answers were recoded to include the rounded minutes (see Table 28; χ2(7,790) = 5.643, p = .570). This indicates that respondents in the closed response format who were on the verge between two responses (i.e., because their commute is approximately 30 minutes) tended to choose the lower value.
There were no significant differences in response behaviour between countries. However, self-employed respondents were significantly more likely to work from home and report the value “0 minutes” in the open-ended numeric format or “less than 15 minutes” in the closed question version than respondents who were employed or in atypical working situations (χ2(16,792) = 55.549, p < .001).
How did respondents arrive at their answer?
Most respondents (80%,
n = 633) gave substantive responses to the probe on how they arrived at their answers. There were no significant differences between the shares of substantive probe responses depending on the question version. However, respondents in atypical working situations were more likely to give non-substantive responses (29%,
n = 84) than respondents who were employed (16%,
n = 45) or self-employed (13%,
n = 30) (χ2(2,792) = 24.153, p < .001).
Among the substantive responses, almost half of the respondents (42%,
n = 267) reported a clear
strategy how they arrived at the number of minutes they spent travelling to and from work each day. These respondents either reported a calculation, mainly by summing up the time spent to get to work and back (i.e.,
“I need 40 minutes in one direction from door to door. I took this value times two“) or based their answer on the distance between their home and workplace (i.e.,
“I live 600 meters from my work”). A second group of respondents (29%,
n = 182) gave short responses based on their everyday
experiences,
such as
“Because that is how long I drive” or
“That is my experience”. In 13% of substantive responses (
n = 83), respondents explained that the time they spent travelling
varied strongly,
depending on traffic, the vehicle(s) they used, or where they had to travel to for work. In some cases, respondents travelled to varying client premises. In other cases, respondents worked mainly from home, but when they travelled, travelled far (i.e., a performing artist wrote:
“We do events all over the country, so we travel a lot when we have a show. The rest of the time we work from home”). Many of these respondents reported an estimated average (i.e.,
“I work in many locations, so this is an average travel time” or
“In extreme cases, I travel more than three hours to markets and fairs. But most of the time I work from home or in the warehouse not far from me, so the average time is rather low”). Finally, 16% of respondents (
n = 101) reported that they did not travel to work at all because they
work from home. Respondents who received the open-ended numeric format all inserted the value “0” as their travel time. Respondents who received the question version with the closed response options chose “less than 15 minutes”, but often commented that they were missing a response option to express that they do not commute (e.g.,
“I work from home, so the answer is 'no time is spent travelling to work' but that was not an option”).
Inconsistency in reporting occurred when respondents seldom travelled to work or had travelled to work before the beginning of the COVID pandemic. For instance, in the case of the performing artist mentioned above, the respondent reported a travelling time of “between 3 hours and under 4 hours”, indicating the average time to a performance. Other respondents who mainly worked from home calculated an average. Among respondents who were working from home due to the pandemic at the time of the web probing study, some reported their current situation, while others reported their regular commute before the start of the pandemic. For instance, one respondent who inserted “0” as the number of minutes spent commuting explained:
“I have been working from home since the start of COVID, so I do not travel. My actual office is a one-hour commute - if I ever do it again.” In contrast, another respondent answered “between 1 ½ and under 2 hours”, clearly only referring to the situation before the pandemic:
“I non-COVID times, my office is 40 miles away and my typical commute is between 45 and 60 minutes each way.”
There were no significant differences in probe responses between question versions, indicating that the cognitive strategies used by respondents to answer the survey question did not differ by question format. However, there were significant differences in probe responses based on the respondent’ working situation (χ2(6,633) = 69.436, p < .001). Employed respondents were most likely to base their survey response on a clear calculation of their travel time and distance, while respondents in atypical working situations were more likely to report strongly varying travel times, and self-employed respondents were most likely to work from home (see Table 29).
Summary:
- Either question format (open-ended numeric or closed) is suitable to capture the time spent commuting.
- Respondents who seldom travelled for work sometimes reported an average across all days, and sometimes only reported their average travel times on days that they travelled.
