The Great Migration Gets Underway: A Comparison of Black Southern Migrants and Nonmigrants in the North, 1920* Stewart E. Tolnay, University of Washington Objective. This article examines the characteristics of black southern migrants in the North near the beginning of the Great Migration and compares them with northern-born African Americans. Methods. Data from the newly available 1920 Integrated Public Use Microdata Series file are used to conduct ordinary least squares regression and binary logistic regression analyses that compare migrants and native northerners on: residential characteristics, economic activity, and family patterns. Results. On the one hand, southern migrants, males and females alike, were more likely to report gainful occupations than native northerners. On the other hand, migrants experienced denser housing conditions and held lower-status jobs than indigenous northerners. No significant differences in home ownership or family patterns were found. Even the statistically significant differences between migrants and northern-born blacks were quite modest. A supplemental “generational analysis” suggests that the relatively minor disadvantages experienced by migrants in 1920 were probably due to a temporary period of adaptation and dislocation resulting from their geographic mobility. Conclusions. When combined with evidence from later stages in the Great Migration, these findings indicate that black southern migrants fared quite well in the North, relative to native northerners. Thus, the generally negative descriptions of migrants by contemporary observers, and some later researchers, should be viewed skeptically. In many respects the Great Migration of African Americans from the South to the North and West1 was the most dramatic population movement *Direct all correspondence to Stewart E. Tolnay, Department of Sociology, University of Washington, Seattle, WA 98195-3340. The research on which this article is based was supported by grants to the author from the National Science Foundation (SBR-9529308) and from the National Institute of Child Health and Human Development (R01 HD34363). Earlier stages of this project were conducted at the University at Albany—SUNY, where support was provided by grants to the Center for Social and Demographic Analysis from the National Science Foundation (SBR-9512290) and the National Institute of Child Health and Human Development (P30 HD32041). The 1920 Public Use Microdata Sample data used in this analysis are available from the Integrated Public Use Microdata Series Project at the University of Minnesota <www.ipums.umn.edu>. I thank Patty Glynn for programming support. Robert M. Adelman and the anonymous Social Science Quarterly reviewers provided valuable comments on earlier versions of this article. 1 For simplicity, throughout the text I will use “North” and “northern” to refer to all locations outside of the census-defined South. SOCIAL SCIENCE QUARTERLY, Volume 82, Number 2, June 2001 ©2001 by the Southwestern Social Science Association 236 Social Science Quarterly to occur within the United States during the 20th century. It resulted in a fundamental demographic shift in the black population between 1910 and 1970, erasing its lopsided concentration in southern states and converting a primarily rural population into a largely urban one (Johnson and Campbell, 1981; Henri, 1975). In addition, the Great Migration had profound effects on the southern communities that the migrants left, as well as on the northern communities that they joined (see, e.g., Lemann, 1991; Marks, 1989; Massey and Denton, 1993; Tolnay and Beck, 1992; Wilson, 1987). Finding their new locations something less than the promised land, some migrants soon returned to the South (Bunch-Lyons, 1997; Long and Hansen, 1975; Marks, 1989). Many more, however, remained and attempted to establish themselves, socially and economically, in the relatively foreign environment of the northern industrial city (Drake and Cayton, [1945]1962; Fligstein, 1981; Gottlieb, 1987; Grossman, 1989; Henri, 1975; Lieberson, 1980; Trotter, 1985). Observers of the earlier stages of the Great Migration often emphasized the migrants’ negative experiences in, and impacts on, northern society (e.g., Drake and Cayton, [1945]1962; Frazier, 1932, [1939]1966; Johnson, 1922; Mossell, 1921). Viewed with suspicion by many black and white native northerners alike, the migrants were often blamed for a variety of social ills that were perceived to be intensifying in northern cities, including crime, sexual promiscuity, family instability, and residential deterioration. In contrast, studies based primarily on census data for the later stages of the Great Migration have reached a somewhat different conclusion about the migrants’ experiences in the North. For example, studies focusing on the 1960s and 1970s show that, despite their lower levels of educational attainment, the migrants were more likely than native northerners to be employed or actively seeking work and less likely to be on public assistance (Lieberson, 1978a; Lieberson and Wilkinson, 1976; Long, 1974; Long and Heltman, 1975). Furthermore, the migrants experienced less marital disruption, and the children of southern-born mothers in the North were more likely to live with two parents (Lieberson and Wilkinson, 1976; Tolnay 1997, 1998a; Tolnay and Crowder, 1999). In this article, I use the newly available 1920 Public Use Microdata Sample (PUMS) to provide an overview of the status of southern migrants relative to that of their northern-born neighbors. I consider three different realms of the southern migrants’ experiences in the North: housing characteristics, economic status, and family patterns. The evidence yielded by these comparisons provides the first systematic, quantitative portrait of southern migrants during the earliest stage of the Great Migration. Southern Migrants in the North African Americans first began to leave the South in significant numbers during the second decade of the 20th century. Between 1910 and 1920, The Great Migration Gets Underway 237 alone, roughly one-half million African Americans left the South (Alexander, 1998) and relocated primarily to urban areas in the Northeast and North Central regions. These newcomers to northern cities often were portrayed in unflattering terms by contemporary observers. Although the migrants were positively selected for education from the general southern population (e.g., Lieberson, 1978b; Tolnay, 1998b), and many had actually moved from southern towns and cities (e.g., Alexander, 1998; Marks, 1989), the “typical” migrant was frequently described as an unsophisticated, illiterate sharecropper with little or no experience at urban living. Northern African Americans could be as critical as whites in their assessment of these new neighbors from the South. The migrants’ strange dress, accents, food, and habits were considered uncouth and embarrassing by many blacks with deeper roots in the North. And their growing numbers were sometimes viewed as a potential threat to the racial status quo that offered northern blacks a relatively comfortable coexistence with whites, if not actual racial equality (Drake and Cayton, [1945]1962; Frazier, 1932; Mossell, 1921). These concerns about the early southern migrants are also evident in the writing of social scientists who studied migrant communities during the first half of the 20th century. Even before the Great Migration had begun, W. E. B. DuBois ([1899]1990:67) referred to the legacy of slavery as a challenge for southern migrants to Philadelphia when he wrote, “The lax moral habits of the slave régime still show themselves in a large amount of cohabitation without marriage.” Based on her study of 100 southern migrants who arrived in Philadelphia in 1917 and 1918, Sadie Farmer Mossell (1921:216) reached the uncharitable conclusion that, “With few exceptions the migrants were untrained, often illiterate, and generally devoid of culture.” According to Drake and Cayton, southern migrants to Chicago during the Great Depression continuously replenished the city’s black lower class ([1945]1962:717), which engaged in a perennial struggle against unemployment, poverty, and family disintegration. Charles Johnson (1922) also considered the southern backgrounds of the migrants to be a handicap as they tried to adapt to life in the northern city. Although many early observers and scholars tended to focus on the negative characteristics of southern migrants, others have offered a more balanced view by describing the migrants’ initiative and contributions to northern society (see, e.g., Meier and Rudwick, 1976:252). Indeed, southern migrants and their children achieved success and prominence in virtually every realm of northern society from business to education to the arts. Still, contemporary descriptions of the southern migrants were more likely to emphasize the problems they presented for northern society, rather than their many contributions to it. Early scholars of the Great Migration, such as DuBois, Mossell, Johnson, and Drake and Cayton, used primarily anecdotal evidence, along with limited survey data, to support their descriptions of southern migrants in the North. They did not have access to the kinds of census data that were used 238 Social Science Quarterly by subsequent researchers to compare the relative statuses of southern migrants and native northerners during the post–World War II period. Those later comparisons suggest that the migrants actually were doing quite well, socially and economically, during the final stages of the Great Migration. For example, male migrants were more likely than their northern-born counterparts to be in the labor force and employed (Lieberson, 1978a; Lieberson and Wilkinson, 1976; Long and Heltman, 1975). As a result of these economic advantages, migrant families were also less likely to live in poverty or to be on welfare (Long, 1974). In addition, southern migrants enjoyed greater family stability than native northerners with less marital disruption (Lieberson and Wilkinson, 1976; Tolnay, 1997, 1998a) and more children living with two parents (Tolnay, 1997, 1998a; Tolnay and Crowder, 1999). The sharp contrast between these earlier and later portraits of southern migrants is intriguing and suggests two potential explanations. First, perhaps the descriptions of migrants from the earlier stages of the Great Migration, which were based largely on anecdotal evidence for a few northern cities, focused too narrowly on the negative experiences and impacts of migrants while deemphasizing their many contributions and successes. If so, then a more objective analysis, using census data similar to those available to researchers investigating more recent time periods, should yield a more favorable portrait of the early southern migrants. Second, perhaps the experience of migrants changed as the Great Migration progressed, with later migrants adjusting more successfully than earlier migrants to life in the North. If accurate, this explanation suggests that a census-based description of migrants in 1920 will show them to compare unfavorably with indigenous northerners and to be more seriously disadvantaged than later migrants. In the analyses that follow I examine the characteristics of southern migrants at the very beginning of the Great Migration. My objective is to provide an overview of the early migrant experience by comparing southern migrants and native northerners on a variety of attributes, including residential characteristics, economic status, and family patterns. These three general attributes represent characteristics that were used by early observers to present generally unfavorable descriptions of southern migrants (Drake and Cayton, [1945]1962; DuBois, [1899]1990; Frazier, 1932; Johnson, 1922; Mossell, 1921) and have been used by more recent researchers to compare migrants and native northerners near the end of the Great Migration (e.g., Lieberson, 1978a; Lieberson and Wilkinson, 1976; Long, 1974; Long and Heltman, 1975; Tolnay, 1997, 1998a; Tolnay and Crowder, 1999; Tolnay, Crowder, and Adelman, 2000). The Great Migration Gets Underway 239 Data, Variables, and Method Data To compare the characteristics of southern migrants and native northerners at the onset of the Great Migration I use the newly available 1920 PUMS, a 1%, nationally representative sample of all households in the United States created by the Integrated Public Use Microdata Series Project at the University of Minnesota (Ruggles and Sobek, 1998). From the larger sample I have selected African Americans residing in nonfarm dwellings in urban areas of nonsouthern states in 1920. I further restrict the study sample when necessary to make it appropriate for the specific characteristic being examined. For example, when studying economic status I restrict the sample to civilian, noninstitutionalized males or females 18–64 years of age. When analyzing family patterns I focus on children between the ages of 0 and 14 who were living with their mothers. My examination of residential characteristics is based on African American heads of households. The 1920 PUMS does have some limitations with implications for this study. First, the 1920 census includes limited information with which to classify individuals by migration history. Unlike most later censuses (e.g., 1940, and 1960 through 1990), the 1920 census did not include a question that asked individuals where they resided five years in the past. As a result, in this study southern migrants are identified by using their reported state of birth. Second, the 1920 census does not offer a wide variety of measures that can be used to compare the social and economic characteristics of migrants and native northerners. For example, it includes no measures of educational attainment, employment status, income, or wealth. Despite these limitations, the 1920 PUMS represents the most powerful and versatile source of data for studying the characteristics of southern migrants during the earliest stage of the Great Migration. Variables Five dependent variables are used to compare the characteristics of southern migrants and northern-born African Americans in the North. Comparisons on residential characteristics are made by examining group differences in density of housing and home ownership. Housing density is measured as the number of households included within a single dwelling. Home ownership is a dichotomy scored 1 if a dwelling was occupied by the owner and 0 if occupied by a renter. Differences in economic activity are examined by comparing the two groups on labor force participation and occupational status. Labor force participation for adult males and females is coded 1 for those individuals who reported any gainful occupation and 0 for those who did not. The Duncan Socioeconomic Index (SEI) is used to 240 Social Science Quarterly represent the occupational status of those adults who reported a gainful occupation (Duncan, 1961; Ruggles and Sobek, 1998). Higher values on the Duncan SEI correspond to higher occupational prestige. Finally, differences in family patterns are considered by examining the living arrangements of young children. Father’s presence is measured for those children 0–14 years who were living with their mothers. Children coresiding with a father are coded 1, and those living with their mother only are coded 0. For the analysis of father’s presence, southern origin is based on the mother’s region of birth. The independent variable of primary interest distinguishes southern migrants from native northerners. This dummy variable, southern born, is coded 1 for those individuals (mothers in the analysis of father’s presence) who reported a southern state as their birthplace and 0 for those who reported a northern birth state.2 Selected control variables are also included in the analysis of each of the five dependent variables in order to avoid drawing inferences about the differences between southern migrants and native northerners that are affected by spurious or suppressed relationships. In order to control for variation in opportunity structures for African Americans by region and size of place, all analyses include two sets of dummy variables that describe the location of individuals and households. The first set distinguishes individuals by the size of the urban area in which they resided. Small city identifies those living in cities of less than 25,000 population and medium city represents residents of cities with populations between 25,000 and 99,999. Those living in large cities (i.e., those with populations of 100,000 or more) serve as the reference category for city size. The second set distinguishes cases by their regional location within the North—Midwest or West (with Northeast as the reference). In addition, the analyses of labor force participation and the Duncan SEI include controls for the individual’s age (and age-squared for labor force participation), literacy, and marital status (married, spouse present versus all others). The analyses of home ownership and housing density include controls for the household head’s age, gender, literacy, and labor force participation. The analysis of children’s living arrangements also includes controls for the child’s age, mother’s age, and mother’s literacy. The Appendix reports descriptive statistics for all variables used in the analyses, separately for southern migrants and native northerners. 2 This definition lumps together migrants from all areas of the South and therefore does not consider the possible effects of intraregional variation in economic organization, culture, and opportunity structures for African Americans on postmigrant adjustment in the North. A thorough consideration of the experiences of migrants from different subregions of the South is reserved for future research. The Great Migration Gets Underway 241 Method Binary logistic regression techniques are used to analyze those dependent variables that are measured as dichotomies (home ownership, labor force participation, father’s presence). Ordinary least squares regression techniques are used for dependent variables that are interval in nature (number of households in dwelling, Duncan SEI). Two models are estimated for all dependent variables. The first model includes only the variable southern born as a predictor in order to determine the “gross” differences between southern migrants and native northerners on the five dependent variables. The second model introduces the appropriate control variables in an effort to describe “net” group differences. The analysis of children’s living arrangements is weighted by the inverse of the number of children in the family in order to prevent large families from having a disproportionate influence on the results. Findings Residential Characteristics One of the first challenges faced by southern migrants upon their arrival in the North, even before looking for a job, was finding a place to live. Many moved in with friends or relatives who had migrated earlier. Others moved into small apartments or “kitchenettes” that offered the bare necessities. Some lived as boarders or lodgers in the homes of strangers. The evidence in Table 1 supports the common description of southern migrants settling into more crowded housing situations once they reached their northern destinations. Model 1 reveals that migrants resided in dwellings containing significantly more households than the “typical” dwelling occupied by northern-born blacks (β = +.871, p < .001). Although the migrant disadvantage is attenuated somewhat by the introduction of control variables, it remains highly significant in Model 2 (β = +.609, p < .001). A better idea of the level of housing density, as well as the net difference between migrants and native northerners, can be gained by estimating predicted values of the number of households per dwelling unit for each group while fixing all control variables at their respective mean values. This heuristic exercise shows that, when equated on all other characteristics included in the model, southern migrants resided in dwellings with an average of 2.8 households, whereas northern-born blacks resided in dwellings that contained an average of 2.2 households. The denser housing arrangements for southern migrants suggests that they were also more likely to be renters than owners. Indeed, the results in Model 3 of Table 1 show that the southern-born were significantly less likely than native northerners to reside in owner-occupied dwellings (β = − .385, p < .001). Although the evidence in Model 4 continues to indi- 242 Social Science Quarterly TABLE 1 OLS Coefficients from Regression of Housing Density and Logit Coefficients from Regression of Home Ownership on Predictor Variables: Households Headed by Blacks in the Urban North, 1920 Housing Density Model 1 Southern born .871*** (.190) Age Female Literate In labor force Place smaller than 25,000 Place 25,000– 99,999 Midwest West Intercept Adjusted or pseudo-R2 N 2.033*** (.159) .007 Home Ownership Model 2 Model 3 Model 4 .609*** (.188) –.019** (.007) .087 (.234) .388 (.293) –.176 (.350) –1.669 (.228)*** –1.738*** (.242) –1.435*** (.177) –2.061*** (.388) 4.255*** (.627) .066 –.385*** (.096) –.128 (.104) .044*** (.004) –.108 (.134) .730*** (.177) –.303 (.185) 1.292*** (.115) .573*** (.134) .414*** (.105) .825*** (.198) –4.254*** (.369) .115 2,944 –1.118*** (.077) .005 3,001 NOTE: Standard errors in parentheses. *p < .05, ** p < .01, ***p < .001 (two-tailed tests). cate a lower probability of home ownership among migrants, the differential is reduced to nonsignificance by the inclusion of control variables (β = −.128, p = .223). Supplementary analyses revealed that controlling for city size was especially important in reducing to nonsignificance the difference in home ownership between migrants and native northerners. On the one hand, as might be expected for those moving to a new location, southern migrants experienced slightly greater density in housing conditions than did indigenous northerners. On the other hand, when appropriate control variables are considered, migrants were no less likely than native northerners to be home owners. The relatively small difference in density of housing between the migrants and nonmigrants, coupled with the nonsignificant difference in home ownership, suggests that the two groups were quite similar in their housing characteristics. That is, the “aver- The Great Migration Gets Underway 243 age” housing situation for both groups consisted of a rented unit in a dwelling that contained more than one household. Economic Activity To compare the economic activity of southern migrants and indigenous northerners, I consider first the labor force activity of adult males. Despite the fact that an overwhelming percentage of all African American males in the North reported a gainful occupation in 1920, Model 1 of Table 2 reveals a significant advantage for southern migrants, when no other characteristics are controlled (β = +.777, p < .001). Though the differential is attenuated when controls are introduced in Model 2, the significantly higher level of labor force activity among migrants persists (β = +.596, p < .01). Once again, predicted probabilities can be used to illustrate the magnitude of the net differential revealed in Model 2. When all control variables are held constant at their means, the predicted percentages of males reporting gainful employment are 98% for southern migrants and 97% for northern-born males. The findings in Table 2 also indicate that the slight migrant advantage in labor force activity did not carry over to occupational prestige. Model 3 reveals that migrants occupied a significantly lower rung in the occupational hierarchy than did northern-born males (β = −1.485, p < .001). Furthermore, Model 4 demonstrates that their lower occupational prestige cannot be explained by the control variables. Indeed, the coefficient for southern birth increases somewhat when controls are introduced (β = −1.813, p < .001). Again, although the two groups differ significantly in average SEI scores, predicted values obtained by fixing the control variables at their means reveal that the difference is not large. Southern migrants have a predicted SEI score of 14.2, compared with 16.0 for native northerners. The evidence for the economic activity of males in 1920 is generally consistent with findings from the later stages of the Great Migration. That is, at both time periods, southern migrants were characterized by higher levels of labor force activity, but lower occupational status, than were males born in the North (Lieberson and Wilkinson, 1976; Long and Heltman, 1975). These results describe a male migrant population that was fully engaged in the northern economy but entered the labor market at a level slightly below that occupied by their northern-born neighbors. In contrast to the findings for males, Model 5 in Table 2 shows that the bivariate relationship between migrant status and labor force participation for females is not statistically significant at the p < .05 level (β = +.109, p = .095). However, when controls are included (Model 6) a significant migrant advantage emerges for females as well (β = +.169, p < .05). Further investigation of the findings for females suggests that marital status functioned as a suppressor variable in Model 5, attenuating the effect of south- 244 Social Science Quarterly TABLE 2 Logit Coefficients from Regression of Labor Force Participation and OLS Coefficients from Regression of Duncan SEI on Predictor Variables: Black Men and Women in the Urban North, 1920 Men Southern born Age Age2 Literate Married, spouse present Place smaller than 25,000 Place 25,000– 99,999 Midwest West Intercept Adjusted or Pseudo-R2 N Women Labor Force Participation Duncan SEI Labor Force Participation Duncan SEI Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 .777*** .596** –1.485*** –1.813*** (.187) (.194) (.465) (.469) .235*** .102*** (.048) (.020) –.003*** — (.001) — .433 4.084*** (.370) (.894) 1.211*** 1.596*** (.239) (.448) .109 (.065) .169* –2.895*** –2.774*** (.076) (.586) (.593) .157*** –.022 (.021) (.026) –.002*** — (.000) — –.081 2.402* (.149) (1.166) –2.165*** –1.110* (.077) (.572) –.147 (.246) –2.125*** (.610) –.276** (.100) –2.153** (.816) –.099 (.265) –.926 (.618) .099 (.102) –.233 (.783) .150 (.193) .514 (.529) 3.129*** –1.741 (.133) (.908) –1.375** (.443) .091 (1.047) 15.825*** 8.931*** (.385) (1.292) .016 .086 4,399 .002 .021 4,281 –.315*** 2.110*** (.072) (.567) –.285 1.287 (.171) (1.297) –.164** –1.177** 16.508*** 14.630*** (.053) (.390) (.481) (1.718) .000 .172 4,212 .012 .022 2,009 NOTE: Standard errors in parentheses. *p < .05, **p < .01, ***p < .001 (two-tailed tests). ern origin on labor force activity. This suppression is due to two factors: Migrant women were more likely to be married and living with a spouse than were northern-born women (60.3% and 56.3%, respectively), and married women residing with a spouse were less likely to be employed, as indicated by the coefficient for marital status in Model 6 (β = −2.165, p < .001). Predicted values for female labor force activity, when the control variables are fixed at their mean values, reveal a larger differential by migration status than was observed for males: 45% for northern-born women versus 49% for southern-born women. This evidence of a higher level of labor force in- The Great Migration Gets Underway 245 volvement for female southern migrants in 1920 differs from the finding by Lieberson and Wilkinson (1976) for 1960 that levels of employment and labor force participation were lower for female migrants than for native northern women, when education and marital status are controlled for. Like their male counterparts in the labor force, female southern migrants were engaged in less prestigious occupations than were northern-born women. When no other characteristics are taken into account (Table 2, Model 7), migrant women averaged SEI scores that were roughly three points lower than those for native northerners (β = −2.895, p < .001). Furthermore, the differential is virtually unchanged when control variables are added to the equation (β = −2.774, p < .001). When predicted values for the Duncan SEI score are derived, averages of 13.7 and 16.4 are obtained for female southern migrants and native northerners, respectively. Family Patterns Observers of the early stages of the Great Migration often commented on the greater instability of migrant families and attributed it to the transplantation of a rural southern family pattern that originated during slavery (Drake and Cayton, [1945]1962; DuBois, [1899]1990; Frazier, 1932, [1939]1966). Although studies of migrant family structure during the post– World War II period show that migrant families were actually more stable than those of indigenous northerners (Lieberson and Wilkinson, 1976; Tolnay, 1997, 1998a; Tolnay and Crowder, 1999), this issue has not been examined empirically for earlier years. The results reported in Table 3 compare the children of migrant mothers with the children of northern-born mothers on their propensity to coreside with a father. Although the children of migrants are somewhat more likely to live with a father, neither the bivariate (β = +.069, p = .55) nor the partial (β = +.120, p = .32) relationship is statistically significant. Thus, the “migrant advantage” in family structure observed during the post–World War II period had not yet emerged by 1920. Conversely, contrary to the frequent reference to a dysfunctional family culture among southern migrants, there is no evidence that migrant families were any less stable than nonmigrant families as the Great Migration got underway. Generational Patterns of Migrant Disadvantage An important limitation of census data for examining the experiences of migrant adaptation is the difficulty, in some cases the impossibility, of measuring the migrants’ duration of residence in their new locations. This 246 Social Science Quarterly TABLE 3 Logit Coefficients from Regression of Father’s Presence in the Home on Predictor Variables: Black Children 0 to 14 Years Old in the Urban North, 1920 Father’s Presence Model 1 Model 2 .069 (.116) .120 (.122) –.029*** (.007) –.046** (.015) –.105 (.229) .069 (.149) –.105 (.158) –.224 (.119) –1.057*** (.249) 3.028*** (.357) .036 Southern born Mother’s age Child’s age Mother’s literacy Place smaller than 25,000 Place 25,000–99,999 Midwest West Intercept Pseudo-R2 N 1.473*** (.094) .000 2,216 NOTE: Standard errors in parentheses. *p < .05, **p < .01, ***p < .001 (two-tailed tests). limitation inhibits our ability to determine the extent to which any observed “migrant disadvantage” is due to (1) something problematic about the characteristics or culture of migrants or (2) the naturally disruptive, but temporary, period of adjustment faced by all migrant populations. As noted above, unlike later censuses (1940, 1960–1990), the 1920 PUMS does not include a question about an individual’s place of residence five years in the past—an imperfect but useful measure of recency of migration. It is possible, however, to use information about the state of birth for an individual and his or her parents to classify northern African Americans into three groups, using categories similar to those used to describe the “generation” of immigrants from abroad. Following this strategy I have created the following three categories of northern residents: First generation southern migrants are those who were born in the South (this is the same group that is identified by the dummy variable “southern born” in Tables 1 through 3); second generation migrants were northern-born but had at least one parent who reported a southern state of birth; and the third generation were born in the North to northern-born parents. The Great Migration Gets Underway 247 Using these (admittedly crude) measures of duration of residence in the North, I reexamine the three characteristics for which southern migrants were found to have a statistically significant disadvantage vis-à-vis native northerners. If the migrant disadvantage reported for these characteristics in earlier tables was primarily due to a temporary adjustment period as firstgeneration migrants settled into their new northern locations, then we should find that second-generation migrants did not experience the same disadvantage. Furthermore, if second-generation migrants were not significantly different from those of the third generation, or if they enjoyed an advantage over the third generation on these characteristics, then it seems unlikely that something inherently problematic about the culture of southern migrants played an important role in the disadvantages experienced by the first generation. A pattern of declining disadvantage between the first and second generations would be more ambiguous, since it could be consistent with both potential explanations. TABLE 4 Coefficients (Logit or Ordinary Least Squares) Describing Generational Differences on Selected Characteristics: Urban North, 1920 Housing Density Third generation Second generation First generation — .683* (.325) .838*** (.218) Male SEI Female SEI — 1.490 (.799) –1.283* (.548) — 1.491 (.977) –2.174** (.711) NOTE: Standard errors in parentheses. The models used to generate these coefficients include the same control variables reported in Tables 1 and 2. *p < .05, **p < .01, ***p < .001 (two-tailed tests). Table 4 reports the coefficients for dummy variables representing firstand second-generation migrants, with the third generation serving as the reference category. Although the same sets of control variables used in earlier models (see Tables 1 and 3) were included in the equations represented in Table 4, the coefficients for those variables are not reported in order to simplify the presentation of results. For two of the three characteristics analyzed in Table 4 second-generation migrants actually enjoyed an advantage over third-generation northerners. Second-generation male and female workers had slightly higher occupational prestige than their third-generation counterparts. Although neither difference is statistically significant at p < .05, the second-generation advantage for males is significant at p = .06, and the coefficient for females is 1.5 times its standard error. In contrast, on only one of the three characteristics do second-generation migrants compare unfavorably with those of the third generation: They experienced greater 248 Social Science Quarterly housing density (β = +.683, p < .05). In sum, the evidence in Table 4 suggests that, in two of the three cases, the disadvantage experienced by firstgeneration migrants was erased or reversed by the second generation. To be sure, the results from this analysis of generational differences among migrants must be treated as only suggestive and far from definitive. However, when considered in conjunction with all of the other findings presented in Tables 1 through 3, the evidence in Table 4 does point toward the strong possibility that any disadvantages experienced by southern migrants may have resulted from their shorter duration of residence in the North and to the disruptive effects of their adjustment process. Discussion The story of the experience of black southern migrants in the North has been a contradictory one. On the one hand, contemporary descriptions of the migrants, especially prior to World War II, were often unflattering, ascribing many negative characteristics to the migrants and blaming them for a variety of social problems in the North. On the other hand, some contemporary descriptions, especially more objective comparisons of southern migrants and indigenous northerners during the post–World War II period, have offered a more favorable portrait of the migrants and have documented a variety of ways in which the migrants were actually better off than their northern-born neighbors. Only recently have the data required for a more systematic investigation of migrant characteristics during the earlier stages of the Great Migration become available. The evidence yielded by my analysis of the 1920 PUMS data does not describe a migrant population with serious social or economic deficiencies, at least when compared with native northerners. Despite having levels of illiteracy that were roughly twice those of the northern-born black population, southern migrants, both males and females, were more likely to report gainful occupations. Furthermore, no significant differences between migrants and native northerners were observed for home ownership or the likelihood that children lived with their fathers. Even on those characteristics for which southern migrants compared unfavorably with native northerners, the magnitude of most differences was not great. For example, differences in average occupational prestige of 1.8 points for males and 2.8 points for females do not suggest two populations occupying significantly different locations in the labor market. Likewise, the difference in housing density of 2.8 households per dwelling for migrants and 2.2 households per dwelling for native northerners does not describe two populations facing dramatically different residential opportunity structures. Rather, a more reasonable conclusion to draw from the evidence presented here is that, among African Americans in northern urban areas at the beginning of the Great Migration, southern migrants and native northerners occupied generally The Great Migration Gets Underway 249 similar positions in the economy and social structure. To the extent that differences existed, they were probably minor and/or fleeting. Two important caveats are in order regarding the 1920 census data on which my analyses are based. First, census data are not useful for assessing the validity of some criticisms that were leveled against the southern migrants, for example, those regarding their greater criminality and promiscuity. Second, the reported contrasts between southern migrants and native northerners might be influenced by selection processes not detectable in the PUMS data. For example, it is possible that the poorest segments of the migrant population were more likely to be undercounted by census enumerators because of their greater mobility. In addition, migrants who experienced failure or disappointment in the North may have returned to the South, leaving behind a “positively” selected population of migrants (e.g., Lieberson, 1978a; Lieberson and Wilkinson, 1976; Long and Heltman, 1975). Although these concerns should not be ignored, it is reassuring that the evidence from the 1920 PUMS is largely consistent with similar evidence from later years that also describes, on average, a generally successful and well-adapted population of southern migrants in the North. Like most migrant populations, their relocation did require a period of adjustment to their new communities and resulted in some temporary dislocation. But the real barrier to the ultimate success of southern migrants in northern society was not their regional origin or their history of geographic mobility. Rather, it was the same barrier that obstructed the progress of their northern-born African American neighbors: their race. REFERENCES Alexander, J. Trent. 1998. “The Great Migration in Comparative Perspective.” Social Science History 22:349–76. Bunch-Lyons, Beverly. 1997. “‘No Promised Land’: Oral Histories of African-American Women in Cincinnati, Ohio.” OAH Magazine of History 11:9–14. Drake, St. Clair, and Horace R. Cayton. [1945]1962. Black Metropolis: A Study of Negro Life in a Northern City. New York: Harcourt, Brace. DuBois, W. E. B. [1899]1990. The Philadelphia Negro. Millwood, NY: Kraus-Thomson Organization. Duncan, Otis Dudley. 1961. “A Socioeconomic Index for All Occupations.” Pp. 108–34 in A. Reiss (ed.), Occupations and Social Status. New York: Free. Fligstein, Neil. 1981. Going North. New York: Academic. Frazier, E. Franklin. 1932. The Negro Family in Chicago. Chicago: University of Chicago Press. ———. [1939]1966. The Negro in the United States. Chicago: University of Chicago Press. Gottlieb, Peter. 1987. Making Their Own Way. Southern Blacks’ Migration to Pittsburgh, 1916–30. Urbana: University of Illinois Press. 250 Social Science Quarterly Grossman, James R. 1989. Land of Hope: Chicago, Black Southerners, and the Great Migration. Chicago: University of Chicago Press. Henri, Florette. 1975. Black Migration: Movement North, 1900–1920. Garden City, N.Y.: Anchor. Johnson, Charles S. 1922. The Negro in Chicago: A Study of Race Relations and a Race Riot. Study sponsored by Chicago Commission on Race Relations. Chicago: University of Chicago Press. Johnson, Daniel M., and Rex R. Campbell. 1981. Black Migration in America: A Social Demographic History. Durham, N.C.: Duke University Press. Lemann, Nicholas. 1991. The Promised Land: The Great Migration and How It Changed America. New York: Knopf. Lieberson, Stanley. 1978a. “A Reconsideration of the Income Differences Found Between Migrants and Northern-Born Blacks.” American Journal of Sociology 83:940–66. ———. 1978b. “Selective Migration from the South: A Historical View.” Pp. 119–41 in Frank D. Bean and W. Parker Frisbie (eds.), The Demography of Racial and Ethnic Groups. New York: Academic. ———. 1980. A Piece of the Pie. Blacks and Immigrants since 1880. Berkeley and Los Angeles: University of California Press. Lieberson, Stanley, and Christy A. Wilkinson. 1976. “A Comparison between Northern and Southern Blacks Residing in the North.” Demography 13:199–224. Long, Larry H. 1974. “Poverty Status and Receipt of Welfare among Migrants and Nonmigrants in Large Cities.” American Sociological Review 39:46–56. Long, Larry H., and Kristen A. Hansen. 1975. “Trends in Return Migration to the South.” Demography 12:601–14. Long, Larry, and Lynne R. Heltman. 1975. “Migration and Income Differences between Black and White Men in the North.” American Journal of Sociology 80:1391–1409. Marks, Carole. 1989. Farewell—We’re Good and Gone: The Great Black Migration. Bloomington: Indiana University Press. Massey, Douglas S., and Nancy A. Denton. 1993. American Apartheid. Cambridge: Harvard University Press. Meier, August, and Elliott Rudwick. 1976. From Plantation to Ghetto. New York: Hill and Wang. Mossell, Sadie Farmer. 1921. “The Standard of Living among One Hundred Negro Migrant Families in Philadelphia.” Annals of the American Academy of Political and Social Science 97:169–222. Ruggles, Steven, and Matthew Sobek. 1998. User’s Guide: Integrated Public Use Microdata Series: Version 2.0. Minneapolis, Minn.: Department of History, University of Minnesota. Tolnay, Stewart E. 1997. “The Great Migration and Changes in the Northern Black Family, 1940 to 1990.” Social Forces 75: 1213–38. ———. 1998a. “Migration Experience and Family Patterns in the ‘Promised Land.’” Journal of Family History 23: 68–89. ———. 1998b. “Educational Selection in the Migration of Southern Blacks, 1880–1990.” Social Forces 77:487–514. Tolnay, Stewart E., and E. M. Beck. 1992. “Racial Violence and Black Migration in the American South, 1910 to 1930.” American Sociological Review 57:103–16. The Great Migration Gets Underway 251 Tolnay, Stewart E., and Kyle D. Crowder. 1999. “Regional Origin and Family Stability in Northern Cities: The Role of Context.” American Sociological Review 64:97–112. Tolnay, Stewart E., Kyle D. Crowder, and Robert M. Adelman. 2000. “‘Narrow and Filthy Alleys of the City?’: The Residential Settlement Patterns of Black Southern Migrants to the North.” Social Forces 78:989–1015. Trotter, Joe William Jr. 1985. Black Milwaukee: The Making of an Urban Industrial Proletariat, 1915–45. Urbana: University of Illinois Press. Wilson, William Julius. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: University of Chicago Press. 252 Social Science Quarterly APPENDIX Descriptive Statistics for All Variables Used in Analysis by Sample Used and Region of Birth Household Heads Mean SD Males Mean SD Females Mean SD Children 0–14a Mean SD Northern Born Owns home .246* .431 Housing density 2.033* 3.691 In labor force .882* .323 SEI 15.071 16.264 Father present — — Age 42.783* 13.398 Mother’s age — — Female .239* .426 Spouse present .644* .479 Literate .935* .246 Place less than 25,000 .256* .437 Place 25,000–99,999 .158 .365 Midwest .583* .493 West .049 .216 Number of cases 914 .258* .438 1.973* 4.104 .937* .242 15.044* 15.308 — — 34.578* 11.437 — — — — .468* .499 .966* .181 .273* .446 2.238* 4.749 .455 .498 7.515* 12.858 — — 34.129 11.278 — — — — .562* .496 .954* .210 .226* .418 .157 .364 .563* .496 .049 .216 1,421 .230* .421 .155 .362 .566* .496 .048 .215 1,423 — — — — — — — — .814 .385 6.278* 4.364 31.859* 8.931 .507 .495 — — .954* .207 .295* .159 .584* .035 761 .451 .362 .488 .182 Southern Born Owns home .182 .386 Housing density 2.904 5.140 In labor force .937 .243 SEI 14.149 14.329 Father present — — Age 41.401 11.985 Mother’s age — — Female .182 .386 Spouse present .739 .440 Literate .886 .318 Place less than 25,000 .151 .358 Place 25,000–99,999 .146 .353 Midwest .517 .500 West .057 .233 Number of Cases 2,088 .179 .384 2.869 5.734 .967 .178 13.995 13.459 — — 35.649 10.856 — — — — .578 .494 .920 .272 .199 .399 3.080 6.230 .485 .500 6.611 10.243 — — 34.753 11.005 — — — — .608 .488 .927 .261 .129 .335 .143 .350 .531 .499 .053 .224 3,007 .121 .326 .136 .343 .485 .500 .047 .211 2,752 — — — — .824 6.853 32.925 .533 — .930 — — — — .383 4.407 8.914 .502 — .256 .142 .351 .141 .350 .504 .503 .038 .193 1,455 NOTE: Number of cases varies slightly by variable. aLiteracy and region of birth refers to mother, not child. *Indicates significant difference between northern-born and southern-born at p < .05 (two-tailed tests).
© Copyright 2026 Paperzz