1/n I mentioned on @MattGrossmann's podcast that the shift towards permissive immigration attitudes can be traced as far back as the mid-2000s. For the unaware, this was a period (arguably beginning in 2005) marked by greater partican conflict over immigration reform and...
@MattGrossmann 2/n (as far I'm aware) the largest wave of mass immigrant protests in US history (in 2006). Thus, one possibility is that the clarification of the parties' immigration positions allowed respective constituents to align their preferences accordingly. The problem with this...
@MattGrossmann 3/n account is that, by some measures, even Republicans saw reductions/increases (though to a more modest degree than democrats) in restrictionist/permissive immigration attitudes. Thus, elite polarization is unlikely the whole picture.
@MattGrossmann 4/n Another possibility, which I suggest here, is that immigration restrictionism became increasingly framed in the media in 'racist' terms. Such would have the effect of activating anti-prejudice norms/considerations in the public at large, but especially among dems and libs.
@MattGrossmann 5/n With some help from @DavidRozado and his machine learning expertise, I was able to find some evidence consistent with this account.
@MattGrossmann @DavidRozado 6/n As depicted below, the marked spike in support for increasing immigration coincides with a similar spike in the lexical relationship between 'immigration' and 'racism' in the NYT 
@MattGrossmann @DavidRozado 7/n I should note that the data above is grouped into decades in the above so as to derive estimates that are less noisy (Rozado tells me that yearly estimates would be based on sparse data). But here it is is in 5 year intervals. Note the large jumps for the period 2005-2009.
@MattGrossmann @DavidRozado 8/n Moving onwards, 'racial resentment' (or what I like to call, when reverse-coded, 'liberal racial sympathy') is a measure often used in models that predict anti-immigration attitudes. Interestingly, I find that its growing relationship with immigration attitudes appears..
@MattGrossmann @DavidRozado 9/n ..to be almost entirely driven by greater movement towards 'racially liberal' responses, particularly among white liberals. Among white conservatives, in contrast, the relationship between racial conservatism and preferences for decreased immigration has remained largely...
@MattGrossmann @DavidRozado 10/n...stable across time. In other words, and at least by these data, the increasing 'racialization' of immigration attitudes in recent years is largely a function of attitude change among liberals (and, if to a lesser extent, moderates).
@MattGrossmann @DavidRozado 11/n Returning to the NYT finding, it's possible that the equation of immigration restrictionism with 'racism' has created an environment where being pro-immigration has become a means of signaling one's anti-racism.
@MattGrossmann @DavidRozado 12/n The above probably won't come as a surprise to many of you, but what's interesting is that the trend appears to have long pre-dated Trump and even the 2014 migrant crisis.
@MattGrossmann @DavidRozado 13/n Needless to say, such a trend is inherently polarizing insofar as it entails (at least in the minds of Democrats and liberals) that immigration restrictionism--an attitude that was 'mainstream' among the public for decades--is fundamentally illegitimate.
@MattGrossmann @DavidRozado 14/n On a more empirical note, I'd argue that recent research (e.g. svmiller.com/research/econo…) showing 'racial resentment' to be the strongest predictor of anti-immigration attitudes is misleadingly framed insofar as it ignores the fact that it's the relationship between...
@MattGrossmann @DavidRozado 15/n ...racially liberal and pro-immigration responses that's doing much of the heavy lifting.
@MattGrossmann @DavidRozado 16/n Here's more data suggesting that the shift began in the mid-2000s (which, below, comes out to 2006)
@MattGrossmann @DavidRozado 17/n Curiously, we observe a similar 2006 dip among white republicans followed by a complete rebound several years later.
@MattGrossmann @DavidRozado 19/n Even 30 years later, an overwhelming majority supported a pause or slowdown
@MattGrossmann @DavidRozado 20/n Hell, 'decreased' immigration was the modal response even among liberals in some years.
@MattGrossmann @DavidRozado 21/n I've previously tweeted this, but I'll add it here (since it also shows a 2006 shift)
@MattGrossmann @DavidRozado 22/n Not directly related, but still an interesting (if unsurprising) finding I just came across in this book: representatives in districts with larger populations of naturalized citizens (undocumented residents) are more likely to vote against (for) restrictive immigration bills
@MattGrossmann @DavidRozado 23/n I don't have too much confidence in these results (e.g. N=6), but they do suggest that states in which immigrant protests were held (in 2006) experienced significantly greater changes in preferred immigration levels.
@MattGrossmann @DavidRozado 24/n There were two additional non-protest states for which the sample sizes were too small to include. One of the critical limitations when examining state-level data.
@MattGrossmann @DavidRozado 25/n Actually, the inclusion of the remaining states (which had fewer than 20 observations) doesn't substantively change the results
@MattGrossmann @DavidRozado 26/n State-level weekly 'immigration' search interest starting becoming more closely associated with 'racism' search interest in 2006.
