Friday, July 15, 2016

Convention Bumps

As we head into the convention phase of the campaign, I thought it would be useful to review the extent to which the convening party candidate receives a bump in the polls during their convention.  I've covered much of this in previous posts so I'll just briefly review a few key points here.

The figure below summarizes the size of convention bumps from 1964 to 2012.  The convention bump is measured as the percentage point change in the convening party's share of the two-party vote, comparing polls taken between six days and two-weeks prior to the convention with polls taken during the seven days following the convention.  Note that this is a short-term measure of the convention bump and does not say anything about the rate of decay in the weeks following the convention. 

There is a lot going on here but there are a few takeaway points.  

1.  Candidates generally get a bump of some sort.  The size of the bump is highly variable but virtually all candidates leave their convention doing better in the polls than when they went into the convention.

2.  The size of the convention bump does not predict the overall winner very well.  Just ask Presidents Goldwater,  Mondale, Dole, or Gore, all of whom had bigger bumps than their competitors.  One of the reasons for this is that candidates running way behind in the polls have an easier time gaining ground during their conventions.  For instance, in 1964 Barry Goldwater was so far behind in the pre-convention polls (averaging 21% of the vote) that it was easy for him to improve his standing by thirteen points during the convention, though he still never come close to being competitive.  At the same time, Lyndon Johnson went into his convention with 69% of the vote in pre-election polls and left the convention with no bump but still with a substantial lead.

3.  Convention bumps aren't what they used to be (see figure below).   Prior to the 2000 election, convention bumps averaged more than six points, but that has fallen to just over two points from 2000 to 2012.  One potential explanation for this change lies in the scheduling of conventions. The 2008 and 2012 conventions were held in late August and early September and were also held on back-to-back weeks.  The norm in other years had been to hold the conventions in late July or early August and to separate them by two to three weeks.  What is probably most important here is holding the conventions on consecutive weeks, which means that the convention messages end up overlapping and may cancel out each other.  Another potential explanation lies in the increased polarization of the electorate.  It is possible that partisans are so much more committed to their candidates now than they were before and there is a much smaller persuadable electorate that can be influenced by events like the nominating conventions.

So what does all of this mean for the 2016 convention bumps?  One of the the key features of the 2016 conventions is that they follow the recent scheduling trend of back-to-back convention weeks, so this might limit the size of the bumps.  On the other hand, since the conventions are being held in late July rather than late August, there might be more persuadable voters than in 2000, 2008, or 2012.   One other factor that could lead to more substantial convention bumps is that both candidates have problems within their own party and the conventions present them with an opportunity to rally the base in a way that no other campaign event can.  The key for both candidates is to have a smoothly run convention that heals rather than exacerbates existing party wounds and projects a positive message to the rest of the country.  

Wednesday, June 8, 2016

The Immigrant Vote and the 2016 Election

Donald Trump's recent attack on Judge Gonzalo Curiel's impartiality because of Curiel's Mexican ancestry brings to the fore the potential role immigrant populations might play in the upcoming presidential election.  Potentially compounding Republican problems on this front, a recent Washington Post article describes a substantial uptick in citizenship applications in the first quarter of 2016, pointing to the anti-immigration tone of the Trump campaign as a likely source of the increase.

Several related issues are addressed in a number of places in my new book, Altered States.  Some of the relevant findings are summarized below.

The Changing Immigrant Population

Chapter 3 addresses a number of different aspects of population migration (both immigration and internal migration) and presidential election outcomes.  One important finding on this front is that, as a group, naturalized citizens have changed in many politically relevant ways over the past forty years.  Whereas naturalized citizens used to "look" a lot like the native-born population in terms of race, ethnicity, party, and ideology, they are are now overwhelmingly non-white and have grown substantially more Democratic and liberal than the native-born population.  These changes make the distinction between immigrant and non-immigrant voters increasingly politically important.

Size of Immigrant Population and Changes in Party Support

The figure below is adapted from Chapter 3 and illustrates the relationship between the size of the naturalized population and changes in party support in the states in presidential elections over the past forty years.  The dependent variable is the change from 1972-1980 to 2004-2012  in the expected Democratic share of the statewide two-party presidential vote (purged of home-state and southern regional effects) centered on the Democratic share of the national two-party vote, and the independent variable is the percent of the  state's citizen voting-age population (CVAP) in 2008 who were foreign born.

Adapted from figure 3.2 of Altered States
There is a clear positive relationship between percent foreign born and change in Democratic vote share, albeit with a bit of a curvilinear pattern.  At low levels of foreign born population there is a mix of political outcomes, with Democrats making gains in some states and Republicans making gains in others.  However, Democrats gained ground in virtually all states with an above average level of foreign born population.  The only exceptions to this are Alaska and Texas (so much for Texas turning blue).  Otherwise, the entire lower right corner of the graph--high foreign-born population, coupled with Democratic losses--is empty.  

All of the states with the highest levels of foreign born population (California, Florida, Hawaii, Illinois, Nevada, Massachusetts, New Jersey, and New York) are also among the states where Democrats made their greatest gains.  It is worth noting that many of these states (California, Illinois, New Jersey) used to be fairly competitive but are now clearly in the Democratic camp, while others (Florida and Nevada) have moved from leaning Republican to being fairly competitive now.

The Increasing Importance of the Immigrant Population

Another of the interesting findings from Altered States (Chapter 5) is that, over time, all manner of state-level demographic characteristics have become increasingly closely connected to presidential election outcomes in the states.  As the figure below illustrates, this is certainly the case with the relationship between percent foreign born and party support in the states:
With the exception of the 1972 election, there was little-to-no relationship between the relative size of foreign born population and presidential outcomes in the states in the 1970s and 1980s, but the relationship has grown in strength steadily over time.  The average election-year correlation was .18 from 1972 to 1980, .29 from 1984 to 1992, .50 from 1996 to 2000, and .56 from 2004 to 2012.  Overall, the strongest correlation was .60, in 2012.

Swing States

The electoral impact of a growing and increasingly Democratic foreign-born electorate likely depends on its geographic distribution, playing a more pivotal role in swing states than in solidly Democratic or Republican states.  Among a group of seven swing states (neither party has more than a two-point advantage) identified in Chapter 6 of Altered States, only one--New Mexico--is among the states with the slowest growth in foreign-born CVAP from 2000-2012, but three swing states--Florida, Nevada, and Virginia--are among the group of states with the fastest growing foreign-born CVAP in the same period.  In a close national contest, these patterns of immigrant population growth could play a key role in determining the Electoral College outcome.

It's Not Just Latinos

Finally, it should be pointed out that while many of Donald Trump's more notorious statements about immigration have focused on Mexican and Latin-American immigrants, the naturalized citizen population (immigrants who can vote) is comprised of just about as many immigrants from Asia (37%) as from Latin America (39%).  This doesn't let Donald Trump or the Republican Party off the hook, politically, however, since Asian-Americans hold overwhelmingly negative views toward Donald Trump and the Republican Party and supported President Obama at slightly higher rates than Latinos did in 2012.

Monday, May 16, 2016

Altered States

I'm please to announce that my new book, Altered States: Changing Populations, Changing Parties and the Transformation of the American Political Landscape, is coming out soon (shipping date set for May 30) with Oxford University Press.  (You can get an early look inside on the Amazon page.)

The book focuses primarily on explaining changes in party support in state-level presidential elections outcomes over time (1970s to 2010s), using two different but related explanatory  frameworks: a compositional model that sees changing demographic characteristics in the states as the source of political change; and a contextual model that focuses on changes in the relationships between state characteristics and election outcomes--in response to elite polarization-- as the source of change.  The analysis supports both models, though the compositional model does a better job of accounting for political change over time.

The book addresses a number of important topics including partisan advantages in the Electoral College, the efficiency of vote distributions across the states, geographic and demographic sorting, population migration, elite polarization, and many other related issues.  I will be posting a few things in the next couple of weeks to highlight some of the findings.

Tuesday, September 16, 2014

Another Look at Electoral Change

Last fall, I posted some data on the changes in Democratic and Republican support in the states in presidential elections, in part in response to observations made by a number of pundits and politicians that there is an increasing Democratic "lock" on the Electoral College.  As I stated then, this sort of chatter even included speculation that Texas could "turn blue" in the not-so-distant future.  Indeed, this idea was recently repeated by Rand Paul, who cautioned Texas Republicans that they need to soften their approach on issues such immigration, or Texas "will be a Democratic state within 10 years."

I return to the issue of electoral change in this post to provide a somewhat clearer visualization of the trends, as well as a more precise account of their electoral impact. 

The strategy used in the previous post focused on plotting the trends in party support from 1972-2012, separately, for each state (see Democratic gains here and Republican gains here).1  I use the same data here but approach the analysis somewhat differently, focusing on the overall change in the centered Democratic share of the two-party voter from the 1972-1980 elections to the 2004-2012 election.   Specifically, I look at the average expected Democratic vote share (based on the trend over time) for each state in each time period and then compare the two periods to ascertain where the parties gained and lost the most support. Centering the vote around the mean state outcome permits a comparison across states without having to also consider factors (national tides) that affect the level of party support from one year to the next.

The figure below captures both the direction and magnitude of changes in party fortunes in the state-level presidential election outcomes.  In this figure, the ubiquitous chromatic devices of red and blue are used to identify states where the trend over time has favored Republicans (red) or Democrats (blue).  The blunt end of the arrow represents the state average during the 1972-1980 elections, and the point at the head of the arrow designates where the states end up, on average, for presidential elections from 2004 to 2012.  It is important to recall that party support is expressed relative to other states, in each year. For instance, in a Republican blowout year, such as 1972, a state like Minnesota, where George McGovern ran about 10 points better than he did across all states, could still be lost by the Democrats.  Despite McGovern's loss there, Minnesota in 1972 is still a strong Democratic state, relative to all other states.  

Changes in Party Fortunes in State-level Presidential Election Outcomes
Changes in Party Fortunes in State-level Presidential Election Outcomes
So what does this figure reveal in terms of a growing Democratic advantage in the Electoral College?  In very gross terms, there is a an important trend in favor of Democratic presidential candidates. Looking just at the direction of movement (ignoring magnitude), there are 29 states that have seen Democratic gains, and 21 states where Republicans have gained strength.  In terms of electoral votes, the states where Democrats have made inroads control 366 electoral votes, while the states with Republican gains control just 169 electoral votes.  This is a substantively large and meaningful difference.  However, it may overstate the case somewhat, since some states in which the parties gained strength were already in the Democratic or Republican column and only became more strongly partisan; and in a few states where a party gained strength (e.g., Mississippi and Georgia for the Democrats, and Minnesota and Wisconsin for the Republicans), their position is improved but they are still at a distinct disadvantage.  And there are a handful of states where movement was very slight, though on balance in one party's favor.  

It's also worth noting that there is no support for the Texas-turning-blue argument, as Texas has gone from leaning Republican to solidly Republican over the past forty years.

Another way of assessing the importance of partisan change is to focus on cases of conversion (switching from one party to another) or changes in swing status (moving from competitive to non-competitive or from non-competitive to competitive.  The figure posted below highlights states that fall into either of these categories.  The vertical dashed lines are are -2 and +2 percentage points, representing a fairly tight band of competitive outcomes.  There is only one state (West Virginia) that can be described as clearly switching from from one party (Democratic) to the other (Republican), while the remaining twenty states moved into or out of the competitive zone.  Across all twenty-one states whose competitive status changed, six states representing 40 electoral votes moved in the Republican direction, while fifteen states representing 188 electoral votes moved in the Democratic direction.

Changes in Party Support, Highlighting "Swing States" 

From this perspective, there has been clear and important movement in the direction of the Democratic party.  The number of states that have moved through this zone in the Democratic direction (and the number of electoral votes associated with them) improves the Democratic position substantially over the past forty years.  A couple of caveats. First, this is only one way to cut the data and the designation of the competitive zone is admittedly arbitrary (as most such designations would be).  Second, this discussion places a premium on a certain type of change and ignores cases in which parties increased their grip on already friendly states.  To be sure, there are a number of Republican and Democratic states where this has happened, and the Republican party has a slight edge in this category.  

The table below uses data on party support and electoral votes to illustrate more precisely the magnitude of the increased Democratic advantage.  The cell entries in the right two columns are the predicted number of Electoral Votes going to the Democratic candidate based on three different national vote models (mean Democratic votes of 48%, 50%, and 52% across the states), along with the expected levels of support (taken from beginning and end points of the arrows in the figures above) in the states and their associated Electoral votes in the 1970s and the 2010s.

In the current period, Democratic candidates have a distinct advantage in close national contests. If the average state-level vote is 50%, the expected Democratic Electoral Vote count is 319.  If the average Democratic state vote drops to 48%,  Republicans would be expected to pick up Colorado, Florida, Nevada, and Ohio, but Democrats would still have a fighting chance, with an expected Electoral Vote count of 257.  And, of course, if Democrats carry 52% on average across the states, they win a comfortable Electoral Vote margin.  What is most impressive here is not just the Democratic advantage, but how that advantage has shifted since the 1970s, where the Democratic Electoral Vote was much more proportional to the national popular vote.  

It is important to remember that the change in the Democratic advantage is not affected just by changes in patterns of party support across the states but also by changes in the Electoral Votes awarded to the states.  In fact, changes in the distributions of electoral votes have muted the shift in Democratic advantage slightly. For instance, at a 50% average state vote the expected Electoral Vote of Democrats in the 2010s would be 333 if there had been no change in the Electoral College since the 1970s.

There is still no evidence of a Democratic "lock" on the Electoral College, but the data presented here do make a clearer case that Republican presidential candidates face an uphill battle, and that their position has deteriorated over time.  The political landscape has changed appreciably in the last forty years and that change is politically consequential. Of course, all of this raises interesting questions about the causes of the changes in party support, questions I will take up in my next post (soon, I hope).

1See the original post for a more detailed explanation of have the expected state vote is calculated. Note that these figures reflect a correction for the trend in North Carolina.

Thursday, October 10, 2013

Trend in Party Favorability

In light of the new Gallup Poll showing Republican favorability at a historic low point, it seems like a good idea to update a post I did on favorability just before the 2012 election. At that point I  concluded that the Democrats held a clear advantage going into the election, although the public was not wild about either party.  I revisit this issue here, almost one year after the original post

The figure posted below summarizes the "favorability" of both parties, using results from publicly available polls that  asked respondents to rate their feelings toward the parties as either favorable or unfavorable. There are several notable patterns here.  First, throughout this time period (February 2012 to October 2013), Democrats have held a distinct favorability advantage: the mean net rating for the Republican Party is -16.4, while the mean net rating for the Democrats is 1.2.  In fact, there is not a single poll in this series of 44 polls in which the Republican party registered a net positive rating, and not a single case in which the net Republican rating was higher than the net Democrat rating. 
Data are taken from and  The dots represent individual polls, 
and the lines are the smoothed lowess trends in the series.

Even worse for the Republicans is that their favorability rating continues to degrade. Their average net rating was -13 prior to the 2012 election and has averaged -22 since then. Meanwhile, the net rating for Democrats has shifted from and average of +.6 prior to the election to an average of +2.3 since the election.  Theses swings in party fortunes are summarized below, where I plot the overall Democratic advantage in favorability.

Here we see that the somewhat pronounced negative turn for Republican party favorability following the election, coupled with the corresponding smaller positive turn for the Democrats, has led to an impressive shift in the Democratic in advantage in favorability.

Of course, this is not say that people are in love with the Democratic Party.  At best, these data say that people are, on balance, fairly evenly split in their view of the Democrats, with some polls reflecting a negative view of them and others reflecting a more positive view.  In political terms, though, these data show that the Democrats fare very well, at least in comparison to the only other game in town, the Republican Party.

What do these data say about the impact of the government shutdown?  Unfortunately, polls that include party favorability questions are too few-and-far-between to be able to pin much on specific events, and the figures presented here include only one poll taken since the shutdown began, the aforementioned Gallup Poll.  In all likelihood, that poll does reflect the impact the shutdown, but it is only a single observation and it doesn't really have much effect on the overall trend in the data.

Instead, I think it is best to focus on the the general pattern in the data, and I think the trend since the 2012 presidential election probably says a lot about the way politics have been conducted and the way people evaluate the parties in light of that conduct.  It could well be that the Republican congressional strategy in the last year will pay off in other ways, but it is pretty clear that it isn't buying them any favor among the general public.

Thursday, September 26, 2013

States of Change

Not long after the 2012 election some observers noted an increasing Democratic "lock" on the Electoral College. Although Republican nominee Mitt Romney's campaign took a lot of heat for losing election, there has developed a general sense that changing demographics in key states are setting the stage for continued Republican difficulties in years to come. By this account, increasing numbers of Latino and Black voters--both heavily Democratic groups--in some areas have moved enough states to the Democratic column that Republican presidential candidates face an uphill battle in anything other than a bad Democratic year.   And there is every reason to expect this demographic trend to continue. This idea has gained enough currency that some Democrats even see Texas, with its increasing Latino population, turning blue at some point in the future.

My interest here is in documenting the changes party fortunes in the states and trying to get a sense of which changes are of greatest consequence to the race to 270.  Of course, I am not only interested in where Democrats have strengthened their position but also where they have lost ground to Republicans; for there are clearly parts of the country where Democratic prospects have dimmed appreciably over time.

I use state-level presidential election returns from 1972 to 2012 to document the trends in party support. The starting date for this type of analysis is somewhat arbitrary, but using 1972 puts us on this side of the beginning of partisan changes in response to party strategies related to civil rights.  I focus on the trend over time in the centered (around the fifty-state mean) Democratic share of the two-party vote, separately for each state.  Centering the vote allows me to focus on each state relative to all other states without worrying about overall swings in party fortunes from one election to the next.  So the focus is not on which party wins or loses a state, but on support for the Democratic party over time in a given state,  relative to all other states.  To gauge the trend over time I regress vote share on year, separately for each state, and also included dummy variables for presidential and vice-presidential home state advantage, as well as one for southern states in 1976 and 2000.  The southern dummy variable is necessary to capture the unnaturally high level of support for the Democratic ticket in the south in response to the candidacy of former Georgia Governor Jimmy Carter.  I then used the results from the state-by-state regression models to estimate the trend in Democratic support over time.  In doing this I set the values for the dummy variables to zero so the "predictions" reflect the trend over time exclusive of these transitory perturbations.

The figures below document the changes in Democratic fortunes.  In each figure, the solid straight line represents the estimated trend in Democratic support over time, and the points represent the actual election outcomes.  It is important to recall that in some cases the trend line does not appear to fit the scatter plot as well as it "should" because it excludes the effects of home state advantages and the southern advantages in 1976 and 1980 (see MS and GA as exemplars of this phenomenon).  It is also worth pointing out that the observations above the zero point (dashed line) are not necessarily cases in which the Democratic candidate won the state; instead, these are cases in which the Democratic candidate fared better than he did, on average, across the fifty states.

The panel of graphs below displays the pattern of partisan change among those twenty-five states in which the Democrats have seen their greatest gains, ordered by magnitude of gains from upper left (across rows) to lower right.  These states generally fall into four different categories.  First there are those states that moved from somewhat competitive to favoring the Democratic candidate: California, Delaware, Connecticut,  Hawaii, Illinois, Maine, Maryland, New Jersey, New Mexico (very slightly), New York, Vermont, and Washington.  Some of these states leaned slightly Democratic in the early 1970s while others were truly toss-up states; but almost all of them have moved comfortably into the Democratic column.  These changes represent an important net gain for the Democratic party. As a group, these states comprise 166 electoral votes that the Democrats can count on now much more confidently than they could forty years ago.

Twenty-Five States with the Greatest Democratic Gains 
in Presidential Elections from 1972 to 2012

Another important group consists of those states that moved from tilting Republican to being truly competitive: Colorado, Florida, New Hampshire, Nevada, and Virginia.  To be sure, these are not states that the Democratic candidate can count on--they are, after all, competitive--but they are states that used to be much farther out of reach for the Democrats and now are up for grabs.  Together these states comprise 61 additional electoral votes in states that have moved toward the Democratic party.

Finally, we have states that were strongly Democratic and became even more so (Massachusetts and Rhode Island), and several states where the change was either very slight and didn't alter the general outlook for either party (Arizona, Georgia, and Mississippi), or that can best be described as flat-liners, impervious to whatever process has driven the changes in other states (Michigan, Oregon, and South Carolina).  Technically, there has been some change in this latter group of states but it is so slight that it is barely discernible in these plots.

Of course, not all states have trended so favorably toward the Democrats.  The panel of graphs below illustrates the pattern of party change in the remaining states, where Democrats either lost  ground to Republicans or just managed to hold their own.  One state that stands out here is West Virginia, which has shifted from a place where the Democratic candidate typically ran ahead of his performance across the other forty-nine states to a place that now appears to be a long shot for Democratic candidates.   Minnesota is another interesting state.  Although the Democratic candidate continues to fare slightly better in Minnesota than in the rest of the country, there has been a very gradual decline in that advantage.  Democrats continue to win there, but Minnesota is somewhat more competitive now than it was thirty or forty years ago.

Probably the most consequential set of states in this figure are those that have moved from being somewhat competitive to being out of reach for the Democratic candidate: Arkansas, Kentucky, Louisiana, Montana, South Dakota, Tennessee, and Texas (so much for Texas turning Blue--see above).   In strong Democratic years, they were once possible pickups for the Democrats, but they now appear to be out of reach.  As a group these states, along with West Virginia and Minnesota,  constitute 92 Electoral Votes. 

Twenty-Five States the Greatest Democratic Losses (or smallest gains)
in Presidential Elections from 1972 to 2012

Of course, there are a number of states that were fairly Republican in the early 1970s and have become more so over time: Alabama, Alaska, Idaho, Kansas, North Dakota, Oklahoma, Utah, and Wyoming.  And there is also another group of flat-liners, which includes states with very, very slight Democratic decline (Indiana, Missouri, Nebraska, North Carolina, and Wisconsin), and a group with almost imperceptible Democratic gains (Iowa, Ohio, and Pennsylvania).   These states (flat-liners and Republican states that became more Republican) don't really have much of an impact on the on the partisan advantage in the Electoral College.

While the figures above provide a lot of state-by-state detail, the same information can also be summarized using a heat map and the by-now-familiar Red/Blue schema.  The interactive map pasted below shows how states have changed over the the last forty years. 

Changes in the Centered Democratic Share of the Two-Party Vote in U.S. 
Presidential Elections, 1972-2012
(Blue=DemocraticAdvantage, Red=Republican Advantage)

In keeping with data already presented, the dominant patterns over time are purplish states turning blue, followed by fewer purple states turning red, some reddish states turning even redder, and a number of reddish states turning purple.1

The movements over time have clearly favored the Democratic Party. Those states that have moved from somewhat competitive to Democratic, or from leaning Republican to fairly competitive, combine for a total of 227 electoral votes. Among those states that have made similar shifts toward the Republican Party (other than those that were in the Republican column to begin with), the electoral vote count is only 92. Perhaps the worst news for the Republican party is that there is only a single large state that has moved in their direction: Texas, which has gone from leaning slightly Republican to being a strong Republican state. Contrast that with the Democrats, who have seen California, Illinois, New Jersey, and New York move from somewhat competitive to strongly Democratic, and Florida has moved from leaning Republican to very competitive.

Still, there is nothing resembling a "lock" on the Electoral College.  A clear trend toward the Democratic Party, yes, but hardly a lock on anything.
Erratum (September 2014):  There was an error in estimating the slope for North Carolina (control for Carter's southern advantage in 1976 and 1980 was not included).  When that error is corrected, North Carolina emerges as another state in which the Democrats have made smallish gains.  See the updated figure here.
1I think. I'm color blind, so this may look a bit different for you.