Saturday, 24 October 2015

Science and zealots: How to detect bad science and how to detect zealots.

Last night, I got banned from Zoë Harcombe's blog. More on that later. Meanwhile, this...

I re-read It's all about ME, baby! (1997 - present) and there's something important missing.

In 2005, I discovered Lyle McDonald. Before this happened, I had the following beliefs:-
1. If something works for me, it must work for everyone else.
2. If someone with qualifications states a fact, it must be true.
3. If someone without qualifications states a fact, it must be false.
4. If a study confirms my beliefs, it must be true.
5. If a study contradicts my beliefs, it must be false.

Sound familiar?
1. is a "Hasty generalisation" fallacy.
2. is an "Appeal to authority" fallacy.
3. is an "Ad hominem" fallacy.
4. & 5. are a "Cherry-picking" fallacy.

Suffice it to say, Lyle bitch-slapped the fallacies out of me. Thank you so much! Read Lyle's site, if you want to learn.

How can studies conflict with each other so much?

Having read a number of conflicting studies, here are some of the tricks that bad studies use:-

1. Fudge the methodology:-
In a meta-study (a study of studies), to make something that's bad (e.g. some saturated fats/fatty acids) look harmless or to make something that's good (e.g. Vitamin D) look useless, fudge the inclusion criteria so that only studies using low intakes or a narrow range of intakes are used, so that the RRs are either close to 1 or have 95% CI values above & below 1. In addition, include studies that show both positive and negative effects (due to them looking at different types of saturated fats/fatty acids, say), so that the overall result is null.

In a study, use a different type of the thing being studied (but bury this fact somewhere obscure so that it's easily missed) to get the opposite result. E.g. To make "carbs" look bad, use a test "carb" that comprises 50% simple carbs (sugars) and 50% complex carbs (high-GI starches, preferably), thus guaranteeing a bad result.

2. Fudge the statistics:- e.g. Regression toward the mean. I'm not a stats nerd, but there are many ways to lie with statistics.

3. Make the abstract have a different conclusion from the full study (which you hide behind a pay-wall), by excluding the methodology and results.

Back to Zoë Harcombe: I left some comments on Jennifer Elliott vs Dietitians Association of Australia.

My M.O. for detecting zealots is by using a slowly, slowly, catchee monkey approach. I left a comment supportive of low-carb diets, because:-

For people with Insulin Resistance, low-carb diets DO ameliorate obesity, postprandial sleepiness and postprandial hyperglycaemia.

Was that loud enough?

I added that I thought the first priority should be to tackle the causes of the Insulin Resistance, because reversing a condition is better than ameliorating it.

My comments were helpful, with links to blog posts showing the above and how to reverse T2DM in 8 weeks. I then went for the throat, criticising Jennifer Elliot, as the article she wrote contained cherry-picked references. I included three more links to my blog as supportive evidence. This resulted in the removal of all but one of my comments (and the comment that remained had the link removed) and the addition of the following:-

"Zoë Harcombe says:
Nigel – too many comments purely trying to get traffic to your site – link above removed; other comments spammed. You’re now spammed.
Best wishes – Zoe"

The correct word is "banned", Zoë! Low-carb zealot successfully detected.

It's not a problem if a lay person becomes a low-carb zealot, but it is a problem if a Health Professional/Fitness Trainer becomes one. Cognitive bias and a refusal to accept contradictory evidence are not good traits for someone who's supposed to be practising evidence-based health/fitness.

No comments: