**5. Conclusions**

This research empirically examined whether the uncertain geographic context problem (UGCoP) and the neighborhood effect averaging problem (NEAP) are encountered in research on individual exposures to traffic congestion. We used crowdsourced real-time traffic congestion data and activity-travel data of 250 individuals in Los Angeles to compare two assessments of individual exposure to traffic congestions: one that only considers commuting trips and one that also considers individuals' non-commuting trips in addition to considering commuting trips (thus taking people's activity-travel patterns into account).

First, the results indicated that spatiotemporal variations in traffic congestion intensity exist in the study area, which calls for the consideration of individuals' activity-travel patterns when assessing their exposures to traffic congestion in future research.

Second, the paired sample *t*-test results revealed that individual exposures to traffic congestion are significantly underestimated when individuals' activity-travel patterns are ignored. Further, the results highlighted that the probability distribution function of individual exposures is heavily skewed but tends to converge to its average value when individuals' activity-travel patterns are considered, which is a manifestation of the neighborhood effect averaging problem (NEAP). These results indicated that both the UGCoP and the NEAP are critical methodological issues in traffic congestion and health studies.

Lastly, we presented space-time visualizations of the traffic congestion exposures of 3 individuals from the same household. The results illustrated that since individuals have idiosyncratic activity-travel patterns, their exposures to traffic congestion vary significantly even if they live at the same residential location. These results corroborate the claim in previous studies that using residence-based methods or data aggregated over areas (e.g., census tracts) may lead to erroneous assessments of individual exposures to traffic congestion or other environmental influences [31].

The results of our research imply that epidemiological studies should pay more attention to individuals' activity-travel patterns when assessing people's environmental exposures. As the results of this study illustrated, ignoring individuals' daily mobility (i.e., activity-travel patterns) may result in erroneous assessments of their exposures to traffic congestion. Eventually, using inaccurately estimated traffic congestion exposures may exacerbate inferential errors or lead to significantly misleading results when investigating the effects of traffic congestion exposures on human health. In addition, the results imply that researchers who study environmental exposures should focus more on individual-level analysis, since this study shows that people who live in the same residential location may have different traffic congestion exposures due to their distinct activity-travel patterns. This implies that using residence-based methods or data aggregated over areas (e.g., census tracts), which are popular approaches in previous studies, may lead to critical methodological issues.

Although this study significantly advances our knowledge about two critical methodological issues (i.e., the UGCoP and the NEAP) in traffic congestion and health studies, it has several limitations that should be addressed in further studies. First, we presumed that individuals used the shortest path (in terms of travel time) to travel from one location to another, which may not fully capture their true activity-travel patterns. Although this assumption is reasonable, people may not necessarily use the shortest path. One possible solution to this issue may be to employ a space-time prism [57], as illustrated in Figure 11. A space-time prism consists of points that individuals may possibly visit given their spatiotemporal constraints [50,58]. Considering possible alternative routes in a space-time prism may help researchers comprehensively assess traffic congestion exposures.

**Figure 11.** An example of a space-time prism.

Second, further studies need to consider the subjective aspects of individual exposures to traffic congestion. Previous studies revealed that exposures to traffic congestion may go through subjective perception filters [12,16,59]. This indicates that although people are exposed to the same level of objective traffic congestion (e.g., 20 min in traffic congestion), the effects of the objective traffic congestion on health may follow different mechanisms for each individual. However, due to the limitations of the survey data used in this study (which did not collect or provide any data on participants' perceptions), we were not able to address this issue. One possible solution may be to integrate in-depth interviews about subjective factors with activity-travel surveys [25,60]. By combining the subjective experiences of traffic congestion with objective measures, future research may further advance our knowledge of the health impacts of exposures to traffic congestion.

Third, more scrutiny is required to unveil the temporal dimension of the effects of traffic congestion exposures on health [24]. We computed total traffic congestion exposure in minutes because we presumed that cumulative exposure may influence health. However, the effects of traffic congestion exposures on health may show "time-lagged response" [24], which means that it may have health effects afterward. Moreover, not only the duration of traffic congestion but also the variability in driving time may negatively affect health. Since several epidemiological studies reported such evidence [15,61], more attention is required to clearly understand the temporal aspect of the effects of traffic congestion on health. One possible solution is to utilize real-time global positioning system (GPS) technology to gain clearer pictures and more detailed understanding of these temporal effects of traffic congestion on health [62,63].

**Author Contributions:** The authors confirm contribution to the paper as follows: study conception and design: J.K., M.-P.K.; data collection and computing: J.K.; analysis and interpretation of results: J.K., M.-P.K.; draft manuscript preparation: J.K., M.-P.K.

**Funding:** Junghwan Kim was supported by a Foster Fellowship. Mei-Po Kwan was supported by grants from the U.S. National Science Foundation (BCS-1832465) and the National Natural Science Foundation of China (41529101).

**Acknowledgments:** Junghwan Kim would like to thank Brennan Borlaug and the Transportation Secure Data Center (TSDC) for their kind support on data access. The authors thank the three anonymous reviewers for their helpful comments.

**Conflicts of Interest:** The authors declare no conflict of interest.
