search
top

Dueling Studies

Stanford Law Professor John J. Donohue have revised a study that shows John Lott is wrong, and more guns actually equal more crime. Looks like his premise is that you can’t count the years that were part of the decline in the crack cocaine epidemic. In other words, if you cherry pick your data, Lott is wrong. John Lott didn’t waste any time pointing out the flaws with the study, noting that Donohue used a statistical method that Donohue himself argued was misleading in a different context. Will you see the media reporting on that? I doubt it.

Based on a Facebook thread I saw about this topic earlier, a lot of people don’t like the idea of basing their rights on the outcome of statistical studies, and believe we ought to keep strictly to a rights argument. The problem with that is a rights argument only appeals to a certain part of the population. There are plenty of people out there who, if you asked them, would agree with a prohibition on speech or actions that “hurt someone else’s feelings.” There are voters out there who are swayed by statistical magic, because it makes them feel smart. We can’t let the other side own that field, because if we do, they own those voters.

11 Responses to “Dueling Studies”

  1. Lance Lot Link says:

    There are plenty of people out there who, if you asked them, would agree with a prohibition on speech or actions that “hurt someone else’s feelings.”

    That’s already happening. A lot of anti-bullying laws are based upon someone being annoyed or harassed by unspecified conduct. Normally, specific actions are prohibited, i.e. thou shall not do XYZ. The anti-bullying laws often say thou may do XYZ if no feelings are hurt, but if thine neighbor’s feelings are hurt, then XYZ is a crime.

  2. Lance Lot Link says:

    and believe we ought to keep strictly to a rights argument

    It doesn’t have to be a one size fits all approach. There are a lot of good arguments, and rights is only one of them. The key is to know which argument to use at any given time.

    Remember, we want to persuade, not argue. Arguing is only one of the means, not the end goal.

  3. BenC says:

    In the end the constant is always going to be “you cant fix stupid”.

  4. Jim Jones says:

    If you want to give yourself even more heartburn, look into the “you can’t hurt my feelings” laws that muslims in Canada and Europe have used to silence critics of the more extreme strains of Islam. Then just look into those stupid rules in general, and you will be doubly glad that we have that pesky constitution.

  5. tincankilla` says:

    Dohohue is a hack on this issue, using many of the emotional tactics of antis. here’s him in january at a debate on the topic: http://www.c-span.org/video/?317256-1/stanford-university-hosts-gun-policy-debate

  6. Drifter says:

    I appreciate John Lott’s work, but they other side will always have Mary Rosh.

  7. ProdigalSon says:

    This is not exactly a “scientific” conclusion, but if you can get such radically different results just by varying the model you use, it seems to me that the relationship is probably not that robust. Lott’s method shows one result, Donohue’s another, and while both are well reasoned, each has its own underlying assumptions that seem to cause very contradictory results.

    That’s really the problem with statistics. You, as the researcher, need to determine what’s relevant, and in doing so, you unintentionally skew the outcome of the model. It’s not really a cherry-picking issue, it’s just part of how the system works. By including something irrelevant, or forgetting something important, you are bound to change your results.

    • TS says:

      You can get a model to say pretty much what ever you want. You only care about its accuracy if you want to use it for predictive analytics. If you want to use it to sway public policy, not so much.

      Lott’s method shows one result, Donohue’s another, and while both are well reasoned, each has its own underlying assumptions that seem to cause very contradictory results.

      The key difference is that one of them has raw data on their side, while the other turned the raw data on its head by “controlling” out factors that they chose.

    • SJ says:

      I’ve heard it said that torture works.

      You can torture statistics, and get them to confess to anything.

  8. Joe Huffman says:

    I think a better refutation of this “study” is that it were true you would see a higher percentage of CCW holder committing crime than the general public. But it is just the opposite.

    To believe this study you have to believe that people getting CCWs causes other people to commit more crimes in response.

top