How 50 Years of Data May Help Predict Party Nominations
It’s commonly said that winning the Iowa Caucuses can be a bad sign for the future of a presidential campaign. Although such predictions may be attributed to political folklore, it’s true that over the past two election cycles, the Republican primaries in Iowa have successfully predicted the eventual nominee zero percent of the time. However, New Hampshire's Republican primary winners went on to be the nominee 100 percent of the time in that same period. To take an objective look at whether such trends hold true over a longer period of time—and what dynamics may be at play—Analytics@American decided to dig in further.
Presidential primaries are a relatively new phenomenon. While some states began holding primaries much earlier, the binding primary process was not nationally implemented until 1968. We combed through Federal Election Commission data from every presidential primary since 1968 to see which states were most consistently predicting the eventual nominee. After eliminating the data from elections when an incumbent president was running for reelection, as they generally didn’t see any substantial competition, we were left with between seven and nine data points per state. While that may not be enough data to make the findings statistically significant, the data does paint an interesting picture. As the primary system evolves, we’ll have more and more of this data to analyze, and therefore, more accurate results.
In the two bar graphs below—one for the Democratic primaries and one for the Republican primaries—we compared the number of times a primary or caucus was held in the state and, of those, the number of times the winner became the presidential nominee for that party. Click on the image to zoom in and see where your state falls:
Embed Our Graphic On Your Website:
What the Data Reveals
In our analysis of the data, the key question we wanted to answer related to determining which states are most reliably choosing the eventual nominee on each side of the isle. In addition to Super Tuesday, which was held on March 1, the early contests typically receive the most attention. According to the data, states like Iowa predict the correct eventual nominee 50 percent of the time for Democrats and 42.8 percent of the time for Republicans. Here is what the data revealed about other early primaries New Hampshire, Nevada, and South Carolina:
- Iowa—Democrats: 50%; Republicans: 42.86%
- New Hampshire—Democrats: 44.44%; Republicans: 75%
- Nevada—Democrats: 28.57%; Republicans: 71.43%
- South Carolina—Democrats: 37.5%; Republicans: 71.43%
Though there has been a great deal of hype surrounding these early contests, the data suggests that the most accurately predictive primaries are yet to be held. Although Super Tuesday is a big day in the election cycle, the 11 states in which Democratic primaries are held on that day, and the 12 in which Republican contests are held, don’t appear to carry as much weight as other states when it comes to the data’s crystal ball. None of these states have accurately predicted the eventual party nominees 100 percent of the time. For Republicans there are seven states with a perfect record—they’ll be hosting primaries between March 5 and June 7. For Democrats, the only state with a perfect record is Kansas, which holds its primary on March 5.
Of course it is important to remember that these results are not perfect predictors. With such limited data existing, there is no way to factor out the outliers. The Democratic data contained more outliers than the Republican’s. While most of the Republican nominees won with about 40 states, the range for Democrats was 20-50 state wins (for Mondale and Gore respectively). This difference also explains why there is so much more variety in percentages between states on the Democratic side than on those of the Republican side.
Why This Type of Analysis Matters
Whether it’s analyzing primary data to better inform the public during an election year or analyzing intelligence and cybersecurity risks in the Department of Defense, public sector analysts produce usable information that allows meaningful public participation and supports decision-making at the community level. The government has access to more data than they can interpret, and they’re looking for interpreters.
The government’s need for individuals with analytical skills has been gaining a lot of attention lately—as evidenced by a recent survey done by Govloop, which found that of 283 public sector professionals surveyed, a whopping 96 percent said their agencies had a data skills gap. This need for data analysts in the public sector is highlighted by the sentiments of Karen Terrell, vice president of SAS Federal:
“There is no question that a data skills gap is threatening America’s future, and in particular, the workforce of the federal government.”
The report further notes the resulting priorities that have been established by the U.S. government:
“As the federal government continues to compete to attract top talent, there is no priority more important than building a skilled workforce capable of meeting the complex challenges of public-sector missions.”
How the Public Sector Data Skills Gap Can Be Filled
The good news is that more and more people are taking advantage of a graduate business degree in analytics. By earning a degree like a Master of Science in Analytics (MSAn), both government employees and those who want to work in the public sector can gain the valuable data analytics skills that are needed to fill this critical gap. A data analytics degree can provide a great benefit to both the federal government and individuals who are seeking to further their careers—and make a positive difference while they do it.
To learn more about a Master of Science in Analytics from American University’s Kogod School of Business, visit onlinebusiness.american.edu/analytics.