Black Attachment to the Democratic Party and Co-Racial Context

I recently read a great working paper titled “Bound by Blackness: Understanding the Maintenance of Black Democratic Party Loyalty” by Laird et al. that sheds important new light on the partisanship of African-Americans. The persistently strong attachment to the Democratic Party among blacks has long been explained as a function of “linked fate” (concern among blacks with what happens to blacks as a group). However, after showing the weak evidence for this theory, these authors use several pieces of evidence to argue for a different theory that center on social sanctions leading blacks to overwhelmingly identify as Democrat. Namely, pressure from group expectations leads to greater compliance with the group norm that is Democratic support among blacks. Some of these same authors offer compelling evidence for this same theory in other research as well.

I thought one natural (and simple) extension of this work would be to check how black political identity varies by local social context. With its large sample of blacks (7,920) and geographic codes (county FIPS and zip code) that could link to social context demographic data from the Census, the 2016 CCES provides a good opportunity to see whether black political identity varies by the extent to which they live near other blacks. In this way, while far from perfect, I could capture a similar mechanism of pressure to comply with group norms driven by the amount of surrounding group members who could influence an individual. My general expectation is thus that blacks should have greater attachment to the Democratic Party in areas that contain more blacks.


In the above plot, I generally find support for this expectation. The x-axis shows 10 intervals representing the percentage of blacks living in black survey respondents’ zip codes, and the y-axis shows the percentage of black survey respondents that identify as Democrat (Democrats + Democrat leaners). In zip codes that range from 0 to 10 percent black, blacks identify as Democrat 74.1 percent of the time. At the highest level of living with co-racial individuals–in the 90-100 percent black zip code range–blacks identify as Democrat at a 89.8 percent rate. Thus, if the social sanctions theory applies here, blacks receive the least amount of group pressure to comply with norms (i.e. Democrat support) in the areas with least amount of fellows blacks (0-10 percent) and that translates into the lowest level of Democratic attachment across all zip code ranges. In a local context where blacks live with many other blacks (90-100 percent), blacks identify as Democrat at the highest rate across all zip code ranges, perhaps because of greater pressure to comply with group norms as part of living around more co-racial individuals. (Note: the proportions at 90-100 and 80-90 are statistically significantly different at conventional levels, but the 90-100 and 30-40 proportions are only different at p<0.10.)

Interestingly, black Democratic attachment does not monotonically increase as percentage black zip code increases. Instead, there’s a peak at the 30-40 percentage range, after which Democratic attachment decreases and then picks up again around the 80-90 and then 90-100 range. The relationship therefore doesn’t perfectly follow my expectation, and so the evidence here is not extremely strong. However, the Democratic identification estimates at the highest and lowest levels of black zip code percentage ranges is telling, and suggest there is some type of context effect that shapes the extent to which blacks support the Democratic Party. In this sense, the results here lend some additional support for the theory of social sanctions and group compliance pressures as an explanation for black partisan homogeneity.

Update 3/22/18:

I plotted the same relationship from the last graph but broke it up by self-described (five-point) ideology below. Interestingly, the relationship between amount of co-racial individuals and attachment to the Democratic Party appears stronger among the two conservative groups (“conservative” and “very conservative”) than among the other ideological groups. As with the overall relationship, this specific result mirrors another one of the findings from Laird et al.’s work: conservative blacks experience this social pressure to conform to group expectations (based on their social context) the most.


Black Attachment to the Democratic Party and Co-Racial Context

Tracking National Attention toward Mass Shootings with Google Trends Data

Many often lament that attention toward mass shootings and subsequent debate they engender is fleeting. In a matter of a week, if not days, national discussion about the tragedy itself as well as measures to prevent future ones (largely centered around gun control) quickly evaporate. However, with the most recent mass shooting at Stoneman Douglas High School, there does seem to be evidence of a different trajectory.

To capture “national attention” toward this mass shooting, I used Google Trends data to track web search frequencies for two sets of searches: “gun control” and the name of the location of the mass shooting. In addition to doing this for Stoneman Douglas, I gathered similar data (using the gtrendsR R package) for all other mass shootings that were in the top 10 most deadliest–including Stoneman Douglas, this amounted to seven mass shootings.

Below are two graphs showing the trajectories for both search terms. For each graph, search volume is placed on a 0-100 scale (where 100 represents the highest volume). First, I show searches for gun control seven days before and six days after each of the seven mass shootings:


Each event follows a very similar path. Before Stoneman Douglas, four of the six saw a spike in public discussion about gun control followed by a dramatic decline into obscurity. The trends following the Sandy Hook and San Bernardino diverged from this pattern, as even about a week after these shootings, debate about gun control persisted. The Stoneman Douglas shooting has followed the trajectory of these latter two events: after falling a bit from its peak, gun control debate–as measured by Google searches, which is a serviceable but not perfect proxy–has persisted in the following week. Moreover, six days out, attention toward gun control in the aftermath of Stoneman Douglas eclipsed that after Sandy Hook and San Bernardino.

A similarly distinctive trend for Stoneman Douglas materializes in the following graph as well, which plots web searches for the shooting location name in the two weeks following the shooting:


In nearly every case, the two-week aftermath saw the shooting quickly fall off the radar. In most cases, it took just a matter of days for public attention to dissipate. Interestingly, for the five days after this most shooting, it seemed like Stoneman Douglas was following this same trajectory. But within the last few days (Days 6, 7, and 8 on the graph), attention toward the Stoneman Douglas shooting has reversed its descent to obscurity, and instead has started to receive renewed attention (now on an upward trend). The distinctive post-tragedy trajectory for Stoneman Douglas–maintaining national attention and spurring gun control debate more than usual–is fairly clear by now, and perhaps owes to the role that the school’s students have played at the center of the national debate on gun control in the week following the tragedy.

Tracking National Attention toward Mass Shootings with Google Trends Data

Vote Validation and Possible Underestimates of Turnout among Younger Americans

Vote validation data appended onto survey data is incredibly valuable. Due to the propensity of individuals to lie about whether or not they voted in an election, self-reported turnout is unreliable. Moreover, as that linked Ansolabehere and Hersh 2012 paper shows, this overreport bias is not uniform across Americans of different demographic characteristics, which further precludes any credible use of self-reported turnout in surveys. Checking self-reported turnout against governmental records of whether or not individuals actually voted provides a much more accurate (though not flawless) measure of whether or not someone really voted in an election. I mention that it’s not without flaws because in order to create this metric–validated turnout–respondents to a survey need to be matched to the voter file (each state has one) that contains turnout information on them. This matching process does not always go smoothly. I explored one case of that in my last post (which has since been fixed). Another potential issue was raised on Twitter by political scientist Michael McDonald:

Aside from the topic of this specific discussion, McDonald is making an important broader point that survey-takers who move (have less residential stability) are less likely to be matched to the voter file; even if they turn out to vote, they may not be matched, and thus would show up as non-voters on surveys with vote validation. Younger individuals tend to move more, and so this flaw could impact them most.

I thought it might be interesting to check for evidence of such a pattern with CCES vote validated turnout by age, and compare those estimates against another commonly used data source to study turnout among different demographics: the Current Population Survey (CPS). For the latter data, I pulled two estimates of turnout from McDonald’s website: 1) CPS turnout with a Census weight (which I’ll refer to as “CPS Turnout”) and 2) CPS turnout with a Census weight and a correction for vote overreport bias (which I’ll refer to as “Corrected CPS Turnout”), more detail on which can be found here. I end up with three turnout estimate sources (CCES, CPS, Corrected CPS) across four age groups (18-29, 30-44, 45-59, 60+), all of which I graph below. The key comparison is between CCES turnout and the two CPS turnout estimates. As McDonald describes, the correction to the CPS turnout is important. Therefore, I pay special attention to the Corrected CPS metric, showing the difference between CCES and Corrected CPS turnout estimates in red above the bars for each age group.


These surveys use very different sampling and weighting procedures, so, on average, they likely produce different estimates. If these differences are constant across each age group, then there is likely nothing going on with respect to the movers/youth turnout underestimate theory. However, the difference–the (CCES – Corrected CPS) metric in red–does in fact vary by age. Most vividly, there is no difference in turnout estimate between these two metrics at the oldest age group, for Americans 60 and older. Each metric says about 71 percent of those age 60+ turned out to vote in 2016. However, for each younger age group, CCES vote validated turnout is smaller than the Corrected CPS one. The largest difference (a 12.4 point “underestimate”) curiously appears for the 30-44 age group. This result doesn’t fall seamlessly in line with the youth turnout underestimate theory–which would suggest the younger you go in age group, the larger the underestimate becomes. But the lack of underestimate for the oldest age group–almost surely the most residentially stable of the age groups–compared to underestimates between five and 13 points for the younger age groups is very telling.

I would need to find data on residential mobility/rate of moving by age group in order to confirm this, but it does seem the most likely to move–the youngest three age groups–see a greater difference between a turnout score built from vote validation and a turnout score that doesn’t use vote validation, the CPS. If that’s the case, I think the theory of vote validation missing some movers and thus likely younger Americans (who are actual voters)  is convincing. This notion would fall in line with takeaways from past research similarly looking at the ties between movers, age, and political participation. Thus, the results here shouldn’t be too surprising, but this possible underestimate of youth turnout is something researchers should keep in mind when using surveys that include vote validated turnout, like the CCES. Regardless, this represents just one (potential) drawback amid an otherwise extremely useful dataset for studying political behavior. Every survey has its flaws, but few have a measure of vote validated turnout, which will always prove more reliable than self-report turnout metrics found in typical surveys.

Vote Validation and Possible Underestimates of Turnout among Younger Americans

Turnout Underestimates and Voter File Match Rate Problems in the 2016 CCES

In versions of the Cooperative Congressional Election Study before 2016, vote validated turnout was consistently higher than actual turnout across states. Grimmer et al. 2017, for example, show this phenomenon here in Figure 1. Matching CCES respondents to individual state voter files to verify whether they voted using governmental records gives a more accurate picture of voter turnout, but the CCES–as with nearly all other surveys–still suffers from a bias where those who take the survey are more likely to have voted than those who did not take it, all else equal.

However, this trend took a weird turn with the 2016 CCES. Unlike the typical overrepresentation of individuals who voted in the CCES, the 2016 version seems to have an underrepresentation of voters. The below graph shows this at the state level, plotting actual voter eligible population (VEP) turnout on the x-axis against CCES vote validated turnout on the y-axis. The closer that the points (states) fall on the 45-degree line, the closer CCES vote validated turnout approximates actual turnout at the state level.


The line of best fit in red clearly does not follow the 45-degree line, indicating that CCES vote validated turnout estimates are very far off from the truth. For comparison, I did a similar plot but for vote share–state level Democratic two-party vote share in the CCES vs. actual two-party vote share:


This result should suggest that it’s not that state level estimates of political outcomes from the CCES are wholly unreliable. Rather, the problem is more specific to state level turnout in the CCES, which Grimmer et al. 2017 stress. That still doesn’t address the switch from average overrepresentation to underrepresentation of voters from 2012 to 2016 in the CCES. In particular, regarding the first graph above, a set of seven states–at around 60-70 percent actual turnout but at around 25 percent CCES turnout–were very inaccurate. I plot the same relationship but change the points on the graph to state initials to clarify which states make up this group:


CCES turnout estimates in seven Northeastern states–Connecticut, Maine, Massachusetts, New Jersey, New Hampshire, Rhode Island, and Vermont–severely underestimated actual turnout. The below table gives the specific numbers on estimated turnout from the CCES, actual turnout, and deviation of CCES turnout from actual turnout (“error”) across these seven states:


On average, CCES turnout in these states underestimated actual turnout by 38.1 percentage points. It is very unlikely that the CCES just happened to sample many more non-voters in these seven states, which marks one explanation for this peculiar result. Another more likely explanation concerns problems with matching CCES survey respondents to the voter file, as Shiro Kuriwaki suggested to me. This turns out to be the likely source for the egregious error. Catalist, a company that manages a voter file database and which matched respondents from the CCES survey to the voter file, had very low match rates for respondents from Connecticut (40.7 percent match rate), Maine (35.6), Massachusetts (32.2), New Jersey (32.1), New Hampshire (38.2), Rhode Island (37.2), and Vermont (33 percent). The below graph illustrates how this affects turnout estimates:


Catalist match rate (the percentage of survey respondents that were matched to the voter file) is plotted on the x-axis, and the difference in CCES turnout and actual turnout (i.e. error) is plotted on the y-axis. These two variables are very closely linked, and for an obvious reason: the CCES treats respondents that are not matched to the voter file as non-voters. Inaccuracies with turnout estimates in fact reflect inaccuracies with voter file match rate. This weird pattern in 2016 is not about overrepresentation of non-voters in the seven specific states but rather about errors in properly carrying out the matching process in those states. The under-matching issue has received attention from CCES organizers and it appears it will be corrected soon:


What’s still strange is that even after ignoring those error-plagued seven states, you don’t observe the usual overrpresentation in the remaining states without a clear matching problem. Many are close to the 45-degree line (that indicates accurate survey turnout estimates) and fall on either side of the line, with more still under the line–suggesting that in several states, the CCES sampled more non-voters than it should have. The estimates remain close to actual turnout, but I still think this is unusual compared to the known consistent overrepresentation of voters in past CCES surveys (again, see Figure 1 here). Perhaps lower-than-usual voter file match rate–while not to the same degree as in the seven Northeastern states–also contributed to a lower than expected CCES vote validated turnout across many other states. However, it could also be that voter/non-voter CCES nonresponse bias occurred to a smaller degree (and even flipped in direction for some states) in 2016.

Update 2/10/18:

It looks like this issue in the CCES has been fixed and the corrected dataset has been posted to Dataverse.

Update 2/14/18:

I re-did the main part of the analysis above with the updated CCES vote validation data. As the below figure plotting actual turnout against CCES turnout shows, considerable less error results. I calculate “error” as CCES turnout rate minus actual VEP turnout rate. The average error is +0.57 points, ranging from -10.8 (the CCES underestimating turnout) to +10.8 (overestimate), and the half of all states have lie between an error of -3.95 and +5.38.


Turnout Underestimates and Voter File Match Rate Problems in the 2016 CCES

Leftovers from “Democrats Are Changing Their Minds About Race, and the Youth Are Leading the Way”

Here is some additional analysis and information for an NYMag piece that Sean McElwee and I wrote.

We used Voter Study Group panel data to track changes in racial attitudes towards blacks–as measured the traditional racial resentment battery–over time. Below is a more standard cross-sectional approach that doesn’t exploit the panel structure, showing same-year racial resentment levels among Democrats vs. all Americans in 2011 and 2016.


The key takeaway above is the shift toward more racially liberal attitudes occurs among all Americans, but happens at a faster rate among Democrats.

Then, we checked whether certain demographic characteristics were most associated with this racial liberalization trend among Democrats specifically. To do so, we restricted our sample–already formed a sample of 8,000 Americans interviewed in both 2011 and 2016–to only respondents who identified as Democrats in 2011 and 2016. Thus, we’re following the same group of consistent Democrats and seeing what other characteristics predict change to more liberal racial attitudes.


The demographic most strongly associated with this change turned out to be age, as the above graph shows, which calculates a net agreement level for each racial resentment item across three key age groups. We see that the youngest individuals–those age 17-29 during the 2011 survey–show the greatest shifts toward more liberal racial attitudes.

We also checked to see if this held up in a multivariate model that accounted for other demographic attributes of respondents. Specifically we regressed a dependent variable–indicating whether a respondent shifted from a non-liberal racial attitude in 2011 to a liberal racial attitude in 2016–on a few key demographic variables. As an example, I’ll describe the components that went into Model 1 from the table below, modeling the battery item that asked whether people agreed that blacks have gotten less than they deserve over the last few years:

  • Dependent variable: Our outcome is a 1/0 indicator. For this particular battery item, agreement (strongly or somewhat agree) represents a liberal racial attitude. Thus, to capture shift towards a liberal racial attitude, this variable takes on a value of 1 if a respondent answered anything other than agreement in 2011 AND said they agreed with the statement in 2016, and 0 otherwise. I’ll note a couple of other points. First, ordinary least squares regression produces the same results as logistic regression, so we stick with OLS as a matter of interpretability. Second, using a binary variable here means we ignore degree of agreement (i.e. we treat “strongly” and “somewhat” agree the same) which still could be important. This is a tradeoff we make, where we place greater value on a simple measurement of an attitude switch. I may try some different modeling strategies that capture degree of agreement–I’ll update this post whenever I get around to that.
  • Independent variables: The predictors here are race (Non-white race with whites as the baseline), age (Age 30-54 and Age 17-29 with Age 55+ as the baseline), gender (Female with Male as the baseline), and education (College grad with Non-college grad as the baseline).


Comparing the size and significance of the coefficients here indicate that the youngest Democrats (Age 17-29) are shifting their racial attitudes in the liberal direction the most. Importantly, the strength of the relationship holds when controlling for other potentially important variables, like race and education.

Update 2/2/18:

I ran the same models but with a continuous (rather than binary) racial attitudes scale as the dependent variable. Results from before hold, as the youngest age group drives overall racial liberalization among Democrats the most. For each item and for each wave (2011 and 2016), I created a 1-4 scale out of the agree/disagree four-point Likert scale, where 4 always represented the most liberal racial attitude and 1 always represented the most conservative racial attitude. I then used the difference between the 2016 scale and 2011 scale to create the outcome measure (indicating racial attitude change in the liberal direction). Below is a plot of the coefficients from the same multivariate regression from above except for the dependent variable which is now this new continuous measure. As an example of how to interpret this result, for the “deservemore” item, Democrats of age 17-29 grew 0.36 points more racially liberal than Democrats of age 55+.


Leftovers from “Democrats Are Changing Their Minds About Race, and the Youth Are Leading the Way”

Does Approach to Coding Party ID Produce Different Over Time Pictures of Partisanship Stability?

Do different approaches for constructing partisanship distribution out of the traditional 7-point party ID scale on a survey result in different pictures of over time partisanship stability? That’s a small question I had after reading a recent Pew Research report on weighting approaches for non-probability opt-in samples. The analysis involved considering weights for political variables, the most important being party identification. Pew used a partisanship benchmark built from a certain approach to treating a party ID variable: coding Independent leaners (those who say they are Independents when first asked but admit that they lean toward a party upon a follow-up question) as Independents, and not as members of a party to which they lean. Usually, this decision is problematic, as these leaners overwhelmingly resemble regular partisans in terms of voting proclivities, ideological self-identification, and issue positions, as I discuss in a past blog post. Given this evidence, I was curious in the decision Pew made in constructing a party ID weighting benchmark that treated leaners as Independents.

Additionally, I wondered if their caution about partisanship weighting due to over time change in partisanship distribution (see page 27) might be shaped by their treatment of Independent leaners. For example, their own data shows–specifically as of late–that an approach of grouping leaners with their parties produces a more stable over time portrait of partisanship than an approach of leaving leaners ungrouped and as Independents. To shed light on this question, I turned to three major surveys that provide over time measurement of the public’s partisanship: the American National Election Studies (ANES), the General Social Survey (GSS), and the Cooperative Congressional Election Study (CCES). Though not all of them extend back as far as the ANES does, for example, trends by survey should be informative. Most importantly, in the below graph, I calculate over time partisanship distribution for each survey year cross-section by survey source and approach to handling Independent leaners–grouped (with parties) or ungrouped (left as Independents). I also compute the standard deviation in over time partisanship measurements by survey source and leaner coding approach, which I interpret as an indicator of variability. (One note: because each survey encompasses varying amount of years, comparisons of SD’s should be made between leaner coding approaches within surveys, not across surveys in any way.)


Data from the ANES provides evidence in favor of the suspicion I had–that coding leaners as Independents inflates the over time variability in partisanship that Pew worries about. While the grouped leaner approach results in a 3.45 SD, the ungrouped leaner approach results in a 5.03 SD. In other words, this approach produces an over time portrait of partisanship that has much more variation than the grouped leaner approach. An implication here could be that if researchers want to weight on party ID but are worried about its variable nature, using the grouped leaner approach is safer.

However, evidence from the GSS and CCES surveys offer evidence in the opposite direction. In the GSS case, the SD is larger for the grouped leaner approach (4.49) than for the ungrouped leaner approach (4.26). In other words, GSS time series data suggests coding leaners as Independents results in a more stable picture of over time partisanship. Likewise, the CCES data would imply the same conclusion, as the grouped leaner approach is more variable (SD = 2.80) than the ungrouped leaner approach (SD = 2.54)

In sum, I cannot really draw concrete conclusions about the best approach to constructing a party identification benchmark on the grounds of choosing how to code Independent leaners. It’s worth noting that the difference in variability–as measured by the standard deviation–is largest in the ANES case, which shows the grouped leaner approach offers the most stable partisanship metric. Still, while not to the same degree in the other direction, evidence from the GSS and CCES support the opposite takeaway. At the very least, though, I can conclude that there does not appear to be a difference in over time partisanship stability that results from different coding decisions. Using a party ID benchmark wherein leaners are ungrouped does not exaggerate over time partisanship variability as I thought it might have–at least not in a consistent manner. This of course is a very simple analysis, but it seems like leaner coding is not too much of a problem for partisanship benchmark construction. At the same, it’s worth keeping in mind that in almost all other cases, researchers are better off sticking to the grouped leaner approach.

Does Approach to Coding Party ID Produce Different Over Time Pictures of Partisanship Stability?

Clarifying the Relationship between Partisanship and 2016 Vote Choice with Panel Data

When looking at polling crosstabs with party identification breakdowns of key variables like vote choice and approval ratings, many often conclude a strong impact of partisanship. This data shows that Democrats largely express support for Democratic candidates in elections and leaders, while Republicans express support for their candidates and leaders. Yet conclusions about this relationship could suffer from the issue of endogeneity (I’ve discussed this before here). Inferences about how partisanship affects an outcome Y need to assume stable partisanship, and that the outcome Y–such as vote choice–does not in turn affect partisanship. The possibility of reverse causation would mean that people first arrive at a decision to vote for Donald Trump, for example, and subsequently update their partisanship to match with their candidate preference. If this occurs, then the previously assumed “exogenous” nature of partisanship as an unmoved predictor becomes dubious. Maybe individuals who were originally Republicans but did not support Trump changed their partisanship, and as a result, partisanship became a mere reflection of vote choice, and not a stable underlying predisposition in a way that would be meaningful.

A similar concern has been raised regarding approval rating polls showing strong intra-party support for Trump. Original party base members may no longer identify as Republicans on surveys, and thus Republican party identification simply comes to mean support for Trump and not a meaningful underlying political trait. Cross-sectional survey data cannot overcome this problem, as it lacks a measure of an individual’s preexisting partisanship. Panel data, on the other hand, can better address this issue. Cross-sectional data uses contemporaneous 2016 measures of partisanship and vote choice (recorded at the same time) to say that 90 percent of Republicans voted for Trump. But the better approach would be to use a pre-2016 measure of partisanship–unaffected by Trump–and calculate how vote choice breaks down along this variable that better represents an underlying indicator of partisanship. Publicly available panel data–with waves in December 2011, November 2012, and December 2016–from the Voter Study Group (VSG) offers such an improvement in capturing party voting (the rate at which partisans vote for co-party candidates).

Specifically, I can compare how party voting looks like using both a 2011 measure of partisanship (before both the 2012 and 2016 elections) and a 2016 measure of partisanship (that is purportedly endogenous to 2016 vote choice). If there are large differences in party voting across these measures, then an endogeneity problem exists, suggesting that partisanship is shaped in response to vote preference. If small or no differences result, then partisanship constitutes a more exogenous variable–in line with the stable over time character that much of political science literature (and other evidence) suggests.

The VSG data offers mixed evidence but mostly lies in favor of the latter conclusion. I’ll start with the broadest perspective–overall rates of party voting: Republicans voting for Trump and Democrats voting Clinton as a percentage of all partisans who reported voting in 2016. If I use the 2016 measure of partisanship, I find that 89 percent of partisans voted their party in 2016. If I use the 2011 measure of partisanship, 84 percent of partisans voted their party in 2016. Thus, it appears that a small percentage of people shifted their partisanship to match their vote preference in 2016, and in a way that would slightly inflate the impact of partisanship on voting. But it’s fairly small, as the party voting rates remain similar.

Breaking these party voting rates by party and candidate reveals a similar picture, but with some additional information. The 2016 measures of partisanship suggest very high rates of party voting, with 90 percent of Democrats voting Clinton and 88 percent of Republicans voting Trump. That rate declines a bit when using a pre-Trump (2011) measurement of partisanship: 83 percent of original Democrats opted for Clinton, while 84 percent of original Republicans went Trump. These percentages are not that different, but at least some partisanship updating is likely at play. What’s more interesting is how this pre-Trump underlying partisanship better captures defection from the Democratic Party (in a way that–not balanced out by similar defection among Republicans to Clinton–could have tilted the election just enough to Trump). If we use a 2016 party measure, then we would conclude that seven percent of Democrats voted Trump. Using the 2011 measure of the original Democratic party base, however, nearly doubles that size, revealing 13 percent of Democrats who voted for Trump in 2016. This panel approach can thus offer a more meaningful estimate of how many original Democrats defected from the party in voting for the out-party candidate in 2016.

Finally, I wanted to further break down this comparison by using the full seven-point party identification scale. The below plot shows how each party identification group (of the seven in total) voted in 2016 when using an individual’s 2016 reported partisanship.


Differences in party voting rates at this partisanship subgroup level appear when comparing the above plot to the same breakdown but with an individual’s 2011 reported partisanship, as the below plot illustrates:


If we use 2016 cross-sectional data, then we end up with an overestimate of how closely “Strong Democrats” and “Not very strong Democrats” adhered to their party affiliation for deciding whom to vote for. This approach would say that 97 percent of strong Democrats and 79 percent of weak Democrats voted their party in 2016, as opposed to 90 and 70 percent respective rates when using underlying (2011) partisanship. Similar differences appear on the Republican side, but as mentioned earlier, the magnitude is smaller. The biggest difference is for the “Lean Republican category,” as while a 2016 measure suggest 90 percent of this group went Trump, a metric capturing original members of this category suggests 83 percent did.

In sum, this comparison does suggest some partisanship updating to accord with vote choice took place, but not to any large extent. Concerns with endogeneity should be tempered. That’s in large part because partisanship remains a very stable over time variable–at both the aggregate and individual level. To underscore this latter point, I used all three survey waves of the VSG (bringing in the 2012 wave that I’ve exclude up until now) to track individual level partisanship dynamics at three different points in time over a five-year span. That results in the following table, which shows the distribution of VSG survey respondents by the different possible party ID combinations they can have across the three survey waves. In each wave, they can express three different partisan affiliations, which makes for 27 unique combinations (3*3*3 = 27).


Two combination groups stand out: people who identified as Democrats in all three years (41.74 percent of all respondents) and people who identified as Republicans in all three years (33.79 percent). That means about three out of every four people (75.53 percent to be exact) are consistent partisans over the course of four years. The next most common group is people who do not reveal any partisan leanings–Independents–which makes up 6.07 percent of all respondents. Thus,  81.60 percent of all people have a consistent expression of partisanship (or lack thereof) across five years at three different points in time, a piece of evidence indicative of strongly stable partisanship.

Moreover, looking further down the table, only 4.64 percent of respondents ever identify with both parties at some point during the three survey waves. Rather, most of the party switching–which is very little to begin with–is into and out of the Independent category (this movement takes up 13.76 percent of all survey respondents). In light of these trends, the lack of substantial endogeneity–changes in partisanship driven by vote choice selection–should not come as much of a surprise.

Clarifying the Relationship between Partisanship and 2016 Vote Choice with Panel Data