I was wondering how many voters switch parties each year? For example do we know how many voters switch from D to R or R to D in 2015 or 2016? Is there a breakdown by state?
I don't think very many deliberately do so, particularly in the off year. Voter rolls are public records in many states. If you have the data for two years, you could compare. You would also spend a lot of time trying to match records (name changes, addresses changes, etc.)
The only real reason to change affiliation would be able to vote in a primary. But in many cases a switch may have to be done well in advance of the primary. In Oregon and California, there has been a drop in Green Party registration before the 2008 and 2016 presidential primaries. This was in part because the primaries were late in the campaign, and voters could anticipate the primary because of results in other states (in 2008, the California presidential primary was in February, so there was little drop off). Otherwise, most changes are more casual. If someone moves, etc. they might choose a new party. This doesn't really mean a switch, because they probably don't remember what they marked on a form five years ago.
An exception is in states without party registration. In Texas, you can choose your party on primary election day, and the only requirement is that you can't vote in both primaries (this is physically possible because the primaries are (or may be) in separate locations, or switch for the primary runoff. You have to affirmatively acknowledge that you have not voted in another primary, or can not do so in the future.
Someone is not a "Democrat" or "Republican", but rather someone who voted in the Democratic or Republican primary. And that status is only through the remainder of the year. So it is fairly ordinary to vote in whichever primary has the more interesting races.
In Ohio, a vote is technically stuck to the primary they voted in previously. But if challenged, they only have to affirm they support the Democratic or Republican party.
This is the only one that I can recall:
Ohio Statewide Data on Voters who Changed Party Affiliation in 2016At the bottom is a link to a spreadsheet. It is pretty interesting in the number of voters crossing over, but also the large numbers without a previous affiliation.
In Oregon, they've begun automatically registering anyone who updates a license at the DMV and is not currently registered. They mail notice to the voter that they have become registered as unaffiliated, and give them an opportunity to (a) choose a party; (b) become unregistered; or (c) do nothing. The overwhelming share do nothing and the number of unaffiliated voters has burgeoned. About as many person take the step to say that they don't want to be registered as those who choose a party.
In Delaware, they introduced on-line registration (for those who have a driver's license). There is a pull-down menu to select a party, which is indicated as a mandatory question. Unenrolled or unaffiliated is way down the list alphabetically, and there has been a surge of registrations into many small parties (Delaware keeps parties around almost forever). "American" is at the top of the list, and might be selected by those who don't want to be Democrats or Republicans.
A similar process is happening in New Jersey, where people who write-in an other party that is illegible are sent a notice to provide clarification. A write-in choice of "Independent" or "Non-Partisan" is apparently considered illegible, and so some voters choose a party, rather than "Unenrolled" on the second chance.
