Comparison of global positioning system (GPS) tracking and parent-report diaries to characterize children's time–location patterns

Abstract

Respondent error, low resolution, and study participant burden are known limitations of diary timelines used in exposure studies such as the National Human Exposure Assessment Survey (NHEXAS). Recent advances in global positioning system (GPS) technology have produced tracking devices sufficiently portable, functional and affordable to utilize in exposure assessment science. In this study, a differentially corrected GPS (dGPS) tracking device was compared to the NHEXAS diary timeline. The study also explored how GPS can be used to evaluate and improve such diary timelines by determining which location categories and which respondents are least likely to record ''correct'' time-location responses. A total of 31 children ages 3-5 years old wore a dGPS device for all waking hours on a weekend day while their parents completed the NHEXAS diary timeline to document the child's time-location pattern. Parents misclassified child time-location approximately 48% of the time using the NHEXAS timeline in comparison to dGPS. Overall concordance between methods was marginal (k ¼ 0.33-0.35). The dGPS device found that on average, children spent 76% of the 24-h study period in the home. The diary underestimated time the child spent in the home by 17%, while overestimating time spent inside other locations, outside at home, outside in other locations, and time spent in transit. Diary data for time spent outside at home and time in transit had the lowest response concordance with dGPS. The diaries of stay-at-home mothers and mothers working unskilled labor jobs had lower concordance with dGPS than did those of the other participants. The ability of dGPS tracking to collect continuous rather than categorical (ordinal) data was also demonstrated. It is concluded that automated GPS tracking measurements can improve the quality and collection efficiency of time-location data in exposure assessment studies, albeit for small cohorts.

Quantifying human exposure to air pollution—Moving from static monitoring to spatio-temporally resolved personal exposure assessment

“…Bias is also a problem when people need assistance to complete the forms (Freeman et al, 1999). Another difficulty can be the language barrier for non-native speakers (Elgethun et al, 2007). In general, time requirements of the survey process need to be kept low in the interest of the participants, encouraging them to fill in the form without getting bored and to reduce recall bias (Crosbie, 2006;Freeman et al, 1999;Freeman and Saenz de Tejada, 2002).…”

“…However, TADs and questionnaires require literacy, a sense of time and a certain degree of commitment. It is useful to train the participants beforehand in how to correctly use the forms (Elgethun et al, 2007;Freeman et al, 1999;Freeman and Saenz de Tejada, 2002) especially when they are presented on electronic devices.…”

“…did not have large time and/or space gaps between logs (Nethery et al, 2008a). The feasibility of using GPS receivers for tracking individuals in their everyday environments has also been studied by Adams et al (2009), (Elgethun et al, 2003;Elgethun et al, 2007), Phillips et al (2001) and Rainham et al (2008) who were looking into methodology, potential and limitations when using GPS sensors for exposure assessment and the validation of TADs. The main problem when using GPS devices is that the satellite signal is often not strong enough for use inside buildings or near certain materials such as steel-reinforced constructions, body panels and other electrically conductive material (Phillips et al, 2001).…”

The German Environmental Survey for Children (GerES IV): Reference values and distributions for time-location patterns of German children

“…A study on 31 US children comparing GPS tracking information to parental time diary data suggests that parents may overestimate the time spent at home while underestimating the 470 time spent in other locations (Elgethun et al 2007). However, directly measuring timelocation patterns with GPS tracking participants or possibly burdensome multi-day timelocation diaries is generally of low feasibility in large-scale epidemiological studies.…”

Analyses of school commuting data for exposure modeling purposes

“…They essentially are constructs of the real situation, useful for testing hypotheses and determining modeling consequences of interrelated variables, but our results can only be used relatively and not in an absolute sense. What is needed is GPS-based school commuting monitoring information similar to that obtained by Elgethun et al (2007) in a study of young children's movements during the weekend and by in a school-day study in Australia. The latter used GIS-based modeling techniques to allocate school trips much as we did, and then compared their allocations with actual routes taken on 59 to-school and 54 from-school (each one-way) commutes by elementary school children using a GPS unit.…”