When discussing the efficacy of various anti-mosquito remedies, quite often someone will suggest one which has been tested and shown not to work, and when this evidence is pointed out to them they'll say "Well, it works for me!" To be fair, that a compelling point - you get bitten, you try a remedy, you don't get bitten, therefore the remedy worked. It's very easy to see how powerful that is in any case, and particularly when it's happened to you.
I think it's important to look at this claim. In the first instance, and indeed almost certainly in every instance, the question should be "Did it work?" And there's no harm asking the question.
Scientific research into mosquito repellents has taken place constantly since the 1950s, and pretty much all suggestions have been tried and tested using accurate experiments and techniques. The onus is, therefore, on anyone making a claim to others that "It works for me" to explain why they believe it does, if the scientific evidence shows it doesn't. You have a stack of science on one side, a few people's experience (possibly our own) on the other, I think it's valid to just pause and ask what might be going on.
Okay, here are the standards expected in statistical research. Let's star with a binary claim - let's say a typical claim that taking a particular tablet (garlic, Vit.B) repels mosquitos.
A double-blind controlled test is necessary. This would probably involve three people. The first person wouldn't have any idea, or opinion. We'd take 2 (we'll start with 2) people, give one a garlic tablet, and the other a placebo. The two people involved wouldn't know which was which, nor which was preferred by the person conducting the experiment. The person who gave them the tablets wouldn't know which was which, just 'Tablet A' and Tablet B'. If we really want to get serious, we can make it a triple blind test, where the person who looks at the results doesn't know which is which either, they can just say A or B worked best. Only after everything is completed is the truth revealed.
At each day's end the subject's arms/legs would be studied and photographed, with any bites logged. The photos would be used to ensure any old bite wasn't counted twice.
For scientific standards of evidence we need to apply a little maths to the results. The experiment should be completed on 10 separate occasions. One of the two samples needs to clearly outperform the other at least 9 times out of 10; this could be done over 10 days. .At that point the efficacy of the successful product can be said to have been proven as working on that person (yes, it could then be said it "works for them") with a 95% degree of probability.
That's the maths. That's what is required in a binary to test to say something works with more than a 95% degree of certainty.
Anything less than 10 separate experiments, and there's not enough data for the test to be statistically valid. Anything less than 9/10 in favour of one product over the other is not considered to be proven. We haven't proven to anyone, not even to ourselves, to any accepted standard that it "works for me". To be clear, unless we've done the above, we've not even proven "it works for me" to ourselves.
Now that would be the basis for an experiment if, and only if, we expected everyone using a suitable repellent to never be bitten, and those without an effective repellent to always be bitten, but we know what's not true. It'd also true that some people are effectively co0mpletely unattractive to mosquitos, and never get bitten, whilst yet others are very susceptible, or highly susceptible, to bites. So we'd need to carry that experiment out with ten pairs of people, selected at random, and get the same at least 9 out of 10 times.
If you have no interest in carrying out the above experiments, that's fine. Frankly, I wouldn't be prepared to do so either, so I've absolutely no criticism of anyone who doesn't fancy it themselves. But in that case, the question is whether we should be offering "It works for me" as a counter-argument to properly conducted scientific trials. What we'd doing in that instance is saying "Ignore the masses of evidence from hundreds of correctly controlled tests, and instead listen to my view, which I'm not prepared to test to any accepted standards of accuracy."
Unless we've done the above, we haven't proven to ourselves to the required standard of scientific accuracy that the remedy in question works, even just for us, let alone anyone else. In that instance, we should not be making that claim to other people. And we certainly shouldn't be offering it as counter-evidence to properly conducted research.
And finally, let's say we've done all of the above, and it does indeed work for us. Even at that point, we've proven nothing other than that. There are some areas where we're a little in the dark with on mosquito repellents, and one of the things we don't know about is why some people attract more bites than others. We do know that there's quite possibly a genetic dimension to this, and that in turn may affect how different repellents work on you. However, we do know from other research which repellents work best for most people.
So, returning to our first experiment, if we conduct that, and find garlic or Vit.B works for us, this does not discredit the decades of research on the matter, it just shows we're an exception to the rule, an odd one out, possibly a genetic aberration.
In that instance, we really shouldn't be offering "It works for me". we should be saying "It works for me, but based on all the available evidence I'm the odd one out and it almost certainly won't work for you, and you should ignore my example - I'm the exception that proves the rule."
