Welcome to our research about “young people on the move”!
This page shares findings from BIFYA's work on issues related to housing and geographic mobility. Below, BIFYA's postdoctoral scholar Seva Rodnyansky and MPP candidate Sydney Bennet describe the effects of moving on young people's voting behavior. Their findings may surprise you!
Young Americans vote less. Is it because they move frequently?
By Seva Rodnyansky, PhD and Sydney Bennet
In the United States, demographic and socioeconomic characteristics influence the act of voting. Young people vote at much lower rates than middle-aged and older Americans. This is true for both Presidential and Midterm elections and has been covered by the news media (e.g., NPR, Washington Post) and in several academic journals (e.g., Smets & Van Ham 2013, Jankowski & Strate 1995, Strate et al. 1989).
Looking at the data on elections from 2000 to 2016 in Figure 1 below, it is clear that the younger you are the less likely you are to vote. These data from the Voter Supplement of the U.S. Current Population Survey show three key trends:
- Voter turnout is 17 percentage points lower for 25-34 year olds and 28 percentage points lower for 18-24 year olds than for those over 50 in Presidential elections
- In midterm elections, this gap grows to 33% for 25-34 year olds and 44% for 18-24 year olds
- The above trends are stable from 2000-2016 regardless of which party held the presidency or majority in Congress
Young persons’ persistent lower turnout reduces their political voice and thus their political representation. Understanding the reasons behind these behaviors could increase the political power and voice of younger generations.
Figure 1. Voter Turnout Higher for Older Voters
Turnout by Age, 2000-2016
Note: Chart includes data on both Presidential (solid line) and Midterm (dashed line) elections
The most popular theories for explaining the relationship between age and voter turnout explore differences in education, personal income, parental income, the degree to which politics was discussed growing up, impediments to voter registration, other resource constraints, and civic apathy. However, none of these have satisfactorily explained the behavior of the youth vote.
What is often overlooked is the relationship between residential mobility (moving from one address to another) and voting behavior. Young Americans move residences more often than middle-aged and elderly persons. In 2018, about 20% of 20- to 29-year olds moved, compared to 8-10% of 45- to 49-year olds and less than 5% of those above age 59 (Figure 2). Many reasons exist for variation in residential mobility including differences by age, renting versus homeownership, household formation and size, rootedness in a community, and physical health, among others.
Figure 2. Mobility Rate by Age Group in the U.S. in 2018
Is moving associated with voting less?
The duration of residence in a particular location is another factor that affects voter turnout. Measuring voter turnout by the number of years a person has lived at their address reveals that the longer a person lives in one place, the more likely they are to vote. As Figure 3 shows, in both presidential and midterm elections, voter turnout was 11 percentage points higher among Americans who lived at an address for 2-3 years versus those who lived at their address for 1 year or less. Recent movers (less than 1 year at address) have voter turnout 23 percentage points lower than long-term residents (5+ years at address) in presidential elections and 34 percentage points lower in midterm elections.
This makes intuitive sense. Recent movers may be less familiar with the politics in their district, may not know their polling location, may not have received voter information guides, or for very recent movers may have other demands on their time related to settling in to their new locale. They also may have experienced less political or civic outreach and may have seen fewer advertisements than longer-term residents. The cumulative effect of these mobility consequences may drive the lower voter turnout.
Figure 3. Voter Turnout Increases with Duration at Residence
Turnout by Years at Address, 2000-2016
Are young people voting less because they move more often?
The charts above tell the following story: Young Americans vote less often, but move more often than middle-aged and older Americans. Recent movers—regardless of age—generally also vote less often compared to longer-term residents. So, does frequent moving better explain the difference in voter turnout by age than other theories?
To test this hypothesis, we conduct a statistical regression analysis to see the effect of the combination of age and mobility on voter turnout. The statistical model controls for the effects of multiple explanatory variables (such as age, mobility, demographics, etc.) on the dependent variable, in our case, voter turnout.
For this exercise, we use publicly available data from the U.S. Census Bureau’s Current Population Survey’s Voter Supplement for the presidential and midterm elections between 2000 and 2016 (Flood et al. 2018). This survey includes about 80,000 respondents who answer questions about their voting behavior, as well as other demographic and socioeconomic variables. This survey is designed to be nationally representative for the whole United States.
To set up the analysis, we separate survey respondents along two dimensions: age and residential mobility. For age, we create two groups: Young (below age 35) and Non-Young (above age 35). For residential mobility, we also create two groups: Recent Movers (lived at current address for 1 year or less) and Stayers (lived at address for more than 1 year). We include respondents’ income, educational attainment, and race in the statistical model, all of which have been shown to influence voter turnout.
The model results confirm voter turnout increases with age and duration of residence at an address independently, but the combination of those variables creates a curious twist. We now highlight model results displayed in Figure 4:
Non-Young Stayers have the highest turnout in every year’s election, as expected, and form the baseline category for this analysis [Gray line on Figure 4]
Age differences account for an 18 (presidential) to 29 (midterm) percentage point difference in voter turnout between Young Stayers and Non-Young Stayers [Blue line on Figure 4]
Residential mobility differences account for 17 (presidential) to 25 (midterm) percentage point difference in voter turnout between Non-Young Movers and Non-Young Stayers [Orange line on Figure 4]
We provide two estimates of Young Mover turnout from the model:
- In the first estimate, we calculate the EXPECTED turnout among Young Movers for each election, while controlling for demographic factors. This provides us with a predicted level of turnout among young movers, holding all other factors equal. Results show a 35 (presidential) and 54 (midterm) percentage point difference in voter turnout between Young Movers and Non-Young Stayers [Green Dotted line on Figure 4]. This would mean that Young Movers were not expected to vote in the Midterms at all based on their characteristics.
- In the second estimate, we calculate the ACTUAL turnout in each election among Young Movers. Assuming there are no other important variables missing from our model, we can interpret the difference between these models as the isolated effect of moving on young people’s voter turnout. The ACTUAL voter turnout among Young Movers is 8 (presidential) and 14 (midterm) percentage points above the EXPECTED rate [Green Solid line on Figure 4].
Thus, a recent move appears to prompt a higher than expected probability of voting among young people, all else equal. The results from these models are statistically significant at the 99th percentile with p-values less than 0.001, meaning that there is less than a 0.1% percent chance that these effects occur by random.
Figure 4. Combined Effect of Age and Residential Mobility on Voter Turnout 2000-2016
Statistical Model Results
Note: Equation used to estimate Figure 4, using the Linear Probability Model (LPM), where Voter Turnout %= β_0*Baseline Turnout+ β_1*Age+ β_2*Residential Mobility+ β_3*(Age*Residential Mobility)+ β_4*Income+β_5*Educational Attainment+β_5*Race+ε
Are there other unaccounted for explanations? We amended the statistical model in several ways to see if the results change.
- Is this a college-student effect? No; limiting the model where Young is 25-34 years old does not change the results.
- Is it due to educational differences? No; running the model for respondents without college degrees does not change the results.
- Is the effect sensitive to the duration at residence definition? No; running the model with a Mover definition of up to 3 years does not change the results.
Is the effect a peculiarity of the data or the survey? No; running the model on two other survey datasets (American National Election Study 2012, 2016 and the Cooperative Congressional Election Study 2016) does not meaningfully change the results.
Young recent movers appear to have higher voter turnout than expected. This finding comes as a surprise given the general voting trends by age and duration at residence.
What are the mechanisms behind the effect? Does the act of moving motivate younger people to participate politically? Do young movers believe their vote counts more in their new precinct? Does a recent move increase voter registration and thus increase voting?
Young Movers mark a bright spot amid the declining political participation among younger persons in the U.S. The size of the Young Mover vote may meaningfully alter the outcomes of close elections. This is especially true in midterm elections, when general participation is lower than for presidential years.
Using our model, we attribute the Young Mover Effect to raising the number of voters in the 2000-2016 elections by about 380,000 – 790,000 voters (Tables 1 and 2, column 2). While these numbers may seem small, they may be enough to sway an election, depending on who they vote for and in which state. In 2000, the number of votes attributed to the Young Mover Effect was larger than the popular vote difference (Table 1). The 2000 presidential election was very close with George W. Bush eventually beating Al Gore in the electoral college after recounts and legal action. Without the Young Mover Effect, the election may not have been that close. In the 2004-2016 presidential elections, where the popular vote difference was larger than in 2000, the Young Mover vote accounted for a potentially meaningful 8-18% of the popular vote difference.
Table 1. Young Mover Effect and Presidential Elections
The Young Mover Effect comprised about one half of a percent of all counted ballots or about 380,000-500,000 votes per year (Table 2). While it is more difficult to judge the effect of this increased turnout in midterms where there are no national-level elections, these numbers likely influenced the outcomes of key local races.
Table 2. Young Mover Effect and Midterm Elections
Flood, S., King, M., Rodgers, R., Ruggles, S., and Warren, J.R. Integrated Public Use Microdata Series, Current Population Survey: Version 6.0 [dataset]. Minneapolis, MN: IPUMS, 2018. https://doi.org/10.18128/D030.V6.0
Jankowski, T. B., & Strate, J. M. (1995). Modes of participation over the adult life span. Political Behavior, 17(1), 89-106.
Khalid, A. May 16, 2016. Millennials Now Rival Boomers As A Political Force, But Will They Actually Vote? National Public Radio. https://www.npr.org/2016/05/16/478237882/millennials-now-rival-boomers-as-a-political-force-but-will-they-actually-vote
McDonald, M. n.d. “Voter Turnout”. United States Elections Project. http://www.electproject.org/home/voter-turnout/voter-turnout-data
Rampell, C. July 23, 2015. Where are all the young voters? Washington Post. https://www.washingtonpost.com/opinions/where-are-the-young-voters/2015/07/23/2781990e-316f-11e5-8f36-18d1d501920d_story.html?noredirect=on&utm_term=.c3b6739abc9a
Roper Center for Public Opinion Research. N.d. “Presidential Elections”. Cornell University. https://ropercenter.cornell.edu/data-highlights/elections-and-presidents/presidential-elections
Smets, K., & Van Ham, C. (2013). The embarrassment of riches? A meta-analysis of individual-level research on voter turnout. Electoral studies, 32(2), 344-359.
Strate, J. M., Parrish, C. J., Elder, C. D., & Ford, C. (1989). Life span civic development and voting participation. American Political Science Review, 83(2), 443-464.
U.S. Census Bureau. “CPS Historical Migration/Geographic Mobility Tables. Table A-1. Annual Geographical Mobility Rates, by Type of Movement: 2018”. https://www.census.gov/data/tables/time-series/demo/geographic-mobility/historic.html
 To obtain these numbers, we multiply the number of Young Movers who voted by the difference between the ACTUAL and EXPECTED outputs in the statistical regression for each year 2000-2016.