Characteristics of the Mobile Population

Abstract

This chapter explores current trends in household mobility by describing correlates of moving and individuals’ reasons for moving, exploring answers to the question, “Who moves and why?” While the research presented in Chap. 2 described household mobility trends in a large, demographic context, this chapter describes sociodemographic variations in household mobility and individuals’ reasons for moving. Drawing on nationally representative data from 2014–2015, household mobility and reasons for moving are explored as they occur differentially across age groups and by social class, race/ethnicity, housing tenure, and other demographic characteristics, such as gender and labor force participation.The primary goal of this chapter is to emphasize the variation in characteristics of movers and their reported reasons for doing so.

Appendix: Additional Data and Analysis Details for Chap. 3

IPUMS-CPS ASEC Sample and Measures

The CPS is a monthly household survey sponsored by the US Census Bureau and the Bureau of Labor Statistics. The sample universe includes noninstitutionalized Americans, including individuals in armed forces but not stationed or living on a military base. An Annual Social and Economic Supplement (ASEC) to the monthly CPS collects data on household and family characteristics, migration, and employment.

Measures

Dependent Variables

Household mobility, also referred to as geographic mobility or migration in the CPS, refers to any change of residence of any distance. The CPS measure is derived from a question about respondent’s residence 1 year prior to the survey date and the respondent’s current residence. This information is collected from all members of the survey household who were 1-year old or over on the date of the survey. Movers are categorized as any individual who reported a different residence in the period between the survey date and 1 year prior. Additional data are collected on whether the relocation took place across county, state, or regional boundaries.

The ASEC also includes information on individuals’ reasons for moving. These reasons are based on the householder’s reported reason for moving, which are assigned to other individuals in the mobile household. In order to look for general differences in why people move, individuals’ reasons for moving are often collapsed into four broad categories: family related, employment related, housing related, and other reasons. Each of the detailed and collapsed reasons for moving used by the CPS are provided in Table 3.3. For a detailed description of the methodological details regarding data collection for this variable, including the history of the measurement and some limitations, see Ihrke (2014:2013–2015). In addition to the aggregate data presented and discussed in Chap. 2, the CPS also provide Integrated Public Use Microdata Series (IPUMS-CPS), which are the individual-level data collected by the CPS in the ASEC each year.

Independent Variables

The models in Chap. 3 explored correlates of household mobility, type of move, and reasons for moving based on a number of sociodemographic variables. Respondent’s age was characterized as young adult (age 18–34), which was the omitted reference; midlife adult (age 35–49); or older adult (age 50+). A dichotomous variable categorized an individual’s housing tenure as either (1) renter occupied or (2) owner occupied. A measure for employment status indicated whether the respondent was employed, which was the omitted reference; unemployed; or not in the labor force.

Respondents’ poverty status was a dichotomous measure for whether their household income was (1) above the poverty line or (0) at or below the poverty line. A variable for level of education indicated whether an individual had less than a high school education, which was the omitted reference category; a high school diploma, some college or an AA degree; a college degree; or a graduate degree.

Gender was assessed with a dichotomous variable for whether the respondent was (1) male or (2) female. Additionally, a dichotomous variable for respondents’ nativity marked whether an individual was (1) native born or (2) foreign born. The race/ethnicity categories indicated whether respondents were White only, which was the omitted category; Hispanic only; Asian/Pacific Islander; Black only; and/or additional races/ethnicities, including multiracial individuals. Marital status was measured as married, which was the omitted category; separated; divorced; widowed; or never married, which includes never-married cohabitors. A dichotomous variable indicated whether an individual had coresidential children or not.

Analysis Details

The models in Chap. 3 are based on IPUMS-CPS data on household mobility between 2014 and 2015. The CPS is unable to account for the household mobility of approximately 10–12 percent of their ASEC respondents each year. As such, in order to account for this missingness, the CPS imputes data for these cases based on the individuals’ predicted propensity to move based on characteristics that match a data profile (for more information, see Kaplan and Schulhofer-Wohl 2012). Additionally, the analyses incorporate replicate weights provided by the CPS in order to account for the complex sample design and provide reliable estimates of standard errors.

For each of the models discussed below, collinearity diagnostics revealed that multicollinearity was low (average VIFs < 1.5). Analysis of the correlation matrices (not shown) indicated that none of the observed relationships between the independent variables in the models were very strong. Tables 3.5 and 3.6 in Chap. 3 present the results of logistic regressions on household mobility (no move or any move) and household mobility type (local or distance move) between 2014 and 2015. Table 3.7 in Chap. 3 presents the results of multinomial logistic regression to explore sociodemographic correlates of reasons for moving. Descriptive statistics for the sample are provided in Tables 3.2 and 3.4. In Table 3.2, percentages are provided separately for nonmovers, local movers, and distance movers. In Table 3.4, percentages are provided separately based on individuals’ reported reasons for moving.

Limitations

Following the rationale of Cooke (2011), the CPS was preferred over other data sources because (a) it is nationally representative and (b) the data set contains county-based measures of household mobility that are not available in other national surveys. There are, however, a number of data limitations that warrant discussion. First, the analyses included individuals in the household who were not the household head, which may have inflated estimates for individuals’ reasons for moving since reasons for moving are based on the householder. Others who have used the CPS to examine reasons for moving have restricted their sample to householders in order to avoid assigning the householders reason for moving to other individuals in the household (Ihrke 2014). However, in order to adequately assess how sociodemographic characteristics (e.g., marital status) influence household mobility and reasons for moving, isolating household heads alone could also bias the results. Additionally, including non-householders allowed for comparable models for the household mobility (Tables 3.5 and 3.6) and reason for moving (Table 3.7) analyses.

Another limitation is that the study of household mobility is complicated by the relative distance of a move. Kivisto and Faist (2010:3) emphasized this problem in terms of international migration, pointing out that “a move from Boston to Los Angeles is, in terms of distance, farther than the move from Puebla, Mexico to Los Angeles.” Along the same lines, the conceptualization of short- and long-distance mobility in this study is imperfect since the distance of a move, when based on county boundaries depends on the size of the county and a household’s proximity to the county line. For example, an intercounty move could be undertaken over a relatively short distance, while a move considered here as occurring “locally” could, in comparison, be undertaken across a relatively greater geographic distance.

An additional and important data limitation is that respondents can only choose a single reason for moving, which is often described as their “primary reason for moving.” Often, individuals have a number of reasons for moving that influence their ultimate decision to do so. Also, as indicated in the chapter, the cross-sectional nature of the data makes it difficult to make causal statements about housing tenure, employment, and other variables since they may have occurred before or after the data were collected. Lastly, only the civilian and uninstitutionalized population is included in the CPS sampling frame. As such, individuals who reside in, or have recently relocated to, formal residential care facilities are not included in the population universe. As a result, these analyses may underestimate the household mobility of later-life individuals who transition into these institutionalized settings.