Factors Influencing Teleworking Productivity – a Natural Experiment during the COVID-19 Pandemic

Of 2174 surveyed adults who were teleworking following the implementation of a Covid-19 work-from-home policy, 23.8% reported an increase in productivity, 37.6% no change, and 38.6% a decrease in productivity compared to working at their prior workplace. After controlling for feelings of depression and anxiety likely caused by pandemic-related circumstances, the socioeconomic characteristics associated with no change or an increase in productivity after shifting to teleworking included being older; not employed in higher education; having lower education attainment; and not living with children. Respondents with longer commute trips in single-occupancy vehicles prior to teleworking were more likely to be more productive but those with longer commute by walking were not. Lifestyle changes associated with increased productivity included better sleep quality, spending less time on social media, but more time on personal hobbies.


Study and Survey Design
We conducted the survey between April and June 2020 in the four counties of the central Puget Sound (Seattle) region, Washington, and asked participants about changes in their work productivity, daily routines, and mental wellbeing since WFH during Covid-19. We used convenience sampling, with a target population of adults older than 18 living in the region's four counties. The survey was managed online and distributed through professional email lists of public agencies, non-government organizations, universities and colleges, as well as other public community groups. There was no financial incentive for participation. We obtained responses from 2174 adults from 83% of the ZIP Codes in the four counties, who had shifted from working away to WFH since Covid-19. Compared to central Puget Sound region general population, 69% of our participants had a household income higher than the region's median; 67% were female (50% in the region) and 52% had graduate degrees or above (32% in the region) (Table 1). Detailed information on the survey was documented elsewhere (Puget Sound Regional Council 2020).
2. The lifestyle changes that helped the WFH population continue to be productive while teleworking Factors Influencing Teleworking Productivity -a Natural Experiment during the COVID-19 Pandemic Note 1 -significant in univariate models.

Variables
The unit of analysis was the participant. The outcome of interest was selfreported change in productivity since WFH. Participants were given three options: no change, a decrease or an increase in productivity. Productivity was treated as an ordinal variable with decrease in productivity as the reference.

Statistical Analysis
Partial proportional odds (PPO) models, also referred to as generalized ordered logit models were used in the analysis, which relax the parallel line assumption for variables having different relations with each pair of ordinal outcome groups (Williams 2016). Brant test was used to screen the variables that did not meet the assumption. In PPO models, for variables that met the parallel assumption, one set of coefficients were estimated; while for others, two coefficients corresponding to decrease vs. no change and no change vs. increase were estimated separately. We first tested univariate models for all hypothesized predictors. A full PPO model was then estimated with all the predictors significant in univariate models. A final, reduced model included only the variables that had remained significant in the full model (Table 2, Figure 1). VIF scores showed no issue with collinearity. For each model, we used listwise deletion where only observations with complete information were included.

Change in productivity since teleworking
Of the 2147 survey respondents, 38% reported being less productive since shifting to WFH; 37.6% reported having no change; and 23.8% reported being more productive than prior to teleworking (   Figure 1. Odds ratio and 95% confidence interval of factors associated with no change or increase in productivity which respondents worked under between April and June 2020 mandated telecommuting every workday, which is a more restricted condition than most WFH where workers may have more choices regarding the number of days and times of teleworking. All these conditions may affect the productivity of WFH.

Domains of influence on teleworking productivity ses
After controlling for mental status, people aged 30 and above had a higher probability of reporting no change or an increase in productivity after shifting to teleworking. Respondents not living with children, and therefore less likely to experience unexpected interruptions, were more likely to report no change or increase in productivity. Those in higher education, including students, faculty, and staff, tended to report being less productive. Those with higher educational attainment (having a graduate degree and above) also tended to report being less productive. This is likely due to the nature of academic work (e.g. teaching and knowledge production), which often relies on specific equipment and facilities not available at home, as well as on in-person communications (Carlino and Kerr 2014;McFadyen and Albert A. Cannella 2016). The population segments associated with no change or increase in productivity are potential targets for future long-term teleworking arrangements. Conversely, those who reported a decrease in productivity after shifting to WFH could benefit from additional support from family, employer, and community if continuing to WHF.

previous commute mode and duration
Participants with longer Single Occupant Vehicle (SOV) trips before having the WFH had a higher probability of reporting no change or an increase in productivity after shifting to teleworking. In contrast, those with longer walking trips, had a higher probability of reporting a decrease in productivity.

lifestyle changes
Sleep quality has long been associated with work productivity and quality of life (Kucharczyk, Morgan, and Hall 2012). We found similar positive relationship between sleep quality and reported productivity after shifting to teleworking. More time spent on social media was linked to a decrease in productivity. This was consistent with the negative effects of the distractions, physical discomfort, and negative emotions associated with social media use as reported in previous studies of workplace and office settings (Priyadarshini et al. 2020;Vithayathil, Dadgar, and Osiri 2020). On the other hand, more time spent on personal hobbies, was linked to maintaining the same level of productivity, or to increasing productivity. This suggested that engaging in nonwork related activities helped compensate for the demands of continuous work (Eschleman et al. 2014).

ACKNOWLEDGMENTS
This work has been funded by the US Department of Transportation's University Transportation Center program through the Pacific Northwest Regional University Transportation Center (PacTrans) and Puget Sound Regional Council (PSRC). The authors would like to thank PacTrans and PSRC for their support. Views expressed in the paper do not represent those of the sponsors and the authors are responsible for all errors that may exist. Factors Influencing Teleworking Productivity -a Natural Experiment during the COVID-19 Pandemic Findings