"The groups observed that the disparity of spaceflight data available for men and women who have flown in space – 477 men vs. 57 women as of June 2013 – makes it difficult to derive concrete conclusions based on sex and gender alone."
Statistics can't work like that, can it? It's surely not harder to derive concrete conclusions from a large, imbalanced 477-57 group than from a smaller but balanced 57-57 group... right?
At the extreme, you could always take a random subsample of 57 men.
Note: I agree that imbalance is bad, I just disagree that it's bad for taking the conclusions listed in the article and other similar conclusions.
Well, 57 isn't really that great of a number by itself. Especially given their exposure to space is relatively transient (compared to, let's say, Earth-based factors) and that you're looking for relatively rare outcomes (like radiation-based issues), 57 is vastly under-powered.
Where the disparity is a real problem, though, is in the matching. You can't just take a random sampling of men and women, you need the populations to match. Think age, race, marriage status, educational achievement, military experience, etc. With so few women, it's hard to get appropriately-matched comparisons. Hence, from the summary:
>Female NASA space station astronauts are on average 2 years younger than male astronauts. While there were no significant differences in the percentage of male (76%) and female space station astronauts (69%) who were married, a significantly greater percentage of male astronauts had at least one child (67% versus 38%) and overall, men had more children than women. From a professional perspective, female NASA space station astronauts have almost twice as many doctorate-level degrees as their male counterparts (50% versus 28%); conversely male NASA space station astronauts had more military experience (73% versus 39%). Sex and gender differences as well as these social determinants could impact adaptation to spaceflight.
You could select 57 random men, but the main issue is that smaller sample == higher variability.
Ever seen a political poll where one candidate has 51%, one has 49%, but the margin is ±3%?
A similar problem is probably present in many of the studies. Let's say you have 57 women. But not all of them experience motion sickness on the way up, maybe only a third. 20 people is a pretty small number, in statistical terms.
And the more uncommon the malady, the worse the problem gets.
Luckily, the solution is simple: more gender balanced space crews will generate more data.
The radiation difference was one I was particularly interested in - fielded a question on it at Bio.SE http://biology.stackexchange.com/a/10084/4101 - but it's a good example of why this is important to study. Just as we saw with heart attacks, where females exhibit different signs and symptoms than males, this kind of study can drastically change our expectations as we look toward longer-term space flight.
It's good that these differences are being studied before they are relevant. It's better than the time gap between determining men's heart attack symptoms and realizing that women experience different symptoms when they're having a heart attack. A similar such an oversight could have threatened the viability of a longer term space mission.
PBS Space Time made a video[1] that drew from this study to argue that it makes sense to have the first mission to mars be all women because of the issues in the study.
Statistics can't work like that, can it? It's surely not harder to derive concrete conclusions from a large, imbalanced 477-57 group than from a smaller but balanced 57-57 group... right?
At the extreme, you could always take a random subsample of 57 men.
Note: I agree that imbalance is bad, I just disagree that it's bad for taking the conclusions listed in the article and other similar conclusions.