In June, the Democratic Party won the primary vote in Georgia by ~9 percentage points, 1.19 million to 0.99 million. Do Senate primary vote totals where both sides are having competitive primaries indicate who/which party will win that seat in November?

I am asking this because I wonder if the fact that the Democratic primary registered an advantage bears significance for November's general.

  • How can a party win a primary vote? Commented Oct 16, 2020 at 15:07

2 Answers 2


No, I don't think it does because of how primaries are held. Depending on the state only members of the party may vote in the primary though Georgia is not one of those states.

You also have to factor in the fact that there was no primary opponents for the GOP for President, Senate, House districts 3, 4, 5, 10, 11, 12 so there was not a lot of reason for some voters to even show up if they had no one in the party to vote for. In contrast the DNC only had uncontested ballots in House districts 2, 3, 6, 8, 11, 14 so they had a lot more reason to show up in greater numbers. NPR election results

With the knowledge some GOP voters had no one from their party to vote for they could have just voted for a DNC candidate to try and influence the results.


Closed Primary

Unlike open primaries, voters in areas with closed primaries have to officially register for a political party before they can vote in the party's primary.

Once the voter has registered with a particular party, the voter can only participate in that party's primary election. For example, Republicans can only vote in the Republican primaries, and Democrats can only vote in the Democratic primaries.

An important feature of the closed primary system is that it forces voters to affiliate with a political party before they can vote in a primary election. Therefore, independent and non-affiliated voters are generally forced to affiliate with either the Republican or Democratic party to cast their vote in a primary election.

Open Primary

When we talk about open primaries, we are essentially talking about primary elections that are open to all voters, regardless of their political party affiliation.

Open primaries give voters the greatest amount of freedom when casting their vote because they can privately vote in either party's primary.

Voters in areas with open primaries can vote for a candidate in their party, or they may choose to vote on the other party's ballot, crossing party lines.


In general, yes, there is a fairly strong correlation between the margin of victory (i.e. the difference between the number of votes cast) in Senate primary elections and the margin of victory in the general. However, if we restrict ourselves to elections where the margin of victory in the primary was less than 20 percentage points, there is far less of a correlation, and even less if we restrict our dataset to primary elections won by less than 10 percentage points. The FEC publishes data on this for every election, available here.

The graph below plots the primary victory margin against the general victory margin in the Senate from 2008 to 2018. If the Democratic party has an advantage in both, the election is plotted in the top-right quadrant, and vice-versa for the Republican party. If a point is plotted in the top-left or bottom-right quadrant, this indicates that the result was split.

As you can see, although there is a clear positive correlation, there is a fairly large number of split results, especially when the result is close. The overall correlation coefficient is around 0.71, however, if we only consider races where the primary results came within 20pp of each other, this drops dramatically to 0.24. Only considering races with a margin of victory less than 10pp, the correlation coefficient drops further to 0.16, indicating that this metric might not be the best for predicting the general winner when margins of victory are small, for reasons Joe W covered in his answer.

enter image description here

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .