Women "Advertise" Fertility

Dik Zak writes with word of a paper published in the journal Hormones and Behavior. A study found that women take greater care over their appearance when they are at peak levels of monthly fertility. The researchers took two photos of each of 30 women, one near ovulation and one at the other end of her cycle. They then showed the paired photos (with faces obscured) to a group of observers, who were asked to judge in which photo the women were trying to look more attractive. The observers chose the "high fertility" subject nearly 60% more of the time than would be expected by chance.

Comments

No, 60% more

[Chris Burke]

Not 60% of the time, but 60% more than expected if it were chance alone. So more likely 1.6 * (30*.5) = 24/30, not 18/30.

But of course the actual number isn't in the article.

Re:No, 60% more

[Surt]

The article is clear on this, the slashdot summary is wrong. It's 60% right guesses for 42 guessers against 30 pictures, over a thousand total guesses, with 60% right instead of 50% right.

Re:No, 60% more

[Surt]

30 is a low sample size, but would not be unusual in psychology studies. There are statistical tests you can perform to find out the minimum effect size to declare significance. I've seen studies with meaningful results in as few as 8 samples.

Nevertheless, this particular study had 1260 samples. 42 guessers * 30 guesses each. More than a thousand samples is plenty for significance.

Re:No, 60% more

[Chris Burke]

Where the word 'more' came from, other than slashdot summary, I don't know.

It came from the New Scientist article which was linked from the summary at the end of the sentence: "The observers chose the "high fertility" subject nearly 60% more of the time than would be expected by chance, according to the NewScientist.com writeup."

So like I was saying, it isn't the slashdot summary, it's Daily Mall and New Scientist which are contradicting each other.

Thank goodness for the real study, though, which makes it clear that it is New Scientist which is incorrect.

Re:60% of 30?

[Surt]

The article and summary are in disagreement. Choosing to assume the article is more likely to be right, it is 60% right guesses vs expected 50% right guesses.

However, also omitted from the summary is 42 guessers guessing on the 30 dress-up-women in the study. That's 42x30 guesses, with a 60% correct guess rate overall. 60% with more than a thousand sample points is well within the usual scientific standard for statistically significant.

FULL ARTICLE

[brian0918]

The full article can be found here.

About Statistical Significance

[flynt]

I am a statistician, and reading through the comments hear, am saddened that many readers claim that "statistical significance" could not have been achieved in this study because of a sample size of 30 women. First, that's only part of the random sample in this study, the other part is men sampled to judge the pictures.

Second of all, I have looked up the actual publication in "Hormones and Behavior", and the p-value associated with their main test is .01, which usually signifies statistical significance.

Ultimately, determining whether some difference in populations is due to chance depends on more than just sample size. It depends on how large of a difference you want to detect, and the variance of the measurements within a group. Of course, larger sample sizes help, but it ultimately depends on what you're studying, and the design of the experiment.

So while I definitely applaud being sceptical of all statistics, I urge you to look up the actual publications, read the methodology, and then decide if the results are something you believe. Kneejerk reactions to n = 30 don't really help anyone though.

I have not read through this publication in its entirety yet.

Re:Not statistically independent, however

[aethogamous]

For sake of argument, assume that there is an objective way to measure who takes greater care of their appearance and that all 42 judges are experts at measuring that and never wrong. In that case, these results boil down to 18 out of 30 women taking better care of their appearances during one of their ovulation phases than during one of their non-ovulation phases. When you combine that with the possibility that some of the judges could be wrong (thus increasing the expected variance), it's even less significant - not more. I'm going to go with Scooter on this one.

The published analysis is more analogous to saying that each women has a score that measures how much more or less attractively they dress during ovulation. In the article's case the score is defined in terms of the percentage of observers who think that a women is more attractively dressed during ovulation, with scores ranging from 0% to 100%. The 42 observers are used to estimate that score for each women. The reported percentage of 59.5% is the mean of these scores, and is not a percentage of the 30 women.

The analysis asks whether this mean score is greater than 50%. Whether or not significance is achieved with 30 observations in this case depends on the distribution of these 30 scores, which is not given in the article. Using only the information about the mean (59.5%), using a t.test (the actual analysis was more sophisticated, and included covariates) we can easily constuct p-values ranging from 0.1404 (12 women score 0%, 18 score 100%) to 5.969e-12 (15 women score 55%, 15 women score 65%)

It is possible (but presumably unlikely) that more women actually looked worse to a majority of the observers during ovulation and still get a mean score of 65% (for example if 22 women scored 45% and 8 scored 100%).

Moderators: parent post is GARBAGE!

[Tim]

Moderators, for the love of....don't just give high ratings to people who post technical-sounding gibberish!!

The parent post is spreading misinformation with regard to the link between libido and di-hydro-testosterone (you'll note that he got the abbreviation incorrect, and attached a spurious "5a" to the front, because he confused the name of the enzyme -- 5-alpha-reductase -- responsible for DHT formation with the chemical itself!)

I'm not going to claim that DHT isn't involved with male sex drive, but it's certainly not the "primary mechanism" behind male libido. One needs only refer to the volumes of studies done on the relationship between selective serotonin reuptake inhibior antidepressants ("SSRIs", e.g. Celexa, Prozac, Paxil) and libido suppression to see that the issue is more complicated than a single hormone imbalance.

What makes this really galling, however, is that the guy has the guts to criticize real scientific research before spouting this crap, and you folks take it as some sort of authority!