Jump to ContentJump to Main Navigation
Altered StatesChanging Populations, Changing Parties, and the Transformation of the American Political Landscape$

Thomas M. Holbrook

Print publication date: 2016

Print ISBN-13: 9780190269128

Published to Oxford Scholarship Online: August 2016

DOI: 10.1093/acprof:oso/9780190269128.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy).date: 25 March 2017

Population Migration and Political Change

Population Migration and Political Change

(p.53) 3 Population Migration and Political Change
Altered States

Thomas M. Holbrook

Oxford University Press

Abstract and Keywords

This chapter takes a close look at how migration patterns alter the political landscape of states, in part by influencing changes in state population characteristics. The migration analysis addresses the effects of both internal (state-to-state) and external (foreign) migration. It shows that population migration is a potentially important source of demographic and political change in the states. But not all states are affected equally by migration, with some state experiencing higher levels of migration compared to others. The source of migration streams is also as important as the volume of migration. Separating the foreign-born population from internal migrants, for instance, shows that states with a high level of foreign-born population are almost uniformly states that also have large increases in support for Democratic presidential candidates. In addition, states with a large number of internal migrants from liberal (conservative) states tend to move in a Democratic (Republican) direction.

Keywords:   migration pattern, political change, political outcome, state population, internal migration, external migration, demographic change, immigrant population

In this chapter I examine how changes in the composition of state populations contribute to changes over time in patterns of political support for presidential candidates across the states. The strategy I use is to take a closer look at migratory patterns to get a handle on the extent to which migration from state to state, as well as immigration from outside the United States, alters the political landscape of the states. This involves identifying characteristics and categories of migrants that are likely to have the greatest impact on political outcomes and then assessing the direct effects of migration on political change. This focus on migration provides insights into an important process that contributes to changes in other population characteristics, which are examined in greater detail in chapter 4.

As discussed in chapter 2, population migration is an important potential source of political change in the states. While population characteristics in a given state can change over time without significant in-migration, perhaps through generational replacement and differential birth rates across groups, they also can change as new people move to the state either from other states or from foreign countries. The impact of migration is likely to depend on the overall scope of migration and the extent to which new state residents are different from state natives in politically relevant (p.54) ways (Gimpel and Schuknecht 2004; Jurjevich and Plane 2012; Robinson and Noriega 2010). In states in which there is relatively little in-migration the differences between migrant and native populations would have to be quite stark in order to produce substantively important political change. On the other hand, in states with substantial levels of in-migration, even somewhat modest differences between migrant and native populations could move the state in one or the other political direction. And of course the most substantial impact would occur in states with substantial levels of in-migration and in which the migrant and native populations were substantially different in politically relevant ways.

The Scope of Migration

The maps presented in Figure 3.1 and the data in Table 3.1 shed light on part of this equation, illustrating the scope and sources of population migration across the states. These data are based on estimates from the 2012 American Community Survey of the percentage of each state’s citizen voting-age population (CVAP) born outside the state.1 The CVAP is used as the base here and for most of the other population statistics in this chapter. This is done for a couple of reasons. First, for some variables related to race, ethnicity, and migration, a significant share of the population of interest is not eligible to participate in elections due to citizenship status. Using CVAP as the basis for measuring relative group size focuses on that part of the population that is likely to have the greatest impact on political outcomes: those who are eligible to vote.2 More detailed information on the measurement of state characteristics is provided in the appendix.

The top map in Figure 3.1 (see also the first column of data in Table 3.1) makes clear that the potential for migration to affect political change varies considerably across the states, based on the scope of in-migration. States with dark shading have the highest levels of in-migration, and states with light shading have the lowest levels of in-migration. As a point of reference, nationwide in 2012, 43% of the CVAP were living somewhere other than the state or nation where they were born, a level (p.55)

Population Migration and Political Change

Figure 3.1 Size and Sources of the Migrant Population in the States, 2012

Note: Each map uses a different scale due to differences in the range of each measure of migration. All maps were created using Choroplethr.

that would register on the lighter end of the scale. There are a couple of different patterns to note here. First, generally migration is tied to population growth: the correlation between state population growth from 1970 to 2010 and the percentage of state population born out of (p.56)

Table 3.1 Breakdown of the Citizen Voting-Age Population by Migration Status by State, 2012


Percentage Born Outside of State

Percentage Born in Other States

Percentage Foreign-Born

















































































































New Hampshire




New Jersey




New Mexico




New York




North Carolina




North Dakota




















Rhode Island




South Carolina




South Dakota




























West Virginia












Fifty-state Average




Note: All data are taken from the 2012 American Community Survey.

(p.57) state in 2012 is .75. This is not universally the case—for instance, Utah and Texas have relatively modest in-migration rates, given their population growth rates—but generally states that grew the most tend to have larger nonnative populations. Second, there is a regional pattern to the size of the nonnative population, no doubt also reflecting differences in population growth: Mountain West and southwestern states, along (p.58) with Florida and several southeastern states, have higher in-migration, and the industrial Midwest and Northeast and parts of the Deep South tend to have smaller state-native populations. Third, in a handful of states the size of the native population is swamped by that of the population born out of state: states with the largest nonnative populations are Nevada (86%), Florida (72%), Arizona (71%), Alaska (68%), Wyoming (67%), and Colorado (66%). At the other extreme a number of states have relatively few residents who were born elsewhere: Louisiana (22%), Michigan (24%), Pennsylvania (27%), Ohio (27%), Wisconsin (30%), and Mississippi (30%). These patterns, in part, reflect the industrial decline of Rust Belt states and increased economic opportunities in the southeastern and western states.

It also is important to make distinctions regarding the sources of these migration patterns. In very broad terms the nonnative population in any given state is composed of people who were born in other states or U.S. territories (internal migrants) and people who were born outside of the United States (foreign-born). In 2012, 34% of the CVAP were internal migrants, and 9% were born in a foreign country. This distinction is important because internal migrants and foreign-born residents are likely to differ on characteristics that could have important political consequences. The distinction is also important because there are substantial differences in the type of state that tends to attract more internal versus foreign-born migrants. The second map in Figure 3.1 (second column of data in Table 3.1) shows the distribution of internal migrants as a percentage of state CVAP. It looks a lot like the top map, with a couple of prominent exceptions: New Hampshire emerges as one of the states in which internal migrants constitute a relatively high percentage of the state’s population (60%), and New York as the state with the smallest concentration of internal migrants, at 17%. Otherwise the pattern is familiar, with the Mountain states, southwestern states, Florida, and southeastern states with high concentrations of internal migrants.

The picture is appreciably different, however, when considering foreign-born population as a percentage of state population (bottom (p.59) map in Figure 3.1, third column of Table 3.1). In this case a handful of states stand out as having the highest percentage of their population born outside the United States: California (22%), New York (18%), New Jersey (16%), Hawaii (16%), Florida (15%), and Nevada (14%). Together these states account for the destination of 55% of the foreign-born CVAP. By comparison, these same states account for only 26% of the internal migrant population. States in which the foreign-born constitute a relatively small share of the state population are found primarily in the Midwest (with the exception of Illinois), the Plains states, and the Border South and Mississippi Delta states. A handful of states (Kentucky, Mississippi, Montana, South Dakota, and West Virginia) had foreign-born CVAP of 2% or less.

Demographic Change and Migration

Although the impact of migration on political change in the states is likely a function of the size of the overall migrant population in any given state, it is also likely to be a function of the extent to which migrants are different from natives in politically relevant ways. These differences vary somewhat from state to state, but we can get a sense of their potential importance by examining them over time for the nation as a whole. Table 3.2 provides a glimpse into potentially relevant demographic and attitudinal characteristics of three different groups: foreign-born migrants, internal migrants, and people who live in the state where they were born (natives). Included here are measures of standard demographic characteristics: educational attainment, occupational status, race and ethnicity, marital status, poverty rate, and age. Also included are two measures of political attitudes: net party identification (percent Democratic identifiers minus percent Republican identifiers) and net ideology (percent liberal identifiers minus percent conservative identifiers). These data are examined in two different time periods—1970–80 and 2004–12—that bookend the changes in party support examined in chapter 1. With this table we are able to examine demographic and attitudinal differences (p.60)

Table 3.2 Selected Characteristics of Migrant and Nonmigrant Populations, 1970–80 and 2004–12





Internal Migrants

State Natives


Internal Migrants

State Natives

% BA







% Advanced Degree







% Management







% Professional







% White







% Black







% Latino







% Other Races







% Single







% Poverty







Average Age







% Dem−% Rep







% Lib−% Con







% of Total







Note: Demographic data are based on estimates from the decennial census for 1970 to 1980 and the American Community Surveys for 2004 to 2012. Data on party affiliation and ideology are taken from the General Social Survey (GSS), using data from 1977–80 and 2004–12. The census data represent estimates for the CVAP, while the GSS data represent estimates for the voting-age population. For party identification and ideology, both of which are drawn from GSS data, the internal migrant data represent the status of people whose current state of residence is different from where they lived when they were sixteen years old. More details on data gathering are provided in the appendix.

(p.61) across migration-based groups and changes in group-based differentiation over time.

Looking first at the early period, there were relatively few substantial demographic differences by migration status: the group differences in educational attainment and occupational status were relatively slight; all three groups were overwhelmingly white, with Latinos constituting a somewhat greater share of the foreign-born population than internal migrants or state natives; and foreign-born were somewhat less likely to be single and were on average about fifteen years older than native-born CVAP. In terms of party identification and political ideology, internal migrants were somewhat less Democratic and more conservative than were foreign-born and state natives, while all three groups tilted Democratic and also slightly conservative to varying degrees.3 All in all, the differences are relatively small, and those differences that do exist sometimes pull in different directions. For instance, foreign-born residents might be expected to be somewhat less likely to support Democratic candidates based on their marital status and age, but they have a higher level of Democratic identification and are (slightly) less conservative than native-born residents, which should point them toward the Democratic camp.

The group differences are much more dramatic in the later time period: both foreign-born residents and internal migrants have higher levels of educational and occupational status than state natives; whites constitute a relatively small minority among the foreign-born population, trailing behind Latinos and “other” racial and ethnic groups; and differences in political attitudes grew substantially in magnitude, perhaps as a result of these demographic changes. At a time when internal migrants and state natives grew more Republican and more conservative, foreign-born residents maintained their allegiance to the Democratic Party and also barely changed on the ideological front. The change in the racial and ethnic composition of foreign-born residents is particularly notable, reflecting a substantial difference between the two time periods in the geographic basis of immigration from other countries. Data from the 1970 and 1980 census studies show that, at that time, fully 61% of the foreign-born CVAP (p.62) came from Europe, 13% from Latin America, 9% from Asia, 1% from Africa, and 16% from other areas (e.g., Canada, Australia). For the period from 2004 to 2012 data from the American Community Surveys show that 37% of the foreign-born CVAP came from Latin America, 35% from Asia, 21% from Europe, 4% from Africa, and the rest from elsewhere. The internal migrant population from 2004 to 2012 provides a bit of a mixed picture: it looks very much like the foreign-born population in terms of education, occupation, average age, and marital status but is more similar to state natives on race, ethnicity, and political attitudes, though leaning somewhat more Democratic and at the same time somewhat more conservative.

On balance, these differences support the expectation that states with a substantial foreign-born population should see increases in Democratic support from the 1970s to the 2010s. Also, given the increasing differences between foreign-born and other citizens over time, we should see important changes in the relationship between the percentage of foreign-born and Democratic vote shares over time. At the same time, it is not clear that the level of internal migration in a state necessarily should benefit one party or the other, at least based on characteristics of the migrating population compared to those who have not moved across state lines. There are some differences between internal migrants and native state residents, but perhaps not large enough to produce significant change without taking a more nuanced look at how internal migrants differ from state to state.

Migration and Election Outcomes

We can begin to get a sense of the impact of migration on political change by looking at the simple bivariate relationships between migration differences among the states and changes in Democratic performance in state-level presidential outcomes. Before decomposing state population by source of migration, it is useful to examine the impact of migration overall on state outcomes. The top left panel (p.63) of Figure 3.2 examines change in Democratic vote as a function of the percentage of the state population born out of state. The change in Democratic vote summarizes the state-by-state regression slopes presented in Figures 1.3 and 1.4. Instead of using the slopes themselves (which can be intuitively cumbersome) as the dependent variable, this figure looks at the change in the average estimated centered Democratic vote from 1972–80 to 2004–12, which is derived from the slopes.4 These averages are also the same as the difference between the starting and end points of arrows representing state change in Figure 1.5. The horizontal axis in the top left panel of Figure 3.2 represents the total nonnative population (CVAP) of the states in 2008, the midpoint of the contemporary election period measured in the dependent variable. This includes both internal migrants and foreign-born state residents. Here we see that, by looking at overall migration without considering the sources of migration, the impact on change in party support is relatively paltry. There is a slight positive trend to these data (r = .34), suggesting that Democrats have made modest gains in states with the highest percentage of residents who are internal or external migrants. But this is not a strong relationship and surely does not explain much of the Democratic gains and losses over time.

This overall view of the impact of migration may be a bit misleading in that it assumes that migration effects are constant across migrant groups, an assumption that is likely in error considering the demographic and political profiles of the migrant groups provided in Table 3.2. Given its distinct racial, ethnic, and political makeup, the foreign-born population would seem to be the most obvious place to start. The top right panel in Figure 3.2 presents the relationship between the percentage of the state citizen voting-age population who are foreign-born in 2008 and change in Democratic performance in state-level residential outcomes from 1972–80 to 2004–12. Here we see a much more dramatic relationship. Generally speaking, states in which the foreign-born population constitutes a relatively large share of the CVAP also tend to be states in which the Democratic Party has significantly improved its position over time (p.64)

Population Migration and Political Change

Figure 3.2 The Influence of Population Migration on Changes in Democratic Support, 1972–76 to 2004–12

Note: The dependent variable is change in estimated Democratic support from 1972–80 to 2004–12, based on the trend in the Democratic share of the two-party vote, centered around the national two-party division (see Figures 1.3 and 1.4). All independent variables are measured based on migration patterns in 2008. Weighted home-state liberalism is calculated using Equation 3.1.

(p.65) (r = .64). Although the prediction line is linear, there is a bit of a curvilinear trend in the data: among states with a relatively small foreign-born population (1 to 5%) there is a mix of political outcomes, though most states saw a decline in Democratic strength; beyond that point (greater than 5% foreign-born) there is a much stronger tendency for higher levels of the foreign-born population to be associated with greater Democratic success. One noteworthy exception is Texas, which had above-average levels of foreign-born population and a substantial decline in Democratic support. Otherwise there are no states with high levels of foreign-born population and declining Democratic electoral strength, leading to the completely empty lower right quadrant of the scatter plot. The size of the foreign-born population appears to provide a particularly good explanation for changes in Democratic strength for a number of states with especially high levels of foreign-born population: California, Florida, Massachusetts, Nevada, New Jersey, and New York all fall very close to a prediction line. One other state that stands out as a bit of an anomaly is Vermont, where Democrats made their largest electoral gains, but where there is also very little foreign-born population to speak of. Obviously something else must be driving political change in Vermont. Otherwise the size of the foreign-born population appears to be an important part of the explanation for why Democrats improve their lot in some states more than others.

The key to this relationship is not just that some states have higher levels foreign-born population, but that higher concentrations of foreign-born population are connected to other politically relevant changes in state characteristics. For instance, Table 3.2 shows clear educational, occupational, and racial and ethnic differences between the foreign-born and native populations. States that experience changes in their foreign-born population should experience concominant changes in these politically relevant characteristics.

The expectations are less clear for the influence of internal migrants on political change, given that they are not much different from the state native populations, at least on racial, ethnic, and political dimensions. This is borne out by the data in the lower left panel of Figure 3.2, (p.66) which uses the percentage of the CVAP in 2008 who were internal migrants as the independent variable and change in Democratic support as the dependent variable. There is no clear pattern in this figure, certainly not one that rises to the level of being either statistically significant or substantively interesting (r = .16, p = .28). This null finding likely is an illustration of the fallacy of considering only the volume of internal migration without regard for the nature of the migrants. This figure ignores the possibility that the characteristics of internal migrants may differ from state to state in ways that have important political consequences. Consider, for example, Robinson and Noriega’s (2010) work on county-level migration in the Mountain West states. In responding to previous work that found overall migration levels were not related to changes in party success at the county level, Robinson and Noriega proffered that it is not enough to simply look at the levels of in-migration; it also is necessary to examine where those migrants came from. Specifically they took into account the partisan leanings of counties from which the migrants came and found that in-migration, weighted by the political leanings of the source counties, was strongly related to political change at the county level.

It is likely that the same sort of process is at work in the states. If a given state has a high level of internal migration, but those migrants come from a diverse set of states with no clear, consistent political orientation, then the level of internal migration is not likely to have much effect. However, if a state draws its internal migrants primarily from conservative or liberal states, then internal migration is likely to play a role in changing state political outcomes, and this effect should grow in magnitude as the level of internal migration grows. Similar to Robinson and Noriega’s argument, the expectation here is that the impact of internal migration is a function of both the level of internal migration and the political context of the states from which those migrants come.

While there are differences in sources of internal migrants across states, some states do stand out as primary sources. As might be expected with any measure based on population, large states lead the way. However, the leading sources of internal migrants tend overall to be centrist or liberal, (p.67) and competitive or Democratic states. New York leads the way, with 6.1% of citizen voting-age internal migrants claiming it as their birth state, followed by California (5.1%), Illinois (3.6%), Pennsylvania (3.3%), Ohio (2.9%), Texas (2.8%), Michigan (2.4%), and New Jersey (2.2%). Given that these states are overall more Democratic than the rest of the country, it is interesting that there is no relationship between internal migrant population and change in presidential votes. One potential explanation is that the Democratic advantage in these states is much greater now (the Democratic share of two-party vote ran about 8 points higher in these states than in the remaining states from 2004 to 2012) than it was when most of the internal migrants were born and raised in those states (only a 2-point difference from 1972 to 1980). This suggests that it is important to rely on historical patterns of political tendencies when trying to account for the impact of migration over the long haul. If we were focused on short-term migration—Where did the migrants live five years ago?—then capturing the current political environment of the states would make more sense.5

In addition, while the eight states listed above are the leading sources of internal migrants, some states draw very heavily from them, while other states draw very few migrants from them. In fact each state has its own set of source states from which it tends to draw the greatest share of internal migrants. To get a sense of these differences and how they can affect political change in the states, consider the experiences of Ohio, Vermont, and Wyoming. In Ohio 24.6% of the CVAP are internal migrants, and the leading sources for this population flow are (in order of magnitude) Pennsylvania, West Virginia, Kentucky, Michigan, and New York. These five states combine to account for 46% of Ohio’s internal migrant population. Given the relatively diverse political background of these states, it is no surprise that Ohio has moved very little politically over the past several decades. Vermont’s experience differs from Ohio’s in two important ways. First, half of Vermont’s CVAP was born in another state, increasing the likelihood that the internal migrant population could have a more profound impact on the movement in Vermont’s political profile. Second, the (p.68) leading sources of Vermont’s internal migrant population—New York, Massachusetts, New Hampshire, Connecticut, and New Jersey (in order of magnitude)—are Democratic and liberal, with the exception of New Hampshire, and constitute 65% of Vermont’s internal migrants. Finally, consider Wyoming’s experience: 61% of that state’s CVAP are internal migrants; the leading sources of this migration are (in order of magnitude) Colorado, California, Utah, Nebraska, and South Dakota; and these states together account for 38% of Wyoming’s internal migrant population. With the exception of California, these source states historically are relatively Republican and relatively conservative. Given the differences in scope and sources, we might expect internal migration to affect these states differently. Wyoming, with a large internal migrant population drawn from fairly conservative states, should move toward the Republican Party; Ohio, with a small proportion of its population being internal migrants drawn from a politically heterogeneous set of states, may not be affected very much; and Vermont, with about half of its population born in states that are more liberal than the rest of the country, should be moved in the Democratic direction. This is exactly what has happened in all three states (see Figure 1.5).

Internal migrants beyond those coming from the top five source states influence each of these states, so it is important to gauge the impact of internal migration on each state by taking into account the overall contributions of each of the forty-nine other states and their political leanings. I use estimates of state political ideology of each source state and weight those contexts by the proportion of the population in a given state coming from each source state.6 One tricky aspect of this measure is the need to capture state ideology not at the current time period but at some period in the past that reflects the political tenor of the birth state when internal migrants were likely to be socialized and become involved in the political process and hence is likely to reflect the predispositions of internal migrants. We don’t know when internal migrants left their birth state, so we have to use some measure of birth state context, accepting that it is a rough estimate of how the birth state milieu of internal migrants influences changes in state (p.69) politics. The average age for internal migrants in 2012 was fifty, which means the first election for the average internal migrant was in 1980. Using this as a reference point, the net ideology (% liberal minus % conservative) of each state, averaged for 1972–80, is used to estimate birth state ideological context, which is assumed to represent something like the expected ideological influence of migration streams coming from each migrant source state.

The method for calculating the weighted birth state ideology (WBI i) is presented in Equation 3.1, where ρ‎ ij is the proportion of the CVAP in state i that was born in state j, and IDj is the net liberal advantage (liberal minus conservative identifiers) in state j from 1972–80.

W B I i = j = 1 n ( ρ i , j * I D j )

This measure takes into account both the magnitude and political context of internal migration: high values indicate, on balance, greater liberal influence from internal migration, and low values indicate greater conservative influence, while a value of zero indicates a wash.

When this measure is calculated for Ohio, Vermont, and Wyoming, the resulting values make sense given the primary sources for internal migrants in each of these states. For Wyoming, a state that drew heavily from Mountain West and midwestern states, the overall value of the weighted internal migrant ideology variable was –.56, indicating, on balance, a conservative influence from internal migration. (The mean value for this variable across all fifty states is .49, indicating a somewhat liberal influence from migration.) For Ohio, a state with a more heterogeneous internal migration stream, the weighted internal migrant ideology variable was –.04, somewhat more conservative than the overall average but very close to a wash. For Vermont, a state that draws its internal migrant population heavily from the northeastern United States, the weighted ideology value was 2.50, indicating a very liberal influence from internal migration.

This measure of the potential political impact of internal migration rests upon the assumption that migrating voters reflect the political (p.70) tenor of the states in which they were born. This is not to say that every migrant born in Massachusetts, for instance, is a liberal Democrat, or that every internal migrant born in Utah is a conservative Republican, but on average people from Massachusetts will be more liberal and Democratic than people from Utah. As Robinson and Noriega (2010) point out, there is not a lot of evidence that people who migrate from a given state are substantially different in political views than people who stay in that state. This is the perspective I take, that, generally speaking, people who were born in (and presumably spent some time in) liberal states are, on average, more liberal than people born in conservative states. When those people move, they bring their political perspective with them and have an impact on the politics of their destination state, especially if migration patterns are consistently liberal or conservative and of significant magnitude.

Ideally we could test this idea by identifying the birth state of survey respondents who are internal migrants and then assess the extent to which their political views reflect the state where they were born. Unfortunately one of the primary sources of data on public opinion and elections, the American National Election Study (ANES), has not recorded birth state information since the early 1990s, and the General Social Survey (GSS) does not identify state of birth or current state of residence due to privacy concerns. However, both the ANES and the GSS ask respondents questions that allow us to identify where they spent part of their childhood. The GSS has a series of items about residential mobility since the age of sixteen, including one that identifies which of the nine census regions respondents lived in when they were sixteen.7 This variable can be matched with the region in which respondents currently live in order to identify those respondents who are regional migrants, having moved between regions since their teen years. While the ANES data do identify the state in which respondents grew up, the samples for most states are quite small, so I aggregated these responses by the same census regions used for the GSS data. This regional focus is a bit different from the way internal migrants are treated in the rest of this chapter, but it does give (p.71) us our best opportunity to see if, on average, internal migrants reflect the politics of the place from which they come.

The following analysis is limited to internal regional migrants. For the GSS this is native-born respondents who currently live in a region that is different from where they lived when they were sixteen, utilizing the 2008–12 GSS surveys. For the ANES this is native-born respondents living in a region that is different from where they say they grew up, utilizing the 2000–2008 ANES studies.8 Current ideological leanings of these respondents are measured using questions that asked respondents to place themselves on a 7-point scale, ranging from extremely liberal to extremely conservative. This question was then recoded into two dichotomous variables, one identifying respondents who provided a “liberal” response and one identifying respondents who provided a “conservative” response. These dichotomous variables were then aggregated by region, using the region in which the respondents lived when they were children. From this we can get the net ideology of internal regional migrants by childhood region. We can correlate this aggregated measure with the overall ideological leanings of those regions from 1972 to 1980, using the net liberal advantage measure that was used to create the measure of weighted home state ideology in Equation 3.1. If internal migrants, on average, reflect the politics of the place from which they came, a historical measure of regional ideology should be related to the average ideology of contemporary survey respondents who lived in those regions during their childhood.

This relationship is explored in Figure 3.3, where the top graph presents the relationship for GSS respondents and the bottom graph for ANES respondents. The horizontal axis is the historical net ideology (liberal minus conservative) averaged across states within each of the nine census regions. The vertical axis represents the net ideology averaged across survey respondents from within the same census regions, except that here region represents the region in which the respondents spent part of their childhood. These averages are restricted only to those respondents who live in a region different from the one in which they grew up (ANES) or lived when they were sixteen years old (GSS). This group represents 21% (p.72) of all respondents for both the GSS and ANES samples, yielding 1,140 respondents for the GSS and 1,003 for the ANES. The relationships are quite striking. Beginning with the GSS data, the mean ideological predisposition of internal regional migrants from 2008 to 2012 is very closely tied to the historical (1972–80) measure of regional ideological predisposition for the regions they lived in when they were sixteen (r = .82). A very similar pattern is found for ANES data, where the correlation between average current ideological orientation of region migrants from 2000 to 2008 and the historical measure of the ideology of region in which they grew up is .79.9 These findings do not necessarily say anything about individual regional migrants but instead speak to the overall outlook of the group. Regional migrants, on average, reflect the political leanings of the regions from which they migrated.

The patterns presented in Figure 3.3 certainly provide support for the expectation that migration from one state to another is likely to bear the imprint of the source state’s politics. If source states are consistently of one political stripe or another, and if the level of migration is of sufficient magnitude, then we should expect to see change in the destination state’s political orientation. Returning to Figure 3.2, we see the relationship between the weighted home-state liberalism of internal migrants and change in support for Democratic candidates presented in the lower right panel. This is a relatively strong relationship, with Democratic vote increasing as the birth state liberalism of internal migrants increases (r = .63). Similar to the relationship for foreign-born migrants, there is a curvilinear trend to the data: when there is relatively little bias to the home-state liberalism for internal migrants (only slightly liberal or slightly conservative), there is a lot of diversity in political changes—some states saw steep Democratic gains, while other states saw steep Democratic losses—but as the political tenor of internal migration grows more liberal, Democrats make substantial gains. Also similar to the pattern for foreign-born migrants, the lower right quadrant of the figure is completely empty, meaning that there are no cases of relatively high levels of internal migrant home-state liberalism and Democratic decline during this time period. It is also interesting to note that there are no states in which the political tendencies of (p.73)

Population Migration and Political Change

Figure 3.3 The Relationship between the Ideological Disposition of Migrants’ Childhood Region of Residence and the Contemporary Ideological Disposition of Those Regional Migrants

Note: ENC = East North Central, ESC = East South Central, MAT = Mid-Atlantic, MNT = Mountain, NE = New England, PAC = Pacific, SA = South Atlantic, WNC = West North Central, WSC = West South Central. See text for description of independent and dependent variables.

(p.74) internal migrant source states are strongly conservative, perhaps reflecting the fact that the leading sources of internal migration tend to be large, Democratic states. This figure drives home an important point: it is unreasonable to expect a substantively strong or interesting relationship between internal migration and political change without considering the political context of the states from which the migrants came. Once that context is accounted for, internal migration emerges as a potentially important source of political change in the states.

It is interesting that the weighted ideology of internal migration and foreign-born migration do not overlap very much as explanations. Overall the correlation between the percentage of foreign-born and the weighted home-state liberalism for internal migrants is just .35. A visual examination of the two scatter plots (bottom two panels in Figure 3.2) clarifies this, at least in terms of data points that are significantly off-trend, and helps illustrate the extent to which the two sources of migration may be complementary rather than competing explanations. Some states that are not well explained by foreign-born migration, such as Maine, New Hampshire, Texas, and Vermont, fit the pattern for home-state liberalism much better, and a number of states that are not particularly well explained by internal migrant home-state liberalism, such as California, Hawaii, Illinois, and New York, are more easily accounted for by the pattern of foreign-born migration.

The combined and relative impact of foreign-born and internal migration on changes in party support is assessed in Table 3.3, utilizing a multiple regression model.10 Both variables are significantly and positively related to changes in party support over time, and the relative effects of the two variables (standardized coefficients) are very similar in magnitude.11 One way to get a sense of the potential impact of these two variables is to estimate outcomes that would be expected to occur for different values of percentage of foreign-born and weighted home-state ideology. Given that the two independent variables are measured on different scales, the coefficients are a bit difficult to compare. The standardized coefficients are useful in this respect. The coefficients in the farthest right column address this issue by expressing how the dependent variable is expected (p.75)

Table 3.3 The Impact of Immigration Streams on Changes in Party Support at the State Level in Presidential Elections from 1972–80 to 2004–12


Δ‎Y, Sx

Weighted Home-State Liberalism




Percentage of Foreign-Born







  • N

  • Adj. R2

  • RMSE

  • 50

  • .59

  • 4.231

Note: The dependent variable is change in estimated Democratic support from 1972–80 to 2004–12, based on the trend in the Democratic share of the two-party vote, centered around the national two-party division (see Figures 1.3 and 1.4). Weighted home-state liberalism is measured in 2008, using Equation 3.1, and percentage of foreign-born is measured in 2008. Both independent variables are based on the CVAP. b/s.e. = slope/standard error; Δ‎Y, Sx = change in the dependent variable for a standard deviation change in the independent variable. Bold = p < .05; bold italics = p < .01 (one-tailed).

to change in response to a standard deviation change in the value of the independent variables. The values of these standardized coefficients show both variables to be of roughly equal magnitude: a standard deviation increase (decrease) in either independent variable generates approximately a 3-point increase (decrease) in the Democratic share of the centered two-party vote.

It is important to understand that at least part of the effect of migration reflects the changes in state population characteristics that flow from migration patterns. Many of the patterns found in the national survey data in Table 3.2 are manifested at the state level as well. Foreign-born as a percentage of the CVAP is fairly strongly related to change in the percentage of the population who have an advanced degree (r = .43), whose occupation is classified as professional (r = .46), and who are nonwhite (p.76) (r = .84). Internal migration (weighted by birth-state ideology) is also related to important changes in state characteristics, including change in percentage with an advanced degree (r = .67), percentage with a professional occupation (r = .44), and change in net Democratic identification and net liberal identification (r = .53 and .58, respectively). This is not to say that migration patterns produced all of these changes in state characteristics, for it is possible that the state contexts created by these demographic and political factors made some states more or less attractive destinations for certain types of migrants. In-migration of both foreign-born and internal migrants is, however, a likely contributing source of change in state demographic and political characteristics, as well as change in electoral support.


Population migration is a potentially important source of demographic and political change in the states. But not all states are affected equally by migration. Some experience high levels of migration, with upward of 60% of their population born somewhere else. These tend to be mostly small states that have experienced rapid population growth over the past forty years. Here Nevada leads the way, with more than 80% of its CVAP born elsewhere. At the same time, a number of states—mostly those with low population growth—have relatively few (less than 30%) residents born elsewhere. These differences in levels of migration set the stage for differences in the impact of migration on state politics. But differences in magnitude are not enough to determine how migration might affect change in the states.

As it turns out, the source of migration is as important as the volume of migration. When considering only the size of the migrant population, there is a relatively modest positive relationship between migration and change in Democratic presidential votes. This sort of broad treatment of migration ignores the fact that migrants are a heterogeneous group and that many of the politically important differences among migrants can (p.77) be connected to the places from which they came. First, the foreign-born population stands out as substantially different from internal migrants along important demographic and political dimensions, especially race and ethnicity, party affiliation, and political ideology. Perhaps most important for studies of political change, these differences are much more stark today than they were forty years ago. Separating the foreign-born population from internal migrants is illuminating: states with a large foreign-born population almost uniformly have large increases in support for Democratic presidential candidates. This points to an important tie-in to contemporary political debates over immigration: although there are no doubt many principled reasons why most Republican presidential candidates favor relatively strict immigration policies and most Democratic presidential candidates favor policies that include a “pathway to citizenship,” it bears pointing out that there are real political consequences to changes in the foreign-born citizen population. Relatively lenient policies that facilitate immigration and eventual citizenship are likely to result in increases in a block of voters who are much more inclined to support Democratic than Republican candidates.

It is also important not to treat internal migrants as a homogeneous monolith. When focusing just on levels of internal migration, there is no discernible difference in political outcomes between states with high and low levels. This is because states do not draw their internal migrant populations equally from the same types of states. In some states internal migrants tend to come from states that are, on balance, relatively conservative, pushing the state in a conservative direction. In other states the sources of internal migrants might be relatively liberal and push the state in a liberal direction. And in yet other states the sources of internal migrants could be politically heterogeneous and have very little net effect on state politics. Importantly the impact of internal migration is likely to be greatest when the source states are relatively liberal or conservative and the internal migrant population is substantial, relative to the state native population. When the liberalism of the birth states of internal migrants is weighted by the size of the internal migrant population, there is a strong positive relationship between internal migration and change in Democratic support over time.

(p.78) Migration patterns are an important part of the explanation of changes in state patterns of presidential support. Their importance derives not just from the direct effects shown in this chapter but also from the fact that migration can be connected to broader patterns of changes in the demographic and political makeup of state electorates, changes that contribute to a fuller account of the transformation of the geographic bases of party support in presidential elections.


(1.) Data from the American Community Survey and other Census Bureau studies were obtained from Ruggles et al. (n.d.). More details on which data files were used in this book are provided in the appendix.

(2.) This does not address the issue of those citizen voting age residents who are not eligible to vote, such as convicted felons, in some states (McDonald and Popkin 2001). Unfortunately gathering data on felon status by population subgroups over time is not feasible. Still CVAP is a better approximation of the likely electorate than simply using the voting-age population (Holbrook and Heidbreder 2010).

(3.) This overall net conservative ideological placement reflects a long-term trend in people being more willing to identify as conservative than as liberal (Clawson and Oxley 2012), even when the Democratic Party is relatively popular or when policy tastes run in the liberal direction (Stimson 2004).

(4.) Since the period averages are taken from linear predictions generated by the slopes in Figures 1.3 and 1.4, the relationships are exactly the same if the actual slopes are used as the dependent variable.

(5.) I have saved the impact of short-term migration changes on short-term political changes for another day. This book is focused on longer-term political change.

(6.) Enns and Koch (2013) use hundreds of thousands of responses from hundreds of surveys that include questions on party identification and ideology to estimate state partisanship and state political ideology. In this book I use a couple of different versions of the Enns and Koch measure, details of which are provided in the data appendix.

(7.) This variable is constructed from one that asked respondents in which state they lived when they were sixteen years old, but the GSS only provides the regional identifier.

(8.) As of this writing (spring 2015) the 2012 ANES responses for the question about where respondents grew up have not been released.

(9.) The correlation between historical regional ideological predisposition and the mean of the 7-point ideology scale for regional migrants is also a very robust .65 in the GSS data and .79 for the ANES data.

(10.) This is a simple model that focuses on just two independent variables. Other influences are considered in later analyses. One omitted variable that might occur to scholars of state politics is a regional control for southern states. As it happens, when a southern control variable is added to the model, it has no significant influence (t-score = –.45), suggesting that whatever uniquely southern pattern there is to political change is accounted for by immigration patterns.

(11.) Given that the scatter plots in Figure 3.2 reveal a somewhat curvilinear relationship, the model was also tested using quadratic terms for both variables (e.g., b1*foreign born + b2*foreign born2). Neither of the squared terms was significant, nor were the two terms jointly significant (F-ratio =.21, p =.89).