Predicting Party ID by Race

This post will analyze how numerous variables correlate with party ID both in the US population at large and within each major US racial group. Several points will be made. First, variation in many variable impact White people’s party ID more than non-Whites. Second, finding single variables, or combinations of variables, which take on values that predict non-Whites groups leaning Republican is surprisingly difficult. Third, it will be shown that several variables which people often talk about influencing party ID that don’t actually matter much at all.

For each variable, I will begin by presenting data from the General Social Survey. The General Social Survey has a 6 point party ID scale ranging from strong Democrat to strong Republican. For most variables analyzed, I’ll present the correlations between each variable and party ID among all respondents and among Whites, Hispanics, Blacks, and Asians. I’ll also present the mean party ID of members of each race at each value of the variable under discussion. Mean party ID (% Republican) will be expressed as the percentage of respondents who identified as independents who lean Republican, Republican, or strong Republican, among the set of participants who answered as leaning towards one of the two major parties. Independents with no lean and third party partisans are excluded from this calculation.

When possible, I’ll cite other research from others that is relevant.

The General Social Survey does not come with an Asian category.  I created one by binning together everyone who answered the question “From what countries or part of the world did your ancestors come?” with China, Japan, Philippines, India, or Other Asian Country.


Here is the GSS data for age:

Age - ChartAge - Table

Whites of all ages identify as republican and non-Whites of all ages identify as Democrats. Age has the strongest linear relationship with party ID among Blacks. Among all other groups, it has a slightly non-linear relationship.

Pew data from 2014 also shows an especially large effect of age on partisanship among Blacks and a non-linear relationship between party ID and age among Whites and Hispanics. (These are ratios of Democrats to Republicans)

Age 2

Pew data from 2016 shows a linear trend among Whites, due to Baby Boomers becoming more republican between the two surveys.

Age 3

It is sometimes thought that there is a general trend for people to move far to the right as they age. This data suggests that this is no true. Other data indicates the same. For instance, between 1994 and 2014 Gen X went from leaning Republican by 2 points to leaning democrat by 11 points. The silent generation, by contrast, went from leaning democrat by 5 points to leaning Republican by 4 points, suggesting that age doesn’t have a reliable impact on a groups party ID.

Generations & Party ID: 2014 vs. 1994

Looking at the general public, we see some evidence for the conventional view.

2014 Partisan Advantages by Year of Birth

However, this is very much an artifact of race. The same data set shows that there is no linear relationship between year of birth and partisanship among Whites.

2014 Partisan Advantages Among Whites, by Year of Birth

Non-Whites are younger than Whites are though, and this creates a pattern in the data such that republicanism increases with age more than it does within any population due to differences in republicanism between populations.


Here is the GSS data on sex and party ID:

Sex - Chart.jpg

White men lean Republican, White women are evenly split, and non-Whites of both sexes lean democrat.

Pew’s previously referenced 2014 dataset shows basically the same picture:

Sex - Pew 2014.jpg

Here is the data from Pew’s 2016 data set:

Sex - Pew 2016.jpg

All three analyses show females being to the left of males in each racial group. This seems to be most true among Hispanics and least Democrats among Blacks.

Among Whites who identify with one of the two major parties, roughly 50% identify as Republican. Because this doesn’t make a strong prediction about how they will vote, I looked at CNN exit poll data for the last three elections and found that White women voted Republican in all three, though in each case by a lesser margin than did White men.

Sex - Exit Polls.jpg

Women lean to the left of men and this is not an artifact of race. Possible explanations for this include sex differences in empathy and moral reasoning, female reliance on the welfare state during single motherhood, some women treating victim groups like children, and feminism.

Formal Education

Here is the GSS data on party ID by the highest degree obtained by the respondent:

Degree - ChartDegree - Table

Exit polling replicates this non-linear relationship for the presidential elections of 2016,  2012, 2008, 2004, and 2000Pew data, however, suggests that this stopped being true for partyID, which isn’t quite the same thing as voting, sometime in the mid-2000s.

Pew Data also shows that conservatism peaks among those with a college degree:

Education - Pew.jpg

There doesn’t seem to be any consistent linear relationship between education and party ID, and republicanism seems to be highest among those with a moderate degree of college education. This might be seen as running contrary to the predictions that would seem to follow from the hypothesis that colleges cause people to shift their views to the left.

There is other evidence to contradict this hypothesis. Mariani and Hewitt (2008) analyzed data on 6,807 college students from 38 colleges and universities. They looked at how the political views of these students changed over time and how that change related to the political views of their professors.

It was found that students moved to the left over the college years, but this was unrelated to how liberal their professors were. It was further shown that people tended to select into colleges that already matched their politics so that liberal freshmen went to schools with liberal professors and conservative freshmen went to schools with conservative professors. Even in schools with a majority conservative faculty, students became more liberal during college, though to a slightly lesser degree than those who went to more liberal colleges. During the college years, women and poor students moved to the left more than men and affluent students. Whites and non-Whites did not significantly differ in how their political views changed while in college. This all suggests that college students moving to the left is probably a function of their age and period of psychological development rather than the university as an institution impacting student’s political views.

Other work finds that the more a student interacts with professors the more moderate their political views become. Change in political ideology during college is largely a function of attending student groups and demonstrations and this effect is seen for liberal and conservative students.

And other research finds that students end up having a more favorable view of liberals and conservatives during their college years to a roughly equal degree.

College Impact 1.JPG

My guess is that college happens to coincide with the years in which people first attempt to form a political ideology that is independent of their parent’s, and that is what explains these trends.

Self Perceived Class and Income

Here is the GSS data on self-perceived class and party ID:

Class - ChartClass - Table

Only low-class Whites are (net) democrats. Non-Whites of all classes are Democrats. The relationship between class and party ID is exaggerated by racial differences in class. Non-Whites average a lower class level than Whites, exaggerating the extent to which low-class people tend to be democrats. Within each race, class explains less than 1% of the variance in party ID.

Here is party ID by income from Pew’s 2014 dataset:

Income - Pew 2014

And here is the data from their 2016 survey:

Income - Pew 2016.png

Like with class, only poor Whites are, on net, Democrats, and non-Whites are net democrats at every level of income. Class made a significant dent in Black party ID, but actual income does not.

The relationship between class and party ID is readily explicable in terms of self-interest. Work in multiple countries has shown that income predicts support for wealth redistribution (Gallup, 2016Sznycer et al., 2017).

A causal explanation is suggested by lab experiments where people are assigned positions and complete projects and it is found that people endorse the idea of participants getting paid in proportion to the amount of work did when they are assigned to positions that do more work than average. When they are assigned to positions involving less work than average, they endorse an equal distribution of payments.

Interestingly, the relationship between income and economic ideology that exists between parties does not exist among democrat voters. Pew data finds that richer republicans are the more likely than poor Republicans to say that the economic system is fair. However, this relationship is reversed among Democrats.

High-income Republicans less likely to say U.S. economic system is unfair

Marital Status

Here is the GSS data on marital status and party ID:

Marriage - Chart.jpg

Among Whites, only married people self ID as republican on net. Among Whites, Hispanics, and Asians, there isn’t much of a difference between those who have never been married and those who have been divorced. This suggests, though does not prove, that being married may have a causal impact on party ID.


Compared to the other variables analyzed thus far, religion is going to turn out to be very important in explaining variation in party ID. To begin with, here is the GSS data relating how religious a person rates themselves as being to their party ID

Religous Person - ChartReligous Person - Table

Several things are worth noting. First, highly religious Asians are just as likely to be a Republican as a Democrat. Secondly, religiosity has basically no effect among Hispanics. Third, religiosity explains nearly 5% of the variation in party ID among Whites. Fourthly, among Whites and Asians, more religious people are more Republican while the opposite is true among African Americans.

There are several reasons why religion might be associated with right-wing politics. The most obvious explanation is religious social values. There is more to it than that. Gill and Lundsgaarde (2004) find that more religious nations have smaller welfare states, plausibly because religion and government compete to supply many of the same welfare services.

Guiso et al. (2003) examine data from 66 nations and find a relationship between religiosity in most religions and various markers of conservative political ideology:

“Religious upbringing increases trust in the government for Muslims and to a lesser extent, for Hindus. Religious participation increases trust in the government for all religious denominations except Buddhists. The effect is stronger for Hindus and Muslims, weaker for Catholics and Protestants.

Religions differ most in their position on the trade off between equality and incentives. Protestants and Hindus are more willing to trade off equality for incentives, while Jews and Muslims are less so. For the other religions, the effect is insignificant.

Religious denominations also differ in their attitudes toward private ownership. Protestants, Catholics, and Hindus want more private ownership, while Muslims want significantly less private ownership. Interestingly, Catholics support private ownership twice as much as Protestants (and the difference is statistically significant at the 10 percent level). Catholics also are more in favor of competition than any other religious group (including Protestants), while religious Muslims and Hindus are strongly against competition.

The relation between religion and intolerance seems to be present in all religious denominations, both for religious upbringing and for attendance at religious services. Only Buddhists are more tolerant. The point estimates for Protestants and Catholics are very similar, while those of Muslims are much higher, and those of Hindus even higher. Actively religious Hindus are 29 percent more intolerant than non-religious people, Muslims 19 percent, actively religious Protestant and Catholics 7 percent more.

As panels D and E show, all religious denominations are associated with a more conservative attitude toward women. However, that effect is twice as strong for Muslims than for any other religion.”

On the other hand, Be’ey and Bloom (2015) find that priming Jews and Catholics to think about God increases their support for income redistribution. In a previous post, I analyzed data on 24 European nations and found a .46 correlation between the prevalence of Protestantism and economic freedom.

In America, the three most important religious outlooks are Protestantism, Catholicism, and atheism/agnosticism/unaffiliated. Around the world and in the US, data indicates that Protestantism has a stronger relationship with right-wing politics than catholicism or non-religiosity.

The GSS asks respondents how they feel, on a scale of 1 to 100, about various religious groups, with 100 indicating a high degree of favorability. Here is how feelings about protestants relate to party ID:

Prot Temp.jpg

The sample sizes for non-Whites here are small. Still, feelings about protestants explained 19% of the variance in party ID among Asian Americans and more than 5% of the variation among Hispanics.

This data would predict that Asian American party ID would increase by .18SD should their feelings about protestants increase to the level of Whites, eliminating over 80% of the White-Asian party gap.

Of course, this assumes a causal relationship between these variables that is probably too simplistic and that the pattern found here would be just as strong in the general population which is probably false.

Here is the GSS data with respect to feelings about Catholics:

Catholic Temp.jpg

This variable has a substantial effect among Hispanics such that feeling positively about Catholics predicts a strong increase in democrat party ID. Among non-Hispanics, the relationship is weaker but still substantial and in the opposite direction.

Here is the GSS data on feelings about Muslims:

Muslim Temp.jpg

It may be surprising that this variable has no effect on party ID among Whites, and has no statistically significant effect among any group.

Here is the GSS data for feelings about Jews:

Jew Temp.jpg

It is unsurprising that liking Jews increasing Republicanism. It is noteworthy though that this relationship is strongest among Asian Americans and not at all present among African Americans.

Religion Among Whites

Among Whites, there is a clear distinction between Atheists and Agnostics, who reliably vote Democrat, Catholics, who sometimes vote Republican and sometimes vote Democrat, and Protestants, who reliably vote Republican. The most Republican religious group in the country is either White evangelicals or White Mormons, depending on the survey looked at.

White - Religion - 1972 to 2008

NYT (2008)

White - Religion - 2000 to 2008

Pew (2008)

White - religion - 2012.jpg

CNN (2012)White - Religion - Pew 2014

Pew (2014)White - Religion - Pew 2016

Pew (2016)

Jewish Americans are very reliable Democrats, but Orthodox Jews are very reliable Republicans.

Shimron (2017)

Religion Among Asians

Turning to Asian Americans, it might be useful to first note that a plurality of Asian Americans are Christians.


Asians who are part of this plurality, that is, Asians who are Christian, also lean Republican. This is more true of protestant Asians than of Catholic ones, mirroring what we saw among Whites.


Pew (2012)

Religion Among Hispanics

In contrast to what we’ve seen among Asians and Whites, among Hispanics, being Catholic has essentially no effect on party ID.

Religion - Hispanics - 2014

Pew (2014)Religion - Hispanics - 2016

Pew (2016)

Religion Among African Americans

Among African Americans we see something even odder: Protestants are to the left of the non-religious.

Religion - Blacks - 2014

Pew (2014)Religion - Blacks - 2016

Pew (2016)

In sum, Protestants are to the right of Catholics among Whites and Asians. Christians are to the right of non-Christians among Whites and Asians but not Hispanics or Blacks.

Why the difference? Here’s one possibility: Christianity moves people’s social views to the right. This only translates into being a Republican among populations whose party ID is significantly influenced by their views on such social issues. This is true of Whites and Asians, but Hispanics and Blacks don’t change their party ID much according to their views on Christian social issues.


To investigate this possibility, let’s start with abortion. Whites who think that abortion should be allowed for any reason are Democrats. Whites who disagree are Republicans.

Abortion - Cjart

This question makes a larger difference in the party ID of Asian Americans as well, but not enough to make them net republicans. It has little impact on the party ID of Hispanic and Black Americans.

Asian Americans are to the left of Whites on abortion so this may play some role in their being Democrats. Hispanic and Black Americans, by contrast, are to the right of White Americans on abortion, and so this surely plays no role in them identifying as Democrats.

Gay Marriage

We see the same picture with respect to gay marriage. Whites on different sides of this issue also differ in terms of the party they are most likely to identify with. This is not true among Asian Americans, but it still has a large impact. Among Hispanic and Black Americans, views on gay marriage have little impact on party ID.

Gay Marriage - Chart

MH table

We can also see that Whites, Asians, and Hispanics, don’t significantly differ in their views on gay marriage, while Blacks are slightly to the right of other populations.

Religion After Controlling for Social Conservatism

We’ve now seen that Asian Americans who take right-wing views on abortion and gay marriage are more Republican than the general Asian American population but are still net democrats. Yet, Christian Asian Americans are net republicans. Given this, Christianities influence on social views cannot fully explain the republicanism of Christan Asian Americans.

To look at this more carefully, I ran a series of regressions looking at how religiosity correlated with party ID before and after controlling for views on gay marriage an abortion. Below you can see the results, with columns 1-2 being an analysis of Asian Americans, 3-4 being Whites, 5-6 being Hispanics, and 7-8 being Blacks.  (A negative ES indicates that more religious people were more Republican).

Religion Regression.jpg

As can be seen, controlling for social conservatism only significantly reduced the relationship between religiosity and party ID among Whites.

We saw before that religion has a relationship with economic views as well, especially in the case of Protestantism. So, let’s look at economic views next and then circle back to religion.

Wealth Redistribution

The GSS asks participants the following question:

“Some people think that the government in Washington ought to reduce the income differences between the rich and the poor, perhaps by raising the taxes of wealthy families or by giving income assistance to the poor. Others think that the government should not concern itself with reducing this income difference between the rich and the poor. Here is a card with a scale from 1 to 7. Think of a score of 1 as meaning that the government ought to reduce the income differences between rich and poor, and a score of 7 meaning that the government should not concern itself with reducing income differences. What score between 1 and 7 comes closest to the way you feel?”

Here is how that question lined up with party ID:

EQ Wlth ChartEQ Wlth Table

Group differences in the mean answer to this question are tiny and so can’t play a large role in the party ID gaps. However, we see that White and Asian Americans who take a conservative view of this issue ID as republican and Hispanics come very close to doing so. In contrast to these three trends, the degree to which Black party ID is unresponsive to this variable is rather striking.

Because Hispanic party ID came close to .50 among those who gave a right-wing response to this question, I decided to investigate voting once again. Looking at Blacks and Whites seemed pointless, so I only investigated Asians and Hispanics. I analyzed the reported voting of those who selected 5, 6, or 7, on the above income redistribution question.

Among Asians, there were very few respondents so I combined the results of all the elections between 1976 and 2012. Even still, the resulting sample was only 122 people, of which 52% reported having voted for the republican.

Looking at Hispanics who voted between 2000 and 2012 (Data for Hispanics only goes back to 2000) gives 192 respondents, 51% of which voted democrat and 46% voted of which Republican.

Bush won these Hispanics in both of his elections. So did Obama.

Since there was a significant jump in the proportion of Hispanics who identified as republican between the answered “5” and “6” on the income redistribution question, I also decided to analyze the votes of each answer given (N=658).


Here is the same analysis for Asian Americans (N=218):


Net republican voting doesn’t take place until an answer of 6 or above in either group, but once this threshold is reached voting favors Republicans.


The GSS asks participants if their income taxes are too high, too low, or just right. Among Whites, those who say they are too high are Republicans while those who give a different answer are Democrats. Among non-Whites, views on taxes have a lesser impact on party ID, and even those who say their taxes are too high aren’t anywhere near net Republicans.

Tax - Chart


Finally, for economic policy, the GSS asks participants whether welfare spending is too little, too much, or just right. As can be seen, Whites who think welfare spending is too high are Republicans. The same is not true for any other ethnic group, but this variable does have a large impact on Asian American party ID.

Welfare Spending

Economics and Religion

Now let’s look at the relationship between religiosity

Religion and Econ.jpg

As can be seen, Asian American religiosity no longer significantly predicts party ID after controlling for economic ideology. Given this, we might propose the following hypothesis: Asian Americans are democrats because they many of them are non-Christian and, as a result, economically to the left of Whites.

This theory doesn’t hold up well though because Asian Americans aren’t much to the left of White Americans on the economy. On the three-point welfare scale, Asians and Whites differed by .13 SD. On the income redistribution question, the gap was .10 SD. Thus, being less Christian has not resulted in Asian Americans being far to the left of Whites. If more Asian Americans became Christian, it is possible that their economic views would shift to the right of Whites and this could increase their Republican identification.

Here’s another (not mutually exclusive) hypothesis: being Christian makes people republican in part because they come to identify with Republicans as a group in virtue of participating in the same religious culture as them. This is true of Whites and Asians but is plausibly not true of Hispanics and Blacks because they have their own Christian churches which are not associated with republicanism at all.

This fits with some of the data we’ve already seen. Liking Catholics makes Asians and Whites more pro-republican but makes Hispanics more pro-democrat. Why? Maybe because Asians and Whites go to Catholic Churches and are familiar with a Catholic culture that is connected to Whites and Republicanism while Hispanics are familiar with a Latin American and pro-Democrat Catholic culture. By contrast, liking protestants turns everyone but Blacks more Republican, and this is plausibly because only Blacks have access to a Protestant culture that is separate from the mainstream, White, and Republican, one.

This is a speculative hypothesis but makes sense of a good deal of data.

General Political Ideology

Let’s now turn to general political ideology. According to data from the GSS, conservative Whites and conservative Asians net identify as Republicans while conservative Hispanics and conservative Blacks do not.

Pol Views - ChartPol Views - Table

Because the party ID of conservative Hispanics was near .50, I decided to check how conservatives of each ethnic group reported having voted in the 4 US presidential elections between 2000 and 2012. Only conservative Whites were found to have usually voted for the Republican candidate, although the analysis of Asian Americans involved a small sample and so should be interpreted with caution.

Pol Views - AsiansPol Views - BlacksPol Views - HispanicsPol Views - Whites

This shouldn’t be to surprising since general ideology is a function of many different sorts of issues and we’ve already seen that Christian-Social views can’t take on values that predict non-Whites being net republicans. With that said, let’s move on to explore other specific aspects of political ideology, starting with issues directly tied to race.


Affirmative Action

The GSS has a question asking them to rate, on a 5 point scale, their agreement with the proportion that “African Americans have been discriminated against for so long that the government has a special obligation to help improve their living standards.” Here is how responses to that question relate to party ID:

Affirmative Action - ChartAffirmative Action - Table

Whites and Asians with right-wing views on affirmative action are net republicans. Hispanics and Blacks with such views are not.


Here is the GSS data on party ID and views on immigration:

Immigration - TableImmigration - Chart

Once again, Whites and Asians with right-wing views are net republicans. Among Asians, this trend isn’t strong enough to be statistically significant. Among Hispanics, the relationship is distinctly non-linear.

Since Hispanics who want immigration reduced by a lot are still net Democrats, it would be surprising if Hispanics became Republicans when they got the change to vote for a pro-immigration Republican. Data bares out that this does not happen. This can be seen by comparing Pete Wilson of California and Geroge Bush of Texas, two governors who, in the 1990s, were seen as taking opposite approaches on how the GOP should deal with Hispanics. Both candidates lost Hispanics by similar margins, and while Wilson’s push to deny welfare to illegal immigrants seems to have pushed Hispanics in CA to the left, they were net Democrats before that (Monogan and Doctor, 2017).

Similarly, analyses of support for congressional candidates based on their NumbersUSA grade finds that Hispanics are no more likely to vote for Republicans when they are pro-immigration (Hawley, 2013). (A is very anti-immigration, F is very pro-immigration).

The same study found that, as the GSS data would predict, White support for Republicans was predicted by their immigration stance with anti-immigrant republicans receiving more of the White vote.

So, that is kind of interesting and might have some really obvious practical implications.

Racial Feelings

The GSS also asks participants to rate, on a nine-point scale, how close they feel to Blacks and Whites. Here is how closeness to Blacks relates to party ID by race:

Close to Blacks - Chart

Close to Blacks - Table

The closer Blacks feel to their race, the less republican they tend to be. The same trend is seen among Whites and Hispanics, the more they like Blacks the less republican they are.

Here is how closeness to Whites relates to party ID:

Close to Whites - Chart

Close to Whites - Table

Unlike Blacks, White party ID does not relate to how close they feel to their own ethnic group. Nor does closeness to Whites have an effect on the party ID of any other group.

The GSS also asks a more generic question about closeness to your own ethnic group. However, the sample size for this question is much smaller. With that in mind, here is how it relates to party ID:Close to Ethnicity

So, this suggests that being close to your ethnic group slightly increases the republicanism of Whites and has a much larger, negative, effect on the republicanism of Hispanics and Blacks. The effect among Asians isn’t statistically significant.

The GSS asks a similar question with a slightly larger sample size:

When you think about yourself, how important is your ethnic 
group membership to your sense of who you are?

Respondents answer in terms of a 4 point scale that ranges from very important (1) to not at all important (4). Here is how that question relates to party ID:

Ethic Importance.jpg

So, this suggests that ethnic identity is unrelated to party ID within each group. It’s also worth noting that Whites have a lower level of ethnic identity than non-Whites.

In sum, racial politics doesn’t seem to have much explanatory power with respect to why it is that Hispanics and Blacks are Democrats. It does, however, seem to play an important role in the party ID of Asians and Whites. While this is true of racial politics, broader racial feelings don’t play much of a role in the party ID of Whites or Asians.

Guns and Weed

Guns and weed seem conceptually related to me and they relate to party ID by race in basically the same way, so here is the GSS on those two issues:

Gun Law.jpg


Right-wing views predict Whites being republican, Asians and Hispanics being more Republican but still net republicans, and Blacks being democrats. With respect to weed, Whites are to the left of non-Whites, so this plays no role in non-Whites being Democrats. With respect to guns,

Foreign Policy

Foreign policy is more interesting. Whites identify as net republicans only if they favor an interventionist foreign policy. Asian Americas are more likely to identify as Republican if they favor an anti-interventionist foreign policy.

F P.jpg

I have no idea why this would be the case.

Knowing Republicans and Democrats

So, that’s it for ideology. Let’s now turn to social variables. The GSS asks respondents “Thinking about all the people you are acquainted with: How many of do you think are Republicans? Would you say almost all (1), most (2), about half (3), a few (4) or none (5)?”. Here is how that question correlates with party ID:

Knowing Republicans

And here is how the same question, but about Democrats, correlates with party ID:

Knowing Democrats

Clearly, these effects are large in all groups. However, the sample size is small for non-Whites and the direction of causality is especially suspect here since to some degree being a Republican will cause you to select for more Republican friends and fewer democrat ones.

Where You Live

The GSS also records where participants live. Here is the breakdown of republicanism by region by race:

Region 1

So, there are regions where Whites are net democrats and regions where Whites are net republicans. In the South, Asians are almost net republicans. Interestingly, there is a strong relationship such that the place where Whites tend to be more republican are also the areas where Hispanics and Asians tend to be republican, but Blacks, if anything, tend to be less republican in such areas:

Region 2.JPG


Where You Were Born

Related to where you live is the question of where you were born. He is party ID by race by whether or not someone was born abroad:


This variable has no impact on Hispanics or Blacks and not much of an impact on Asians. However, it makes the difference between Whites being net republicans and net democrats.

Pew’s 2016 dataset finds that native-born Hispanics lean Democrat (30 – 62) by a lesser margin than foreign-born Hispanics (20 – 66). Other Pew data, from 2012, finds Native born Asians lean Democrat (31 – 54) by a larger margin than foreign-born Asian Americans (27 – 49).  So maybe there is an effect, but it isn’t very large.

The GSS also asks people if their parents were born in the US. Here is how that relates to party ID (among people who were themselves born in the US):

Par Born

So, once again we see that Whites who are not from the US, or whose parent’s aren’t, identify as Democrats. There is a similar but weaker effect among non-Whites.

Finally, the GSS asks people how many of their grandparents were born outside the US.

Gran Born

This data suggests that fourth generation Asian Americans may be near net republicans. It also highlights the interesting fact that Whites who were born here, and whose parents were born here, are net democrats if most of their grandparents were born abroad.

Narrow Ethnicity

It is also worth noting that within each race there is a good deal of variation, in terms of party ID, by country of origin. For instance, among Hispanic Americans you see net republicans among the Cubans.

Table: Party Affiliation of Hispanic Voters by Ethnic or Nationality Group, July 1999


Cubans have been becoming less Republican with time:

Political Affiliation Among U.S. Cubans

In fact, Pew reports that “Over half (56%) of Cubans ages 18 to 49 identified with or leaned toward the Democratic Party compared with 39% of those 50 years and older. Conversely, older Cubans tend to identify with or lean toward the Republican Party more than younger Cubans, by 44% to 23%. Even so, the share of older Cubans who are Republican has declined over time. In 2002, among all Cubans, some 68% who were 50 and older said they identified with or leaned toward the Republican Party.”

Still, the majority of Floridian Cubans voted for Trump:

Among Asian Americans, Indians are the most pro-democrat, and the Vietnamese and the Filipinos, while net democrats are to the right of the general public:

And finally, I have a whole post on the party ID of White ethnic groups. This chart gives you the general picture:


White Genes

We’ve reached the last variable on my list: White genetic admixture. There is a weak -.18 correlation between how White a Hispanic country is and how democrat immigrants from that country are.

This would predict a difference in democrat advantage of 22 points if a population went from 0% White to 100% White, suggesting that it cannot explain the majority of the difference between Cubans and other Latin American groups.

Country % White Dem Advantage
Peurto Rico 0.64 41.5
Mexico 0.42 22.2
Cuba 0.72 -6
Dominican Republic 0.47 57.5
El Salvador 0.15 21.5

Among Hispanic Americans, there is a negative correlation between skin darkness, a proxy for White admixture, and republicanism too.

However, even the lightest skinned Hispanics seem to slightly lean democrat. So European admixture may play an important role in Hispanic party ID, but, so far as I am able to tell, it alone can’t make Hispanics republican.

Having light skin color also correlates with republicanism among Asian Americans, but even very light skinned Asian Americans are net democrats.

By contrast, skin color which has basically no relation to party ID among blacks:

It does, however, have a relationship with ideology, with darker skin predicting more conservative views.

But, as we’ve already seen, conservative Blacks overwhelmingly democrats, so this doesn’t translate to party ID, which itself only translates moderately to voting.

So, White admixture might make some groups a little more Republican, but it probably isn’t a huge effect.

The Impossibility of Black Republicans

In the 1970’s, the GSS asked the following question:

"Which party, the Republican or the Democratic, do you feel 
will do more to help blacks in the next few years, or do you 
think there isn't much difference between the two?"

This question was asked to 170 Blacks. Only 3 answered with the Republican party, and of those three, two, or 66% of them, self-identified as Democrats. There were 45 Black respondents who said that there wasn’t much of a difference, and 38, or 83%, of them, identified as Democrats. And then there were the 122 respondents who said the democrat party, of which 97% identified as Democrats.

This data fits a general theme that has stood out in this data: there are no policy views, demographic variables, or anything else I can find, that would predict Blacks being net republicans.

Nevin (2017) found that Republicans running Black candidates didn’t make much of a dent in the Black vote either. Astonishingly, Kidd et al. (2007) studied a congressional race between a Black Democrat and a Black Republican and found that even Blacks who self-identified as Republicans only voted for the Black Republican 12.3% of the time. Views on gay marriage and abortion weren’t predictive of Black voting, and to get the probability of voting Republican over 50%, Kidd et al had to restrict their analysis to Blacks, in a race featuring a Black Republican, who self-identified as Republicans, and who self-identified as evangelic Christians, and who supported the Iraq war, of whom 65% voted Republican.

Now, when I read this I was pretty surprised by the idea that Black Republicans would on net vote for Democrats. So, I went to the GSS to check, and these are small samples, but they do show that Black people who self ID as Republicans also report voting for the Democrat, on net, in 6 of 12 US presidential elections.

Year Democrat Republican N
2012 82% 11% 13
2008 68% 32% 27
2004 18% 76% 34
2000 35% 64% 36
1996 62% 28% 48
1992 49% 42% 48
1988 20% 80% 33
1984 30% 65% 26
1980 78% 18% 23
1976 69% 26% 22
1972 36% 64% 33
1968 47% 53% 14


So, what can we conclude from all this? First, White people’s party ID is super flexible, and their positions on single issues can predict changes in their party ID. Secondly, Asian Americans can be predicted to be net republicans on the basis of religion and/or views on affirmative action or the economy. The same might be true of Hispanics with respect to economic views. Blacks can’t be predicted to be republican on the basis of any variable I know of, and can’t even be predicted to reliably vote Republican on the basis of self-identifying as a Republican.

4 thoughts on “Predicting Party ID by Race

  1. Pingback: On White Nationalism | Ideas and Data

  2. Pingback: Ben Shapiro, Conservative Inc, and the “Groyper” War of 2019 | Ideas and Data

  3. Pingback: Jewish Influence on American Politics | Ideas and Data

  4. Pingback: Are Hispanic Americans Becoming Republicans? | Ideas and Data

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s