Type Errors and Election Integrity

(This is an analogy that came to me while I was sleep-deprived, so it may not be very good.)

In testing any intervention, you have two goals: make sure that if an intervention (for instance, a medication) has no effect (for instance, on recovery from a disease) then the experiment will show it has no effect. On the other hand, If the intervention does have an effect, then you want the experiment to show that effect.

If you fail at your first goal and mistakenly reject the null hypothesis (conclude there is an effect when there isn’t one) then you have made a type I error, or a false positive. If you fail at your second goal and fail to notice an effect when there is one, you have made a type II error, or a false negative.

These goals are at odds. Statistically speaking, if your test is too strict, if your bar is too high, then you might not notice an effect that is really there (and thereby give up the chance to help people with your intervention). But if your test is too loose, you will start seeing effects everywhere that are just the result of random chance (and thereby waste a lot of time/money/effort on useless interventions).

A lot of areas of policy are similar. There are tradeoffs, and you want to choose the strategy or policy that walks the line between two failure modes that mirror each other. (Or maybe you want to choose the meta-policy or meta-strategy that results in the fewest number of errors overall, just as you design your statistical methods to avoid coming to false conclusions of either type.) A lot of times, there is no perfect answer, and you just have to swallow the least-bad tradeoffs.

Many people are concerned about election integrity, and it comes up frequently in New Hampshire politics. It is relatively easy to vote here. We have same day voter registration. If you don’t have a photo ID, you can sign an affidavit promising that you are who you say you are. This leads some people to worry about false positives. They worry that there are people voting who should not be, and that this threatens our election integrity.

A policy change that would make it more difficult to vote in New Hampshire would decrease our chance of false positives and improve our election integrity in one respect, but it would simultaneously increase our chance of false negatives and harm our election integrity, just in the opposite direction.

There are some bills we’re facing in this legislative session which increase our risk of false negatives – that is, they prevent people who should have the right to vote in New Hampshire from doing so. For example, HB1543, which has not yet been scheduled for a hearing in the Election Law Committee, would restrict most college students from voting.

Why do I think college students should be able to vote here? A majority of their year is spent in New Hampshire, even if their parents live in another state, which means that for most of the year, they have to obey all the laws of New Hampshire, face New Hampshire courts if they commit a crime, and pay New Hampshire’s taxes, such as our meals and rooms tax. They are more affected by New Hampshire’s political decisions than the decisions of any other state. I could list some other reasons, but this is the biggest and simplest reason, for me.

When it comes to deciding between policy tradeoffs, our government has already pre-committed, via the State and Federal Constitution, to a preference for false positives when the question regards retaining a legal right. That is, given the option, we would rather citizens retain their rights when they ought not to instead of losing their rights when they shouldn’t. The most obvious example is the principle of “innocent until proven guilty.”

From my experience, most of the vocal supporters of tightening up New Hampshire’s election laws and reducing false positives at the voting booth would balk at the prospect of tightening up New Hampshire’s gun control laws and reducing false positives at the gun shows. “We want to make sure nobody votes who shouldn’t be able to vote” sounds good to them, but “We want to make sure nobody has a gun who shouldn’t be able to have a gun” sounds threatening. The problem is, both voting and gun ownership are legal rights guaranteed by our Constitutions. The presumption is that everybody has those rights unless the government has proven that they they have forfeited them. The burden of proof is not on the person exercising a right.

What is my ask, here? I guess my main request is for my Republican friends to recognize that “election integrity” can be compromised in two different ways, and that the one way street of tightening ID and domicile restrictions is not a simple win for the goal of election integrity. Preventing somebody from voting who should have the right to vote is a rights violation and if you consider the way you feel when your political opponents talk about raising the bar to purchase a firearm, you might understand why they get so concerned when you talk about raising the bar to vote.