Throughout 2015, and even after the GOP primaries had begun, most conservative pundits were confident that Trump would not secure the nomination, or even come close. Some of the reasons for this were understandable, others were based on a clearly faulty understanding of the Republican electorate.
One big mistake that the punditry made was assuming that massive numbers of Republicans identified as "true conservatives," and would thus be turned off by a candidate that clearly had no interest in conservatism as such. The reality of course is that most voters do not use an ideological litmus test when determining their preferred candidate. Furthermore, most Republican voters would fail such a test themselves.
As more data comes in over the next year (I know I am not the only person eager to get a hold of the 2016 ANES, NAES, and CCES), we will be able to make better sense of what motivated Republicans to vote for Trump in the primaries. In the meantime, however, we can gain some understanding by taking a look at the 2016 ANES pilot study -- the full study will be out some time next year. The part of the survey that is available now was conducted in January, just as the primaries were getting under way, and well before Trump had secured the nomination.
There are lots of useful questions in there. In particular, there are lots of new questions on racial attitudes that I will say more about later.
For now, one interesting thing to consider is whether it was true that Republicans that were more ideoleologically conservative were more opposed to Trump. That is, whether self-described conservatives in the electorate shared the conservative punditry’s antipathy to Trump.
As we typically see in surveys of this type, the 2016 NES pilot study showed that most Republicans identified as conservatives of some type; about three quarters identified as “closer to conservatives,” “somewhat conservative,” or “very conservative.” Were these conservatives less likely to support Trump than moderate and liberal Republicans? Not really.
Among the small number of Republicans that identified as some kind of liberal, about 42 percent said Trump was their preferred candidate. Among those Republicans that identified moderate, that is, as “neither liberal nor conservative,” about 41 percent preferred Trump. Among self-described conservatives, about 39 percent preferred Trump. Even the smaller category of Republicans that called themselves “very conservative” were generally supportive of Trump -- about 37 percent of them said Trump was their preferred candidate. Thus Trump had a plurality of support from every ideological category within the GOP.
Again, this survey was conducted in January, when there was still a very large number of GOP candidates in the running. As the number of candidates shrank in the following months, the support for Trump clearly grew among all of these ideological groups. The main take away from these numbers, however, is that conservative journalists and intellectuals who thought that their fellow conservative Republicans in the electorate would be their firewall against a Trump nomination were clearly mistaken.
Perhaps this simple tally of ideology and candidate preferences gives a misleading picture. It is possible that we see a stronger relationship between ideology and support for Trump if we run a more sophisticated model, including a number of control variables. To investigate this question, I generated a logistic regression model, in which support for Trump was the variable being predicted. I restricted the analysis to Republicans. The independent variables were ideology, race, gender, age, income, and education (a six point variable indicating the highest level of education the respondent had achieved). This model did indicate that being more conservative slightly increased the odds that a Republican supported a candidate other than Trump in the primaries, but the difference was not statistically significant. Only two variables in this model achieved statistical significance. Unsurprisingly, white Republicans were about two and a half times more likely to back Trump than non-white Republicans. And better educated Republicans were less likely to support Trump.