Doctoral Dissertation Research: Memory Bookmarking: An Approach to Improve Recall in Surveys

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

This research project will advance survey methods for time-use measurement. Social scientists widely use retrospective survey questions (questions about events that occurred in the past) to gather information about respondents' recent autobiographical events. Recall of even recent events is burdensome for respondents, however, and it often leads to uneven data quality due to recall error. This project will experimentally evaluate a new approach called Memory Bookmarking (MB) to address recall error in surveys. The MB approach collects information about respondents' experiences while they are underway. This information is later fed back to respondents as memory cues or bookmarks. Depending on the experimental condition, respondents will be asked to provide bookmarks about everyday events, in the form of text messages, photographs, or locations. The approach should not only improve memory for the cued events but also help respondents remember other events that are temporally adjacent and semantically related to the cued event -- much like an actual bookmark allows readers to see what is on the previous and next pages. The results of this project will advance current thinking about autobiographical memory, especially about the formats (verbal, visual, and spatial) in which people primarily represent the events of their lives. The project will use techniques from data science to analyze unstructured data and thus contribute to ongoing efforts to integrate data science techniques into survey methodology. As a Doctoral Dissertation Research Improvement award, support is provided to enable a promising student to establish a strong, independent research career. The MB approach will be evaluated in a factorial experiment in which respondents are asked to reconstruct the full sequence of events they have experienced over the prior forty-eight hours. MB respondents are "pinged" on their smartphone at designated times and asked to provide information via their smartphones about the event that is in progress at that moment. This information is then presented to the respondents as memory cues to help them reconstruct all their activities, not just the pinged events, during the prior forty-eight hours. Respondents' recall will be compared between the experimental conditions created by crossing the number of pings with the bookmark formats, and between the bookmark conditions and a no-bookmark control condition. The research will address the following questions: Does the MB approach lead to more accurate and complete recall than when no cues are provided? How effective are the three formats of memory cues (verbal, visual, and spatial) for different types of events? Will respondents reply to pings in a timely manner during the forty-eight hours, especially as the number of bookmarks increases?