138 THE REVIEW OF ECONOMICS AND STATISTICS Our theoretical framework hypothesizes that immigration the MSA/county-10%definition.Alternative measures are policy can affect individuals'factor income according to discussed in the robustness checks. their skill levels.To test whether skills are a key determinant We also constructed several measures of non-economic of immigration-policy preferences,for each individual we determinants of preferences.Following previous work in the construct two commonly used skill measures.First,respon- political-economy literature,we include the following mea- dents were asked to report their occupations coded accord- sures in our baseline analysis:gender,age,race,ethnicity, ing to the three-digit 1980 Census Occupation Code classi- personal immigrant status,party identification,and political fication.From the U.S.Department of Labor (1992,1994, ideology.Gender is a dichotomous variable equal to 1 for 1996)we obtained the 1992,1994,and 1996 U.S.average females.Age is a continuous variable.For race,we construct weekly wages for each three-digit occupation.Under the the dichotomous variable Black,which is equal to 1 if the assumption that the average market returns for a given respondent is African-American.For ethnicity,we construct occupation are determined primarily by the skills required the dichotomous variable Hispanic,which is equal to 1 if for that occupation,these average wages,called Occupation the individual self-identifies with a Hispanic ethnic group. Wage,measure respondents'skill levels.As a second skill Immigrant is a dichotomous variable equal to 1 if the measure,the NES survey also records the years of education respondent or his/her parents were immigrants into the completed by each respondent,Education Years.Educa- United States.Party Identification is a categorical variable tional attainment is another commonly used measure of ranging from1for“strong Democrat'”to7for“strong skills,so we use it as an alternative skills variable. Republican."Finally,Ideology is a categorical variable As discussed earlier,Citrin et al.(1997)primarily inter- ranging from 1 for“extremely liberal'”to7for“extremely pret educational attainment as a demographic variable rather conservative."In addition to these variables,for certain than a skills variable.Below,we present strong evidence specifications we included additional regressors which we discuss below. that education measures labor-market skills once other con- siderations such as gender and political ideology are con- B.Missing Data and Multiple Imputation trolled for.Also,our mapping of occupation categories into average occupation wages captures skills across occupa- Upon constructing the variables described in subsection tions much more accurately than do the occupation categor- IV A and combining them into individual-level data sets for ical variables in Citrin et al. each cross-sectional survey,we observed that there was a In addition to skill measures,we need measures of where significant amount of missing data.In each survey,some respondents live combined with information about gateway individuals did not report either occupation or educational communities.For each respondent,the NES reports the attainment;thus,for these respondents,we could not con- county,state,and (where appropriate)metropolitan statisti- struct skill measures.Missing data also existed for some of cal area(MSA)of residence.We combine this information our non-economic determinants of immigration-policy pref- with immigration data to construct several alternative mea- erences.Across the range of models that we estimated, sures of residence in a high-immigration area.First,we when we simply dropped observations with any missing defined local labor markets two ways:by a combination of data,we generally lost between 25%and 45%of the total observations. MSAs and counties,and by states.In our MSA/county definition,each MSA (with all its constituent cities and This standard approach for dealing with missing values, known as listwise deletion,can create two major problems. counties)is a separate labor market;for individuals living One is inefficiency caused by throwing away information outside an MSA,the labor market is the county of residence. that is relevant to the statistical inferences being made. Following the extensive use of MSAs in area-analysis Furthermore,inferences from listwise-deletion estimation studies and Bartel's (1989)finding that immigrants arrive can be biased if the observed data differs systematically mostly into cities,we prefer the MSA/county definition but from the unobserved data.In our case,inefficiency was try states for robustness.Second,for each definition of local clearly a problem.We also had little reason to think our data labor markets,we try three different definitions of a high- were missing at random,so we worried about biased infer- immigration labor market:5%,10%,and 20%shares of ences.(See King et al.(2001)for a detailed discussion.) immigrants in the local population.These immigration and Alternatives to listwise deletion for dealing with missing labor force data are from the 1990 decennial census as data have been developed in recent years.The most general reported by the U.S.Bureau of the Census (1994).Alto- and extensively researched approach is multiple imputation gether,for each of our six primary measures,we construct a dichotomous variable,High Immigration MSA,which is In 1990,immigrants accounted for 7.9%of the overall U.S.population. equal to 1 for residents in high-immigration labor markets. Accordingly,our 5%cutoff might seem too low,but for completeness we In the tables,we report the results for our preferred measure, estimated the specification.Also,the 1990 Census MSA data are orga- nized by 1990 MSA definitions,but the 1992 and 1994 NES surveys locate individuals by 1980 MSA definitions.Using unpublished informa- "taking"jobs from natives during adjustment to a new full-employment tion on 1980-1990 MSA changes obtained from Census officials,we equilibrium. corrected discrepancies as best we could. This content downloaded from 202.120.14.193 on Mon,15 Feb 2016 10:04:26 UTC All use subject to JSTOR Terms and Conditions138 THE REVIEW OF ECONOMICS AND STATISTICS Our theoretical framework hypothesizes that immigration policy can affect individuals' factor income according to their skill levels. To test whether skills are a key determinant of immigration-policy preferences, for each individual we construct two commonly used skill measures. First, respon￾dents were asked to report their occupations coded accord￾ing to the three-digit 1980 Census Occupation Code classi￾fication. From the U.S. Department of Labor (1992, 1994, 1996) we obtained the 1992, 1994, and 1996 U.S. average weekly wages for each three-digit occupation. Under the assumption that the average market returns for a given occupation are determined primarily by the skills required for that occupation, these average wages, called Occupation Wage, measure respondents' skill levels. As a second skill measure, the NES survey also records the years of education completed by each respondent, Education Years. Educa￾tional attainment is another commonly used measure of skills, so we use it as an altermative skills variable. As discussed earlier, Citrin et al. (1997) primarily inter￾pret educational attainment as a demographic variable rather than a skills variable. Below, we present strong evidence that education measures labor-market skills once other con￾siderations such as gender and political ideology are con￾trolled for. Also, our mapping of occupation categories into average occupation wages captures skills across occupa￾tions much more accurately than do the occupation categor￾ical variables in Citlin et al. In addition to skill measures, we need measures of where respondents live combined with information about gateway communities. For each respondent, the NES reports the county, state, and (where appropliate) metropolitan statisti￾cal area (MSA) of residence. We combine this information with immigration data to construct several alternative mea￾sures of residence in a high-immigration area. First, we defined local labor markets two ways: by a combination of MSAs and counties, and by states. In our MSA/county definition, each MSA (with all its constituent cities and counties) is a separate labor market; for individuals living outside an MSA, the labor market is the county of residence. Following the extensive use of MSAs in area-analysis studies and Bartel's (1989) finding that immigrants arrive mostly into cities, we prefer the MSA/county definition but try states for robustness. Second, for each definition of local labor markets, we try three different definitions of a high￾immigration labor market: 5%, 10%, and 20% shares of immigrants in the local population. These immigration and labor force data are from the 1990 decennial census as reported by the IJ.S. Bureau of the Census (1994). Alto￾gether, for each of our six primary measures, we construct a dichotomous variable, High Immigration MSA, which is equal to 1 for residents in high-immigration labor markets. In the tables, we report the results for our preferred measure, the MSA/county-10% definition. Alternative measures are discussed in the robustness checks.9 We also constructed several measures of non-economic determinants of preferences. Following previous work in the political-economy literature, we include the following mea￾sures in our baseline analysis: gender, age, race, ethnicity, personal immigrant status, party identification, and political ideology. Gender is a dichotomous variable equal to 1 for females. Age is a continuous variable. For race, we construct the dichotomous variable Black, which is equal to 1 if the respondent is African-American. For ethnicity, we construct the dichotomous variable Hispanic, which is equal to 1 if the individual self-identifies with a Hispanic ethnic group. Immigrant is a dichotomous variable equal to 1 if the respondent or his/her parents were immigrants into the United States. Party Identification is a categorical variable ranging from 1 for "strong Democrat" to 7 for "strong Republican." Finally, Ideology is a categorical variable ranging from 1 for "extremely liberal" to 7 for "extremely conservative." In addition to these variables, for certain specifications we included additional regressors which we discuss below. B. Missing Data and Multiple Imputation Upon constructing the variables described in subsection IV A and combining them into individual-level data sets for each cross-sectional survey, we observed that there was a significant amount of missing data. In each survey, some individuals did not report either occupation or educational attainment; thus, for these respondents, we could not con￾struct skill measures. Missing data also existed for some of our non-economic determinants of immigration-policy pref￾erences. Across the range of models that we estimated, when we simply dropped observations with any missing data, we generally lost between 25% and 45% of the total observations. This standard approach for dealing with missing values, known as listwise deletion, can create two major problems. One is inefficiency caused by throwing away information that is relevant to the statistical inferences being made. Furthermore, inferences from listwise-deletion estimation can be biased if the observed data differs systematically from the unobserved data. In our case, inefficiency was clearly a problem. We also had little reason to think our data were missing at random, so we worried about biased infer￾ences. (See King et al. (2001) for a detailed discussion.) Alternatives to listwise deletion for dealing with missing data have been developed in recent years. The most general and extensively researched approach is multiple imputation "taking" jobs from natives during adjustment to a new full-employment equilibrium. 9 In 1990, immigrants accounted for 7.9% of the overall U.S. population. Accordingly, our 5% cutoff might seem too low, but for completeness we estimated the specification. Also, the 1990 Census MSA data are orga￾nized by 1990 MSA definitions, but the 1992 and 1994 NES surveys locate individuals by 1980 MSA definitions. Using unpublished informa￾tion on 1980-1990 MSA changes obtained from Census officials, we corrected discrepancies as best we could. This content downloaded from 202.120.14.193 on Mon, 15 Feb 2016 10:04:26 UTC All use subject to JSTOR Terms and Conditions