Skip to main content Accessibility help

The politics of immigrant policy in the 50 US states, 2005-2011

  • James E. Monogan (a1)

This article asks what shaped immigrant policy in the 50 states between 2005 and 2011. Theoretically, politicians are influenced by electoral considerations as they craft laws. Law-makers consider both current public opinion and how the electorate is likely to change, at least in the near future. Empirically, the article analyses an original dataset on immigrant-related laws enacted by the states with a Bayesian spatial conditionally autoregressive model. The analysis shows that state immigrant policy is affected primarily by legislative professionalism, electoral ideology, state wealth and change in the foreign-born population.

    • Send article to Kindle

      To send this article to your Kindle, first ensure is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about sending to your Kindle. Find out more about sending to your Kindle.

      Note you can select to send to either the or variations. ‘’ emails are free but can only be sent to your device when it is connected to wi-fi. ‘’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

      Find out more about the Kindle Personal Document Service.

      The politics of immigrant policy in the 50 US states, 2005-2011
      Available formats
      Send article to Dropbox

      To send this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Dropbox.

      The politics of immigrant policy in the 50 US states, 2005-2011
      Available formats
      Send article to Google Drive

      To send this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you use this feature, you will be asked to authorise Cambridge Core to connect with your <service> account. Find out more about sending content to Google Drive.

      The politics of immigrant policy in the 50 US states, 2005-2011
      Available formats
Corresponding author
James E. Monogan IIIAssistant Professor Department of Political Science University of Georgia 413 Baldwin Hall Athens, GA 30602 USA Email:
Hide All

1 Intriguingly, Hero and Preuhs (2007) find that government ideology is a strong predictor of TANF laws, but citizen ideology is not. By contrast, the appendix re-estimates the model of Table 2 including both legislative and citizen ideology, but finds that citizen ideology is a strong predictor but legislative ideology is not. Perhaps TANF policy was sufficiently technical that legislators could more easily adopt their own preferences relative to other aspects of immigrant policy.

2 As a robustness check in case of endogeneity bias, I re-estimated the model of Table 2 using two-stage least squares. In this model, I introduced Erikson, Wright and McIver's measure of public ideology from 1995 to 1999 as an instrument for ideology during the time frame of interest. Presumably, the old values of ideology could not have been affected by current policy, but they could affect later values of public ideology. Public opinion ideology continued to show a strong positive effect in this alternative model, with a coefficient of 0.056 and a 90 per cent confidence interval of [0.027,0.084].

3 Source: Table on percentage change in foreign-born by state generated by Terrazas and Batalova of the MPI Data Hub,, accessed 22 June 2010.

4 Of new legal admissions of immigrants in 2008, 43.1 per cent came from Latin America and 34.6 per cent came from Asia (United States Department of Homeland Security 2009, Table 3).

5 See the appendix on data and analysis for a more complete source reference. Also, the example law synopses of the next paragraph are all based on the cited reports by NCSL. I focused on laws that were enacted by each state between the years 2005 and 2011.

6 As a check on the coding scheme, a second coder independently coded all of the 2008 laws (36.7 per cent of the total) for comparison with the author's coding. For the binary tone of the law, Krippendorf's α = 0.789, and for the ordinal scale of scope, α = 0.723. Both exceed 0.7, indicating acceptable inter-coder reliability between the author and the second coder.

7 The appendices describe more thoroughly the coding rules for placing each law into a category and shows the number of laws every state adopted in each of the scope and tone categories. Omnibus laws were broken up such that each provision was coded as a separate law. The appendix includes a table listing the laws that were split up.

8 The appendices list descriptive statistics of this and all other measures in Table A.1, scores for each state with their components in Table A.3, and a density plot of the measure in Figure A.1. All estimates were computed in R 2.14.0 (R Development Core Team 2009). Regarding Equation 1, the addition of one extra welcoming and one extra hostile law prevents any undefined ratios or logs. The main findings of the analysis remain intact even if this measure is constructed with raw counts of laws (rather than weighted counts), using percentages (rather than ratios), and by adding 2 or 3 to the numerator and denominator (rather than 1). The data were pooled across all years to prevent unrepresentative scores due to micronumerosity.

9 For more detail, Banerjee et al. (2004, Chapter 3) describe methods for areal data, or data defined by geographic boundaries, that account for spatial correlation. This discussion includes a technical description of the CAR model.

10 More specifically, Squire captures salary and benefits through the base legislative salary reported by each state, time demands through the number of days a legislature meets per year, and staff and resources through legislative staff figures gathered by the National Conference of State Legislatures (specifically total staff during the session, including permanent and session-only staff). A complete description of the measure is available in Squire (2007).

11 For more details on the data sources for ideology and the other input variables, see the appendix. The Pearson correlation between this CCES ideology measure and a 1995–1999 aggregation of Erikson 1993's data is 0.832. As an alternative specification of the model in Table 2, public ideology was measured with a factor analysis of self-reported ideology, Berry et al. (1998) measure of citizen ideology, and presidential voting in 2004. The results were similar to those reported.

12 The hyperpriors on the precision of the two random effects terms ($$)(-->$<> 1/{{\sigma }_h} <$> <!--$$ and $$)(-->$<> 1/{{\sigma }_c} <$> <!--$$) are based on the recommendation of Best et al. (1999), who studied the effect of changing the distributional form of the smoothing prior. The goal is to assume beforehand that the unexplained variance is split evenly between the clustering term and the heterogeneous term, thereby preventing the hyperpriors from shaping how the variance is distributed (Carlin and Pérez 2000; Banerjee et al. 2004). I also considered the recommendation of Bernardinelli et al. (1995) and recovered a similar posterior distribution for the two variance terms. Also, for modelling purposes, Alaska is treated as a neighbour of Washington, and Hawaii is treated as a neighbour of California. All other neighbours are defined by whether the two states share a border.

13 Specifically, the probability that these variables have an effect in the expected direction exceeds 0.95. Also, the Gelman-Rubin (Gelman and Rubin 1992) statistic converged to 1 for all coefficients, indicating the absence of evidence of nonconvergence.

14 All results are interpreted as if the mean value of a parameter holds. Interpreting effects in terms of percentage increases is possible because the dependent variable is logged, so taking the exponential of an unstandardised coefficient provides a multiplicative factor by which the ratio of welcoming to unwelcoming laws increases. For this particular coefficient, Squire's professionalism measure is based on proportions of legislative capacities compared to the US Congress, so an increase of 0.01 in this variable could be interpreted as a percentage point increase in professionalism. Since the square root of the measure is taken, the marginal effect of the variable changes based on the level of professionalism. Moving from the mean of professionalism (0.19) to a percentage point higher (0.20) amounts to an increase of 0.011324 in the square root measure. exp(1.506 × 0.011324) = 1.0172.

15 As an alternative specification, I also estimated this model with an additional control for public opinion on immigration issues. To do this, I incorporated Lax and Phillips's (2012) measures of state public opinion on the issues of issuance of driving licences to illegal immigrants, prohibition of bilingual education, providing in-state tuition for children of illegal immigrants, and requiring the state government to verify citizenship status before making hiring decisions. These measures of public sentiment were constructed using the novel method described in Lax and Phillips (2009). I combined these four variables into one measure of immigration opinion using principal components analysis and included this measure in a model otherwise identical to the one reported in Table 2. In this alternative model, symbolic ideology maintained its robust and positive effect with a posterior mean of 0.037 and a 90 per cent credible interval of [0.013,0.061]. The posterior coefficient for issue-specific public opinion had a positive mean of 0.071, but the effect was not very robust as the 80 per cent credible interval was [−0.105,0.247]. This implies that legislators are more concerned about symbolic ideology than issue-specific immigration opinion.

16 More formally, Best et al. (1999) define $$)(-->$<>\psi \: = \:\frac{{sd\left( \phi \right)}}{{sd\left( \theta \right)\: + \:sd\left( \phi \right)}} <$> <!--$$ as a measure of the share of random effects variability due to clustering. Table 2 shows that, for this model, the posterior mean of ψ is 0.332.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

Journal of Public Policy
  • ISSN: 0143-814X
  • EISSN: 1469-7815
  • URL: /core/journals/journal-of-public-policy
Please enter your name
Please enter a valid email address
Who would you like to send this to? *



Altmetric attention score

Full text views

Total number of HTML views: 0
Total number of PDF views: 0 *
Loading metrics...

Abstract views

Total abstract views: 0 *
Loading metrics...

* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.

Usage data cannot currently be displayed