Conventional wisdom (of old wives, anyway) holds that the more children of one gender a couple has, the more likely it is that the next child will be of that gender.

I’ve had the darndest time trying to debunk/confirm this notion on the internets, so I thought I’d prevail upon the Teeming Masses for their unimpeachable wisdom. Is it the TM’s view that (assuming no deliberate attempts to affect the outcome) gender mix in families is more or less purely randomly distributed? I’d have thought there’d have been massive epidemiological studies to answer this question, so if anyone can point me to one or more such studies, it’d be much appreciated.

From a purely statstical point of view, assuming a 50% probability for a boy or a girl, a couple with, say, a boy as the first child who have 2 more children is 3 times as likely to end up with more boys than girls - the (equally likely) possible outcomes are BBB, BBG, BGB, BGG - so 3/4 chance of more boys than girls - is that along the lines of answering your question?

It should always be 50/50. Males have an X and a Y chromosome (“XY”) and females two X (“XX”) barring genetic abnormalities we won’t go into, a child will inherit one chromosome from each parent. From the mother (who has an XX) this will always be an X, as she only has X chromosomes to give. From the father there is a 50% chance his sperm (that is to say, specifically the one which fertilises the egg) will have an X or Y chromosome, hence the chance is always going to be 50/50 for either sex.

As for distribution; Remember the gambler’s fallacy, the belief that the universe gives a **** about what’s already happened. Every time a couple conceives a child there is an equal chance of it being XX (and thus female) or XY (and thus male) and no previous conceptions will affect this.

If births are 50-50 and independent, then it makes no difference. However there is some evidence (as I recall) that there are slight genetic tendencies to favor one or the other though it is so slight it was not significant. If this is true, the same sex baby would be slightly more likely to follow.

I have also heard, but don’t have a cite, that couples can have tendencies one way or another. That is, the combinations BBB and GGG are more likely than BBG and GGB. I see no logical reason why this should be ruled out - after all, we know that baby gender is not entirely a random 50/50 chance, because there’s a slight but statistically significant bias in favour of boy births (51:49 IIRC).

Counting among those of my acquaintance for whom this is relevant (ie, more than 3 kids, first two same gender) yields 7 families in the “same gender” bucket and 4 in the “other gender”. I’m not sure how many families you’d have to survey to get a statistically significant sample, but we’ve probably got enough people on the dope to give it a whirl.

Oh, and there also are people who claim that it’s possible to increase your chances of one or the other by methods such as intercourse timing, but this appears to be controversial and unproven

Again, there being a slight skew in the ratio proves nothing. Probability is an average of what is to be expected, not a foregorn conclusion. For example, while you should, in theory, have a coin come up heads or tails, 50 times each after 100 perfectly fair coin tosses, there’s nothing stopping you having 100 heads.

Even if there is some truth to it, you won’t find it statistically in most cases, as you’d need more births than most couples have to reject the null hypothesis that they’re as likely to produce boys as girls.

If you flip a coin and get heads 100 times in a row, your most probable explanation is not “oh well, it’s probability, it happens”, but that you have a rigged or double-headed coin.

The way to find out if the speculation in the OP is true is to do a survey of a random population. Pure logic won’t help us, because it’s quite possible to have a 50:50 sex ratio AND have individuals who are predisposed one way or another - so long as you have equal numbers of girl-predisposed and boy-predisposed individuals in the population

The problem with this is that if you just do a simple survey, you will find families with many boys or many girls purely by chance even if there is no predisposition.

100 boys in a row would suggest a predisposition, but nobody has a hundred kids. If you find a family with five boys, or five girls, does that prove anything? Well, the problem is that it’s not really THAT unlikely - a family with five kids has a 1 in 16 chance of having five of the same gender just by luck. So heavily weighted families are going to be all over the place anyway, whether it’s genes or luck.

The biggest issue you have is that the sex of a baby isn’t really a random event.

I have known lots of families where they’ve only had boys for generations and generations. I had a neighbor they had five boys, all five grew up got married and all of the their kids were boys.

They had to go back generations to find a girl in that family. So that is the first issue, having a baby isn’t really a random event.

You’re sperm should be roughly divided between X and Y but in reality it doesn’t work that way, there are other factors involved, such as Y’s tend to swim faster.

The difficult thing about problems like these is you are really asking two questions without realizing it.

For the sake of argument we’ll assume girl and boy babies will be equal and random.

In that case each birth has a 50/50 chance of boy or girl. BUT and here’s the thing, each birth, in additon to being a unique event is also part of a subset.

So if I said What are the chances of having a girl. If it was my first child it’d be 50/50

Then if I said what are the chances of my next child being a girl. The odds would still be 50/50.

BUT now follow this, suppose I asked “What are the chances of my second child being a boy?” By saying next we are qualifying that as a unique event, but by saying “Second” What we’re really doing is making this a subset of total births which is two.

Also remember statistics don’t tell you what WILL happen they only tell you what is LIKELY to happen. Just because something isn’t likely to happen doesn’t mean it can’t

So you collect a sample of families with three children, and see whether the set of families with three of the same is equal (or close) to 1/4 of the total. With a large enough sample and some elementary knowledge of statistics, you could decide whether to accept or reject the null hypothesis with as much confidence as you’d like. You don’t need a family with 100 boys in a row; you might be able to get by with just 100 (or 1000, or a million) different families.

Thanks for the replies. My query wasn’t as precise as would have been ideal, but I am interested in the question of whether a set of parents’ producing only male or female children is evidence of a particular genetic or other proclivity for doing so or merely a case of having the more-or-less evenly weighted procreative coin come up heads (or tails) 3, 4 or however many times in a row. I did track down a summary of an academic article on the subject (http://www.dartmouth.edu/~chance/chance_news/recent_news/chance_news_10.11.html#item13), but it surprises me that it seems to be regarded as a relatively esoteric topic, instead of one of general interest. I’d think it would be worthwhile grist for mainstream publications on family planning, pregnancy, etc.

Can there be environmental factors at work? There’s a city nearby where I’ve noticed a lot of families with three little girls but very few with three little boys.

It isn’t because as far as we can determine, the effects, if they exist, are very small. Maybe some men have healthier boy sperm, maybe some women possess an environment more hostile to girl sperm, but the consequences drown in the 12,5% of families with three children who statistically should have three boys.

The only thing we can prove from statistics is that just over 51% of all babies are boys.

My findings, based on a massive sample of 108, is that there is a tendency for the second child to have the same gender as the first. Perhaps you can add the extra records to your poll…