¹ Carlsson, Costa and Karlsson, Katarina, ‘Age, Cohorts, and the Generation of Generations’, American Sociological Review, XXXV (1970), 710–18CrossRefGoogle Scholar; Lansing, John B. and Mueller, Eva, The Geographic Mobility of Labor (Ann Arbor: Survey Research Center, 1967), pp. 39–47Google Scholar; and Taeuber, Carel E., Chiazze, Leonard Jr., and Haezszel, William, Migration in the United States: An Analysis of Residence History, Public Health Monograph No. 77 (Washington D.C.: US Government Printing Service, 1968).Google Scholar

² Middleton, R. and Putney, S., ‘Student Rebellion Against Parental Political Beliefs’, Social Forces, XLI (1964), 377–83Google Scholar; Jackman, Mary R., ‘Social Mobility and Attitudes Towards the Political System’, Social Forces, L (1972), 462–72CrossRefGoogle Scholar; and Niemi, Richard, ‘Political Socialization’, in Knutson, J., ed., Handbook of Political Psychology (San Francisco: Jossey-Bass, 1973).Google Scholar

³ Campbell, Angus, Converse, Philip E., Miller, Warren E. and Stokes, Donald E., The American Voter (New York: John Wiley and Sons, 1960), Chap. 16Google Scholar; Wolfinger, R. E. and Greenstein, F. I., ‘The Suburbs and Shifting Party Loyalties’, Public Opinion Quarterly, XXII (1958), 473–82.Google Scholar

⁴ Converse, Philip E., ‘On the Possibility of a Major Realignment in the South’, in Campbell, A., Converse, P. E., Miller, W. E. and Stokes, D. E., Elections and the Political Order (New York: John Wiley and Sons, 1966), Chap. 12, pp. 227–33.Google Scholar

⁵ An exception was Southern rural blacks who migrated to the urban North between the wars and after the Second World War. Black Southern migrants eventually came to reflect the pro-Democratic partisan stance of the Northern urban milieu. Their adaption, however, was attributed to a lack of any meaningful political socialization prior to migration. Since there were no pre-existing political orientations to overcome, blacks were readily mobilized by the political forces of the new environment. Campbell, et al. , The American Voter, Chap. 12, pp. 451–3Google Scholar; Marvick, Dwaine, ‘The Political Socialization of the American Negro’, The Annals, 361 (1965), 112–27.Google Scholar

⁶ Campbell, et al. , The American Voter, Chap. 16, pp. 448–9.Google Scholar

⁷ Sears, David O. and Whitney, Richard E., Political Persuasion (Morristown: General Learning Press, 1973), pp. 8–11.Google Scholar

⁸ Berelson, Bernard R., Lazarsfeld, Paul F. and McPhee, William N., Voting (Chicago: University of Chicago Press, 1954), Chap. 6, pp. 93–101Google Scholar; Miller, Warren E., ‘One-Party Politics and the Voter’, American Political Science Review, L (1956), 707–27CrossRefGoogle Scholar; Butler, David and Stokes, Donald, Political Change in Britain (New York: St Martin's Press, 1976) Chap. 6, pp. 92–6Google Scholar; Sprague, John, ‘Three Applications of Contextual Theses: Cross Section, Across Time Parameters’ (unpublished manuscript, Washington University, 1973)Google Scholar; Sprague, John, ‘Estimating a Bourdon Type Contextual Model: Some Practical and Theoretical Problems of Measurement’, Political Methodology, III (1976), 333–54Google Scholar; Wright, Gerald C. Jr., ‘Contextual Models of Electoral Behavior: The Southern Wallace Vote’. American Political Science Review, LXXI (1977), 497–508CrossRefGoogle Scholar; Boyd, Lawrence H. Jr., and Iversen, Gudmund R., Contextual Analysis: Concepts and Statistical Techniques (Belmont, Calif.: Wadsworth, 1979).Google Scholar

⁹ Berelson, et al. , Voting, Chap. 6, pp. 97–8.Google Scholar

¹⁰ Levin, M. L., ‘Social Climates and Political Socialization’, Public Opinion Quarterly, XXV (1961), 596–606CrossRefGoogle Scholar; Maccoby, E., Matthews, R. E. and Morton, A. S., ‘Youth and Political Change’, Public Opinion Quarterly, XVIII (1964), 23–9.Google Scholar

¹¹ Determining who migrates appears to be a straightforward task. On close inspection, however, there is no clear distinction between changing dwellings and migration. Classification must depend on the specific research problem under study. For a move to have the potential for altering the relevant influencing environment it must place the migrant at such a distance from the family, relatives and friends that old patterns of social contact are eliminated. This analysis uses the Bureau of the Census' definition of migration. They classify a move as a migration when at minimum an individual crosses a county boundary. This type of movement customarily puts the migrant in a completely different residential community which forces the migrant to form new relationships. See Shryock, Henry S., Population Mobility Within the United States (Chicago: Community and Family Study Center, University of Chicago, 1964)Google Scholar. Some county boundaries are situated in the middle of expanding urban areas which may consequently result in the incorrect classification of ‘short-hop’ moves as migrations. This definition may also obscure moves between communities of very different size which, while technically not migrations, do involve important changes in the social setting of the mover.

¹² Respondents were asked, ‘How long have you lived here (name of community)?’ (If less than all of life), ‘Where did you live before you moved here (city, state)?’ Since respondents were asked about their prior community, and not prior residence, the chance of reporting a ‘short-hop’ move is considerably lessened, and the number reporting changes of residence that qualify for migration is maximized. This operationalization automatically excludes persons currently in the process of migrating – those who were between residences for the few months that the study was in the field. This is not a serious problem since internal migration in the United States is less likely to take place during the fall and winter months, the period in which this study was conducted. Also eliminated from consideration are those who may have migrated in the past, but who did not move far enough in their last move to be considered migrants.

¹³ Aggregate data for both the previous and the current environments were made available by Jerome M. Clubb, Director of the ICPSR. A special thanks is due to Erik Austin of the ICPSR Historical Archives for his advice at several stages of the contextual coding. The 1970 American National Election Study utilized in this article was made available also by the ICPSR under a grant from the National Science Foundation. Neither the original collectors of the data nor the Consortium bear any responsibility for the analyses or interpretations presented here. Coding for the contextual data was made possible by a grant from the Horace H. Rackham School of Graduate Studies at The University of Michigan.

The reliance on congressional data for the construction of the previous and current political environments was necessitated by two constraints. The first was a desire to construct a uniform measure of partisan environment which matched as closely as possible the true, long-term political milieu to which the individual was exposed. Races for governorship, the Presidency and the senate were excluded from the contextual measures largely because of the presence of short-term disturbances, unacceptably long time periods between races of non-comparability across states. For instance, the time interval between Presidential elections (four years) and senatorial elections (six years) limited their usefulness for establishing a precise picture of the previous and current context. Data for other statewide election contests did not occur with sufficient regularity across all states to be used in a study based on national survey data.

In addition, there were serious operational problems. Information for the previous environmental measure had to be hand-calculated separately for each migrant from raw election returns. This requires going through congressional election returns scattered over sixty years and across county level data for fifty-two states. Given the cross-checking that was required to ensure accuracy, calculations of alternative measures of the previous environment would have far outstripped the resources on hand. Thus the current contextual measure was limited by the operational constraints on measuring the previous milieu.

¹⁴ See the Appendix for a discussion of self-selection. In addition, the exact influence process underlying individual change that may result from environmental change will not be discussed. For a critical assessment of models of influence resulting from group effects see Erbring, Lutz and Young, Alice A., ‘Individual and Social Structure: Contextual Effects and Endogenous Feed-back’ (unpublished manuscript, University of Michigan, 1977).Google Scholar

¹⁵ A country or congressional district is considered to be dominated by a political party when an average of at least 55 per cent voted for the congressional candidate in the respective election years used to compile the current or previous political environment. A Republican environment therefore has less than 45 per cent voting for the Democratic candidate while a Democratic environment is where at least 55 per cent voted for the Democratic congressional candidate. A competitive environment is defined as a county or congressional district with between 46 and 54 per cent voting for the Democratic candidate.

¹⁶ Migration within competitive environments has been excluded from the analysis because the expected individual partisan effects for such movement are ambiguous. For a discussion of the impact of highly competitive environments on individual political behaviour, see Converse, P. E., ‘The Concept of the Normal Vote’Google Scholar, in Campbell, A. et al. , Elections and the Political Order, Chap. 2Google Scholar. For a discussion of the effects of migration within competitive areas, see Brown, T. A., Changing Political Environments, 1981, forthcoming.Google Scholar

¹⁷ It should be stressed that the dissonant category includes those who have moved from dominant to competitive areas. Empirical investigations show that those who experienced such change behave just like those who migrated to an environment dominated by the opposition. The reason follows from the assumptions about contextual effects. Those who earlier had lived in a dominant political milieu and are currently in competitive setting have had the uniformity of political cues broken by migration. On the other hand, those moving from competitive to dominated political environments are not as startled by the migration because they had previously received information about, and presumably had contact with, both political parties.

¹⁸ Butler, and Stokes, , Political Change in Britain, Chap. 2, pp. 25–6Google Scholar; and Budge, Ian, Crewe, Ivor and Farlie, Dennis, Parly Identification and Beyond (London: Wiley, 1976).Google Scholar

¹⁹ There is one small difference worthy of mention. For those migrating toward a Democratic environment, the proportion voting Republican declines monotonically with the Democratic homogeneity of the previous context. This does not appear to be the case, however, in the case of those migrating toward a Republican environment. Migrants coming previously from competitive environments and moving into dominant Republican settings – a mixed experience – are as likely to support the Republican congressional candidate as migrants remaining within a Republican environment.

An explanation may be found in the fact that a third of all internal migration takes place within the South, a high percentage of which occurs within highly Democratic environments. When averaged with the overall patterns of migration and environmental change, Southern Democratic migrants probably affect the estimates for consonant migrants in Democratic areas by culturally unique factors not necessarily found in other sections of the country. Once the presence of Southern migrants is removed, the pattern for the Democratic side of the relationship is identical to that for movement within or toward a Republican environment.

²⁰ Stokes, Donald E. and Miller, Warren E., ‘Party Government and the Salience of Congress’, Public Opinion Quarterly, XXVI (1962), 531–46CrossRefGoogle Scholar; Nie, Norman, Verba, Sidney and Petrocik, John R., The Changing American Voter (Cambridge, Mass: Harvard University Press, 1976), Chap. 2, pp. 31–2.Google Scholar

²¹ Brown, Thad A., Some Individual Political Effects of Population Movement in the United States (Ph.D. dissertation, University of Michigan, 1977), Chap. 5, pp. 167–72.Google Scholar

²² Converse, P., ‘The Concept of the Normal Vote’, pp. 34–9.Google Scholar

²³ To facilitate easy interpretation of the normal vote procedure the percentage differences between the expected and the observed proportion of the Democratic vote have been presented across the categories of environmental change. Difference scores greater than zero indicate that the expected proportion voting Democratic was higher than the observed Democratic vote and that consequently the net effect within the environmental change category was to pull migrants in the direction of the Republican party's congressional candidate. Entries less than zero indicate that the observed proportion voting Democratic was greater than the expected proportion based on the normal vote estimation, and that therefore the pull of the environmental forces was to the benefit of the Democratic candidate, in that more Democratic votes were received than normally expected on the basis of the migrant group's self-identified partisanship. As with previous illustrations, the categories of environmental change have been separated according to the partisan complexion of the current environment.

²⁴ As the Appendix shows, geographical selection on the basis of party identification does not occur. However, other social and economic attributes may be contributing to the self-selection of migrants to a given area. The eventual political consequences may indeed lead to a process of self-selection rather than one of adaptation. To ensure that the posited relationship between voting behaviour and environmental forces is correct, a series of linear corrections was used to partial out the disturbing effects of socio-economic status and mobility, subjective social class, income and education prior to estimation of the normal vote differentials. Multiple regression equations were constructed with demographic and social variables as predictors of the expected vote and the actual vote. These regression equations produced estimates (bs) of the impact on the vote of specified socio-economic and demographic factors. See Theil, Henry, Principles of Econometrics (New York: Wiley, 1971) Chap. 3, pp. 111–24Google Scholar. The general model used was as follows:

where

AV = the resultant score on the adjusted variable for any given migrant;

OV = original score on the variable for any given migrant;

b_i.j.…n = regression weight for the corresponding demographic variable for which OV is being corrected;

ZX_i.j.…n = the migrant's standardized score on the ith, jth,…, nth demographic characteristic;

a = the regression constant;

= Mean of the original variable.

The differences in the resulting scores based on this adjustment technique and those reported in Figure 2 are so slight that it is safe to conclude that the non-political self-selection is of little importance to the general conclusion.

²⁵ Campbell, et al. , The American Voter, Chap. 16Google Scholar; Wolfinger, and Greenstein, , ‘The Suburbs and Shifting Party Loyalties’, pp. 473–82.Google Scholar

²⁶ The correlation between age and the length of time in the current environment is 0·7 (tau-b). To remove the effects of this age bias and the socio-economic factors that are implied by an age bias, a series of linear adjustments of the migrant's party identification scores was performed. Respondents' current age, socio-economic status and mobility, subjective social class, income, education and current region were included in a multiple regression model predicting their party identification. The scores in Figure 3 are the standardized residuals of this estimation procedure. This statistical controlling method is described in fn. 24.

²⁷ The selection of eleven years or more as the time that migrants should show adaptation to the current partisan environment is not an arbitrary selection. After a decade the individual has been exposed to five congressional elections and two presidential races, and therefore the cues of the partisan environment are more likely to be felt. Empirical research suggests that ten years is the period of time needed for the uprooting effects of migration to diminish. Nearly a decade has to pass before the individual approaches the level of life satisfaction, personal competence and political efficacy of the stable population. See Newman, Sandra, ‘The Residential Environment and the Desire to Move’ (Survey Research Center, The University of Michigan, 03 1974).Google Scholar

²⁸ Andersen, Kristi, The Creation of a Democratic Majority 1928–1936 (Chicago: University of Chicago Press, 1979)Google Scholar; Petrocik, John R., Parly Coalitions (Chicago: University of Chicago Press, 1981).Google Scholar

²⁹ Migration has been suspected before of being an important agent of socialization. Jennings and Niemi's panel study of Americans who were adolescents in 1965 and young adults in 1973 made a special effort to monitor each place a respondent lived during the intermediate eight years. See Jennings, M. Kent and Niemi, Richard G., Generations and Politics: A Panel Study of Young Adults and their Parents (Princeton, N.J.: Princeton University Press, 1981)CrossRefGoogle Scholar. Research is currently underway which attempts to isolate the role that geographical mobility has on the political attributes of young adults in the light of other agents of socialization. See Brown, , Changing Political EnvironmentsGoogle Scholar. This research attempts to separate out the effects of social mobility and geographical mobility, both of which show a similar mid-way adjustment for individuals experiencing change. Abramson, for example, shows that those experiencing upward social mobility have party identifications midway between those of the originating class and those of the class of destination. See Abramson, Paul R., ‘Intergenerational Social Mobility and Partisan Preference’, Comparative Political Studies, VI (1973), 224–31Google Scholar. Yet clear differences exist between the micro-political effects of social mobility and geographical mobility: (1) migration effects appear to be present even when social class mobility occurs, and (2) in the case of social mobility very little of the attitudinal adjustment occurs in the adult years (see Sears, D. O., ‘Political Behavior’, in Lindzey, G. and Aronson, E., eds., The Handbook of Social Psychology, Vol. 5, 2nd edn. (Reading, Mass.: Addison-Wesley, 1969), Chap. 41.Google Scholar) It is possible that the impact of social mobility on political attributes is more a product of self-selection than a response to a changed political milieu afforded to the individual by changing positions in the structure.