Comment by jamii | Hacker News Reader

Comment by jamii | Hacker News Reader

Parent Full thread

Let's look at the first experiment.

http://www.hbs.edu/faculty/Publication%20Files/Enacting%20Ri...

Voluntary enrollment, knowing in advance that they have to do this stupid ritual. ~10% dropped out, and another ~20% didn't fill out their food diaries properly. That's selecting for discipline.

Relied on self-reported food diaries ie opening room for placebo effect and/or guilt-motivated lying.

Only lasted 5 days. Pretty much every dietary intervention works for short periods of time.

Relies on calorie-counting, which is wildly inaccurate.

Finally:

> Despite the effectiveness of rituals, participants tended to think the ritual was not very helpful and reported being unlikely to continue it.

I think it's pretty plausible that rituals can have useful effects on behavior, but I don't think experiments like this tell us anything about it, especially once you add in the effects of publication bias etc.

Comes back to http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.392...

> Null hypothesis testing of correlational predictions from weak substantive theories in soft psychology is subject to the influence of ten obfuscating factors whose effects are usually (1) sizeable, (2) opposed, (3) variable, and (4) unknown. The net epistemic effect of these ten obfuscating influences is that the usual research literature review is well-nigh uninterpretable.

Despite your many reasons to discount the first experiment, their conclusion over six separate experiments is more or less the same as yours: "it's plausible that rituals can have useful effects on behavior" vs "Our body of evidence suggests that rituals can have beneficial consequences for self-control."

What evidence is there that people in this study might have been lying?

I think it's interesting that you completely dismiss calorie counting, which is the only long term strategy that I've seen work consistently for myself & others for weight loss. Why do you say it's wildly inaccurate? How inaccurate is wildly? How accurate does it need to be in order to work?

> What evidence is there that people in this study might have been lying?

People tend to systematically misreport food intake (eg https://academic.oup.com/jn/article/133/3/895S/4688054).

People are really bad at estimating portion sizes (eg https://academic.oup.com/ajcn/article-abstract/35/4/727/4812...).

If you run a non-blinded experiment using a self-reported measure that is well known to be wildly misreported and only get a small effect size to boot, it's on you to prove that you actually affected behavior and not just reporting.

I totally buy both of those claims (and thanks for the sources!), but I don’t see how that necessarily affects this study. The human bias is (unsurprisingly) toward under-reporting consumption in both studies you cite. People aren’t randomly inaccurate, they are consistently underestimating. The self control study showed a decrease in reported consumption, so it doesn’t matter if people under-report food intake or portion sizes, what matters is the numbers went down anyway, right? The bias is present before, during, and after the self control study.

Calorie counting can be wildly inaccurate while still producing the desired effect of weight loss. It probably has more to do with mindful consumption of food than getting the numbers exactly right.

And an unknown number of participants most likely self-reported incorrect information. People don't like admitting they didn't do something that was asked of them.

Self-reported data is shaky data at best and basing scientific results on it makes the science shaky at best.

> Relies on calorie-counting, which is wildly inaccurate.

Some sources on this? (not that I doubt you, it's just that you probably have better references than a quick search might bring up)

Also, if it's inaccurate, what's the alternative?

People are really bad at estimating portion sizes - https://academic.oup.com/ajcn/article-abstract/35/4/727/4812...

Calories in food vary depending on how it is prepared - https://dash.harvard.edu/bitstream/handle/1/5283945/Carmody_...

Calories reported on food labels are often incorrect, because we digest some foods poorly and (because food companies are allowed to use multiple methods and report whichever result suits them best) - http://www2.centralcatholichs.com/copied%20articles%20to%20r...

Individual calorie absorption varies between people depending on gut biome - https://www.researchgate.net/profile/Steven_Heymsfield/publi...

> Also, if it's inaccurate, what's the alternative?

Depends on what you're doing with it. For dieting, counting calories is probably better than not counting calories [citation needed].

For a non-preregistered non-blinded experiment testing an intervention with small effect sizes on a measure where people are known to be bad at self-reporting, well, you're already screwed. There are more direct methods of measurement (eg https://www.ncbi.nlm.nih.gov/books/NBK233774/) but they are costly.

Calorie counting diet helps by forcing you to stop and look at the food before you eat it, so you end up eating less compulsively.

Another way to eat less is to reduce unjustified feeling of hunger, which is often time produced by a sugar crash, specifically a vicious cycle of crash-gorge-crash that’s promoted by sugary foods. If you exclude all sugar and bread from your diet, the appetite will stabilize and you will only ever eat as much as you need, no self control required.

The difficult part of this diet is that pretty much all prepared foods have tons of sugar and bread is also all over the place. The other difficulty is that sugar is addictive, so it takes a few weeks before you stop craving it.

It's inaccurate because people tend to be biased towards under-counting their own calories. However, calorie counting is very effective -- knowing the absolute number of calories you consume is not that important, as long as the reported amount consistently decreases over time.

Reply on news.ycombinator.com