Rules for Assessing Data Quality of Activity-Scheduling Survey Respondents

The development of standards and guidance for applying quality measures in travel surveys is receiving increased attention. The most commonly reported quality measures are unit nonresponse and trip rates. How such a standard would be applied for assessing the quality of a single respondent's data remains unexplored. Furthermore, new quality indicators are needed for emerging travel survey methods, such as multiday and computerized scheduling surveys that capture observed activity travel patterns as well as information on underlying activityscheduling decision processes. A new set of rules is proposed for assessing data of individual respondents by using results from a recent survey of this type. The use of computers and a focus on decision processes led to developing several new quality indicators related to respondent login durations, data entry delays, entry timing, and recall rates. Analysis of these indicators was used to propose six rules for identifying goodquality data. A combined single rule was also developed by using principal component analysis. Further analysis of this combined indicator revealed that sociodemographic characteristics of the respondents were largely insignificant—people of all ages, genders, marital status, and employment or student status were equally able to provide good data. It also suggests that use of the combined indicator to filter out poorquality data will not necessarily introduce any additional bias into the resulting reduced sample set. The weak correlation of respondent activity or trip rates with the combined indicator implies that their usefulness for assessing data quality at an individual level is questionable.