Combining Disparate Surveys Across Time to Study Satisfaction with Life
Combining Disparate Surveys Across Time to Study Satisfaction with Life
Co-Principal Investigator: Giovanni Circella, Senior Research Engineer, School of Civil & Environmental Engineering
Project Duration: 12 months
Project Budget (Federal UTC Funds): $109,458
Project Budget (Cost-share): $54,791
Abstract
In 2011, the United Nations General Assembly passed a resolution recognizing happiness and well-being as a fundamental human goal, and followed this up in 2013 by establishing an official International Day of Happiness. These actions attracted much attention from the international community, and especially from those within academia, generating a surge of popular news and academic pieces on well-being and its variants. However, psychologists and social scientists have been studying happiness and subjective well-being (SWB) for decades based on large-scale longitudinal surveys. For example, Harvard Medical School’s Study of Adult Development is the longest-running study of adult life (ongoing since 1939), and focuses on well-being during adulthood (McLaughlin et al., 2010; Waldinger et al., 2007). The World Values Survey (WVS) is another well-known longitudinal study, originating in 1981, spanning almost 100 countries, and spawning numerous contributions to the SWB literature due to its open availability (Kim, 2018; Sarracino, 2010). Other established sources of longitudinal well-being data include the British Social Attitudes Survey (BSA; Dean and Phillips, 2015), the European Social Survey (ESS; Welsch and Kuehling, 2017), the U.S. General Social Survey (GSS; Ifcher and Zarghamee, 2014), and the International Social Survey Program (ISSP; Levin, 2014).
These large-scale longitudinal studies have allowed researchers to model the effects of general variables such as demographic characteristics, and selected values and behaviors on SWB. However, because these longitudinal surveys are broad in nature, they do not facilitate the examination of SWB within specific contexts or with the help of more diverse explanatory variables. As a result, researchers within assorted fields have taken to studying SWB using cross-sectional surveys, which are more commonly available and facilitate investigation from specific perspectives (e.g., effects of health, occupation, etc. on well-being). A number of scholars have used cross-sectional surveys to examine the impacts of transportation, especially commuting, on well-being (Mokhtarian, 2019, De Vos et al., 2013; Dickerson et al., 2014; Lorenz, 2018; Martin et al., 2014; Smith, 2017; Sweet and Kanaroglou, 2016).
In this proposed study, we plan to combine the longitudinal and cross-sectional approaches to studying well-being, creating a fused data set that includes common variables from five travel-behavior-oriented cross-sectional surveys conducted across a 27-year period. The PI of this study was heavily involved in all five of these surveys, while a co-PI was heavily involved in the most recent three. Accordingly, each survey includes an identical SWB question, as well as numerous other common variables across the individual datasets. Since these surveys were originally designed to serve travel behavior modeling purposes, the development of this fused dataset will allow a unique examination of SWB within a transportation context.
Despite the continuity of some design factors across the five cross-sectional surveys, there are inevitable inconsistencies stemming from question wording differences and evolving survey design techniques over the years. In this study, we will demonstrate an approach for addressing and ameliorating such inconsistencies using a combination of survey fusion and model development techniques. As such, one contribution of this work will be to provide an evolution of variables, in this case, SWB, over time. Accordingly, this study will both: (1) provide a detailed examination of SWB from general as well as transportation-oriented perspectives; and (2) provide an example of combining cross-sectional survey datasets for longitudinal studies.