A Long Horizon: Some Observations About Party Identification Over the Past 25 Years

Pollsters famously maintain their findings represent attitudes as “a snapshot in time.” But, examining data points over a long period provides a great deal of insight into trends in public opinion. One of the most significant measures of public opinion is voters’ partisanship, and trends here can portend a lot, so we looked at party identification in every January from 1990 through 2015.

Since the George H.W. Bush Administration, the NBC–Wall Street Journal poll has been measuring party identification with the following question: “Generally speaking, do you think of yourself as a Democrat, a Republican, an Independent, or something else?” Respondents are asked a follow-up question. Those who respond either “Democrat” or “Republican” are asked: “Would you call yourself a strong (Democrat/Republican) or not a very strong (Democrat/Republican)?” Those who answer the first question “Independent” are asked: “Do you think of yourself as closer to the Republican Party, closer to the Democratic Party, or do you think of yourself as strictly independent?”

There are two ways to measure party identification. The first is without leaners; that is, including only those who say they think of themselves as a Democrat or a Republican and excluding Independents, even those who lean toward a political party. The second is adding the leaners to the Republican and Democratic totals. To track party identification over time, we looked at the self-professed partisanship on the first survey in January of each year. Examining 25 years of party identification in January, the trends are clear:

  • Party identification is not fixed and fluctuates over time.
  • The percentage of adults identifying as either a Democrat or a Republican, excluding leaners, has been steadily dropping.
  • That said, Democrats have enjoyed a clear and consistent advantage except immediately after Newt Gingrich’s 1994 Republican Revolution and for the few years immediately following the September 11th attacks.

Key Findings

Party identification is not static. Party identification is a reflection of respondents’ attitudes surrounding the political environment; it is not a demographic measure that is locked in place, but rather an often-fluid attitudinal measure of how voters respond to the politics of the day.

  • Democrats have enjoyed advantages over Republicans as high as 14 points without leaners or 17 points with leaners (1993). Only once—in 2004 without leaners—have we measured a Republican advantage in January, and even then it was only by one percentage point.
  • Averaging partisanship over presidential terms, President Obama has presided over the largest partisan gap.
  • The percentage of adults identifying as Republican (excluding leaners) dropped during President Bush’s second term from 27{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} in January 2005 at his second inauguration to 21{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} in January 2009 when he left office. It has remained within two points of 21{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} every January since.

 chart 1

Fewer adults are identifying as Republicans or Democrats. Over the last 25 years, the percentage of adults saying they think of themselves as either a Democrat or a Republican has fallen from a decade-long average of 58{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} in the 1990s to only 50{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} so far this decade, excluding leaners. This year’s partisanship was the second-lowest January number we ever recorded (2010 was lowest), as only 47{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} of adults said they think of themselves as a Democrat or a Republican.

 chart 2

Using merge data since 2010 helps us better understand party identification. Since 2010, we have been merging together all the registered voters by year and studying the party identification of many different sub-groups. This gives us large sample sizes (N = 10,805 in 2010; N = 11,546 in 2012; and N = 8,963 in 2014) and small margins of sampling error for the total sample (between +0.91{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222} and +1.04{09f965da52dc6ab4c1643a77bd40d1f729d807040cd8db540234bb981a782222}, though error is larger for the sub-groups). The topline data with leaners has not changed much since 2010.

chart 3

Looking at the merged data also allows us to see with confidence which sub-groups are most Republican (e.g. Tea Party supporters, conservatives, and White Southerners) and which are most Democratic (e.g. liberals, African Americans, and Latinos).

 chart 4

We can also examine changes in attitude over time. Even though party identification was essentially the same in 2010 and 2014, there have been interesting changes below the topline numbers that reflect shifting political sentiment. Two of the most interesting shifts have been by gender and by region.

The gender gap was noted as long ago as the 1980 presidential election. It continues to remain important in American politics and is certainly not settled or static. Looking at our merged data sets from 2010 through last year, the gap has continued to widen, as men have trended more Republican while women trended more Democratic.

chart 5

We also see regional differences. The battleground-heavy Midwest, home to six of the 11 states decided by single-digits in the 2012 election, has shifted from a being slightly Democratic to breakeven. Meanwhile, the West, home of three states decided in 2012 by single-digits, has become more Democratic.

chart 6

Click here to look at party identification with leaners by sub-group in 2010, 2012, and 2014 among 48 different sub-groups.



Polling for NBC and The Wall Street Journal is conducted by Republican pollster Bill McInturff and Democratic pollster Fred Yang. This analysis is our own and does not necessarily reflect the views of NBC, The Wall Street Journal, or Hart Research Associates.


Public Opinion Strategies