Factors Influencing Full-day, Part-day and Overtime Telecommuting: An Investigation of Northern California Workers

Telecommuting has long been viewed as a prospective way to reduce commute and work-related travel, and considerable research related to telecommuting has been conducted in previous years. This study contributes to the literature on the adoption of telecommuting in two ways. First, it divides telecommuting into three different types from the perspective of their influences on travel: full day telecommuting (full day TC), part day telecommuting (part day TC), and overtime telecommuting (overtime TC). Secondly, this study examines the effects of home and workplace neighborhood built environment (BE) variables on telecommuting, along with those of socioeconomic and attitudinal variables. Furthermore, when investigating the impacts of the BE on telecommuting adoption, we take into account the possibility that those impacts will differ between pro-low density and pro-high density people (using non-centered factor scores to identify each case as pro-low or pro-high density). Specifically, we hypothesize that for pro-low density people, lower densities and a more residential character at the home location will increasethe propensity to telecommute, while those same traits at the worklocation will decrease the propensity, and conversely for pro-high density people.;We develop separate binary logit (BL) models of the three types of telecommuting, as well as a multivariate probit (MVP) model of the joint choice of the three types. We found that both sets of models have reasonably strong goodness of fit and predict the marginal probabilities very well, but the MVP model does a much better job than the BL models in predicting the joint probabilities. The BL models and MVP model share a number of significant variables, but there are also variables that are only significant in one set of models or the other. Household income, being a clerical worker, and commute distance are significant in all full day TC, part day TC, and overtime TC models. Variables interacted with the attitude toward density also enter into the models, all having signs consistent with our hypotheses: distance to maintenance activities at the workplace for pro-high density people and population density at the workplace for pro-low density people appear in the full day TC model; share of residential area within a 1/2 mile radius of workplace for pro-high density people appears in the part day TC model; and share of residential area within a 1/2 mile radius of home for pro-high density people appears in the overtime TC model. In general, a given BE variable is significant to one segment (either pro-low or pro-high density) but not the other. In one instance, however, the same BE variable influences pro-low density and pro-high density people oppositely; specifically, employment density in the home area influences pro-high density people positively and pro-low density people negatively in the choice of overtime TC.