Same sex parents and adopted children

by on July 7, 2014 at 12:32 pm in Data Source, Education, Law | Permalink

The largest-ever study of same-sex parents found their children turn out healthier and happier than the general population.

A new study of 315 same-sex parents and 500 children in Australia found that, after correcting for socioeconomic factors, their children fared well on several measures, including asthma, dental care, behavioral issues, learning, sleep, and speech.

Do note this:

Perceived stigmas were associated with worse scores for physical activity, mental health, family cohesion, and emotional outcomes. The stigmas, however, were not prevalent enough to negatively tilt the children’s outcomes in a comparison to outcomes across the general population.

There is more here, from German Lopez, the study itself is here.

1 Brittany Allen July 7, 2014 at 12:48 pm

What kind of economist trusts self-reported surveys?

2 XVO July 7, 2014 at 6:53 pm

Hahahahahhaha……. That’s like asking why are social studies not science?

3 dead serious July 8, 2014 at 3:33 am

“The largest-ever study of same-sex parents found their children turn out healthier and happier than the general population.”

If instead this read:

“The largest-ever study of corporate boards of directors found their CEOs generate more value than non-CEOs.”

You’d have this same group of commenters twisting themselves into pretzels defending the “value” a CEO brings in justification of the outlandish comp, while unable to prove one iota of causality.

Such is the mind of a plutocratic bootlicker I suppose.

4 dead serious July 8, 2014 at 3:38 am

Point being, no rabid screams of “Where are the controls?!” and a dearth of “Self-selection!” accusations when assertions of “value” or “causation” are made.

5 Brian Donohue July 8, 2014 at 10:43 am

It’s true. ‘plutocratic bootlicker.’ Good one. What colorful description do you use for lefties who engage in precisely the same dance of mood affiliation? How about ‘boot-licking apparatchik’?

6 dead serious July 8, 2014 at 1:43 pm

Happy you agree. I’m fine with that.

7 TheObeseDog July 7, 2014 at 12:52 pm

Same-sex parents only have kids if they want them and go to great lengths to get them whereas a certain number kids raised by straight parents were unplanned. How did they control for that?

8 Richard July 7, 2014 at 12:55 pm

Comparing adopted kids to the general population is obviously invalid. We already know that adopted children do better generally. The proper study would be adopted kids in traditional families to adopted kids with same-sex parents.

9 Z July 7, 2014 at 1:09 pm

Not only that, but you have to account for a natural in the mix. Some non-trivial number of “gay parents” are one natural parent and some third party.

What this study shows, maybe, is that homosexuals are better than heterosexuals at selecting babies for adoption. That and it gives the usual subject a reason to hoist the rainbow flag and march around in public blowing their horns.

10 Blanchard July 7, 2014 at 1:27 pm

What the study shows is that when a parent is given a survey about their children they are going to respond positively. No interview, no follow up, just drop a few hundred envelopes in the mail and hope they respond. I’m pretty certain that if you repeated this “research” with nearly any cohort you would achieve a similarly positive result.

11 Z July 7, 2014 at 1:43 pm

@Blanchard: I’m not willing to suspend disbelief. This is not a study. It is agit-prop. I’m sure some other group opposing homosexual adoption could manufacture results pleasing to their cause. The exercise of sending out questionnaires was just so they could pretend to be doing real research, when the results were known in advance. I say this as someone indifferent to homosexual adoption.

12 Boonton July 7, 2014 at 2:15 pm

OK but then why would different sex parents respond less positively to such a survey about their children?

13 Steve Sailer July 7, 2014 at 4:30 pm

“after correcting for socioeconomic factors”


If you did a study of heterosexual single mothers, would it make sense to “correct” for the factors that on average they tend to be poorer and live in crummier neighborhoods and can afford worse schools?

14 Boonton July 7, 2014 at 5:08 pm


For the same reason clinical studies are normalized for age. For any given condiction, like heart disease or cancer, it’s almost always better to be younger than older. But you can’t choose your age. So it’s not helpful to know something like your chance of heart attack goes up after 50. What’s helpful to know is things like given that your risk goes up when you get older, will eating more fish result in less risk or not? Exercising 100 minutes a week? 200? And so on.

Presumably most people would choose a higher income and higher socioecnomic class. (Go ask people, “would you rather make more than $100K or less than $100K?” and you’ll get a much higher % opting for the first than who actually make $100K+), the question being asked is if we take socioeconomic class as a given, what increases or decreases good outcomes?

But that does raise a valid question. If someone shows same-sex parents have a 1% higher risk of bad outcomes, does that make it unethical for SSM coulples to have children? How about middle class or lower middle class couples who may have a 25% higher risk that gets ‘normed out’ in these studies?

15 Cliff July 7, 2014 at 5:10 pm

It would if you were trying to isolate their parenting skills from the general crappiness of having no money.

16 Tummler July 7, 2014 at 6:20 pm

@ Cliff
Why on earth wouldn’t ability to acquire material wealth be considered a parenting skill?

17 Scott Cunningham July 7, 2014 at 7:33 pm

““after correcting for socioeconomic factors”
If you did a study of heterosexual single mothers, would it make sense to “correct” for the factors that on average they tend to be poorer and live in crummier neighborhoods and can afford worse schools?”


If your question is what’s the causal effect of same sex parentage on child outcomes, then yes.

If you question is rather “how are same sex parents different than alternative household structures”, then no.

18 Boonton July 7, 2014 at 3:47 pm

“What this study shows, maybe, is that homosexuals are better than heterosexuals at selecting babies for adoption.”

Does this sound like even a quarterway plausible hypothesis? You’re saying that maybe gay parents have ‘babydar’ that can spot ‘good’ versus ‘bad’ babies when viewing pictures at the adoption agency? Maybe heterosexuals should hire gay ‘adoption consultants’ to help them pick out the ‘best babies’.

19 Steve Sailer July 7, 2014 at 4:32 pm

So, this says that Jodie Foster-style eugenics works? Richer lesbians put a lot of thought into finding sperm donors with the hereditary traits they want in a child.

20 Alise July 7, 2014 at 1:09 pm

Right on, Richard.

21 Nikki July 7, 2014 at 1:11 pm

It seems like a reasonable assumption that in most of those families, one of the parents is biological.

22 Cliff July 7, 2014 at 1:28 pm

This should result in “better” children in general since you can easily pick a very successful third party to supply the other gamete

23 Boonton July 7, 2014 at 2:59 pm

I think same-sex parents have an advantage when one parent is biological in that the non-biological parent has less interest in creating their own children. In many blended families trouble comes when parents start favoring their own biological kids over their stepkids. Even when there are no kids, both parents probably anticipate that potential problem on some level.

It would be interesting if we could compare pure adoptive families (no biological connection) to blended straight and gay families.

Fact is, though, if gay or straight parents made a huge difference in itself it would be easy to pick it up in the data. The fact that it isn’t IMO is very bad for the ‘think of the children’ argument from anti-SSM advocates.

24 Locke July 7, 2014 at 1:11 pm

I think the point of the study is to determine not so much whether same-sex parents are superior at parenting, but rather if the kids are doing alright relative to kids growing up in both the best and worst conditions.

Decent human beings should be happy for these families.

25 T. Shaw July 7, 2014 at 1:28 pm

Ergo, I am not a decent human being.

26 Boonton July 7, 2014 at 3:54 pm

No you weren’t a decent human being before the study came out either

27 GiT July 8, 2014 at 12:27 am

That’s been clear for awhile now.

28 Scott Cunningham July 7, 2014 at 4:37 pm

That’s a good point. And a good research design too. The authors should try to address that selection bias by matching with adopted children in straight families of comparable parental income and education, and so forth. As it is, the paper is just suffering from numerous selection problems including that it is a convenience sample.

29 A Definite Beta Guy July 7, 2014 at 1:29 pm

80% of children had a female index parent while 18% had a male index parent

What does this mean, exactly?

30 Cyrus July 7, 2014 at 2:12 pm

If in the long run, data were to show that lesbian couples made better parents than hetero couples, and gay couples worse parents than hetero couples, does the traditional marriage lobby shift to defining marriage as between a woman and one other person?

31 Chris S July 7, 2014 at 5:16 pm

Agreed. Sample bias.

32 Boonton July 8, 2014 at 9:04 am

Should that be controlled for? Sounds like a feture rather than a bug? Fact is different sex families have a high danger of having kids ‘for the wrong reason’ and since it’s very easy for heterosexual couples to create kids either by accident or on purpose, it seems very easy for heterosexuals who either aren’t serious about parenting or aren’t prepared to be good parents to end up with kids. Since SS-couples have to ‘work harder’ to get kids, that would seem to be a natural filter to block at least some types of bad parenting.

33 joshua July 7, 2014 at 12:59 pm

Convenience sample.

34 ibaien July 7, 2014 at 1:02 pm

this thread is going to go south, fast.

35 Jan July 7, 2014 at 1:22 pm

Interesting. I’ve never seen a study indicating children adopted into same sex couple families do worse than average kids or worse than ones adopted by heterosexual couples. Until we see some evidence to refute this finding, I think the anti-gay adoption lobby will have to update their priors. Or possibly deny the findings and nitpick the study.

36 The Other Jim July 7, 2014 at 1:22 pm

Having gay parents reduces asthma, huh?

Is there anything that homosexuality can’t do?

37 T. Shaw July 7, 2014 at 1:35 pm

Halt climate change . . .

If you don’t have a sense of humor, stop right here.

With apologies to Elsa, the heterosexual lioness.

Born Gay

Born gay, as gay as the wind blows
As gay as the grass grows
Born gay to follow your heart

Live gay and beauty surrounds you
The world still astounds you
Each time you look at a star

Stay gay where no walls divide you
You’re gay as the roaring tide
So there’s no need to hide

Born gay and life is worth living
But only worth living
‘Cause you’re born gay!

38 cheesetrader July 7, 2014 at 2:17 pm

Aaaauuuuuuuuuuuggggghhhhh earworm hell! Where’s the trigger warning? Make it stooooooooop

39 mulp July 7, 2014 at 2:00 pm

Well, the claimed benefits for heterosexual marriage are wildly overstated and never proved.

I remember the days of shotgun weddings and nearly impossible divorce and in those days, marriage meant poverty just as often as being unmarried. Now that those in poverty can avoid the bad marriage or get out of it, marriage has become the means to get rich, according to conservatives.

40 Boonton July 7, 2014 at 2:18 pm

Is there anything that homosexuality can’t do?

Improve comment quality on most blogs.

41 CMOT July 7, 2014 at 1:25 pm

Is it possible that a study that didn’t super-duper affirm anything gay would be published by mainstrem academics? Given its rigged design, doesn’t this study tell us more about culture of academic researchers than it does about gay parents?

I think there’s a weak anthropic version of “gay [anything] studies”, namely, if academic researchers were willing to study the subject honestly they wouldn’t be academic researchers.

42 Boonton July 7, 2014 at 3:01 pm

So you’re saying your inability to come up with data to support your argument just indicates how great the conspiracy is to suppress such data.

43 CMOT July 7, 2014 at 3:16 pm

I make no argument about gay parenting, only about academic culture.

But if you think what the study produced was *data* well … “You keep using that word, I do not think it means what you think it means.”

44 Boonton July 7, 2014 at 3:35 pm

Your statement seems to imply lots of studies were done but if the results came back ‘bad’ the researchers would just opt to not publish the results….despite the fact that consuming lots of time and money and not publishing any results is well known to be harmful for one’s academic career.

Now you seem to be saying the study did not collect data in a proper manner, you should be more specific here and explain your charges.

45 GiT July 8, 2014 at 12:30 am

Regnerus was published despite his rigged design. There are plenty of social conservatives who do sociology of the family. It’s a nice place to smuggle in your traditionalist priors.

46 Chris July 7, 2014 at 1:29 pm

This is sure to silence all those conservatives who opposed gay marriage for fear it causes asthma.

47 Marie July 7, 2014 at 11:02 pm

That was brilliant.

48 Jack PQ July 7, 2014 at 1:32 pm

Endogeneity bla-bla-bla… The study does *not* establish cause and effect. The findings merely confirm that gay parents are not a random sample of all parents, or for that matter, of all gay people.

Prof. Cowen, why do you link to bad research? I do not mean the results are false. I mean there is no cause and effect. And cause and effect is what the authors and especially the media want to claim.

49 CMOT July 7, 2014 at 1:37 pm

I think this is just another turn in the seemingly endless logrolling for Vox. The actuals study is just a means to that end. And don’t think Vox hasn’t noticed and isn’t logrolling back! Tyler’s favorable links from Vox have been growing …

50 Chris S July 7, 2014 at 5:21 pm

I noticed that just today.

At least he isn’t swiping The Browser’s content as much anymore, though he did have a juicy link from there to here.

51 Beliavsky July 7, 2014 at 2:01 pm

According to a summary by the Family Research Council of a study by Mark Regnerus, there are numerous negative outcomes for the children of homosexual parents, compared to those of intact biological families.

Compared with children raised by their married biological parents (“intact biological family,” or IBF), children of homosexual parents (LM and GF):

Are much more likely to have received welfare growing up (IBF 17%; LM 69%; GF 57%)
Have lower educational attainment (IBF 3.19; LM 2.39; GF 2.64)
Report less safety and security in their family of origin (IBF 4.13; LM 3.12; GF 3.25)
Report more ongoing “negative impact” from their family of origin (IBF 2.30; LM 3.13; GF 2.90)
Are more likely to suffer from depression (IBF 1.83; LM 2.20; 2.18)
Have been arrested more often (IBF 1.18; LM 1.68; GF 1.75)
If they are female, have had more sexual partners-both male (IBF 2.79; LM 4.02; GF 5.92) and female (IBF 0.22; LM 1.04; GF 1.47)

52 Nikki July 7, 2014 at 2:20 pm

Having had more sexual partners is now a negative outcome?

53 Jarosław Hirny July 7, 2014 at 2:54 pm

Before citing Regnerus study, please read this:

“Among the problems they cited: The study classifies as “lesbian mothers” and “gay fathers” any people who have had same-sex relations since becoming parents. More than half of the subjects who Regnerus holds up as victims of same-sex parenting are, in fact, the products of heterosexual marriages that fell apart—they are part of his dataset because a parent later went on to have a same-sex partner, casual or otherwise. Regnerus judges the effect of “same-sex relationships” by looking at subjects who, for the most part, were not raised by a same-sex couple.

Many experts concluded that Regnerus had merely documented the well-established effects of broken families on kids—and nothing unique to same-sex parenting. That was the essential conclusion of the American Sociological Association, which has rejected the study’s findings and said publicly: “If any conclusion can be reached from Regnerus’s study, it is that family stability is predictive of child well-being.”

54 Justin July 7, 2014 at 5:57 pm

Thanks for linking to a reasonable objection to the Regnerus study. Regarding the current study, I wonder if the referenced institution will report on the obvious data issues that many have already commented on. I suspect they won’t, which is I think goes to the point that some are making regarding academic bias. Personally I think these studies are all noise and no signal.

55 P July 8, 2014 at 9:21 am


56 GiT July 8, 2014 at 12:32 am

Get with the times. The Regnerus study was marred by some pretty obvious flaws with its “inact biological family” categorization.

57 lesbian wife & aspiring parent July 7, 2014 at 2:19 pm

The only thing you could expect to discover from a study like this is whether the apocalyptic predictions of people against same sex parenting are true.

They aren’t. That’s worth noting.

There are enough people in my own family who are endlessly concern trolling about my future child, worried that without a male parent, s/he will be doomed to psychological malformation. They may cite the Family Research Council’s study above, but more likely, they’ll just murmur something about male influence and father figures and imply that our baby deserves better than two moms.

To be able to quantify actual possible harms is a tremendous asset to us. This doesn’t rule out all the doomsday scenarios we hear, but it excludes enough of them to make people think twice (or once, if they hadn’t done it the first time.)

58 Jack PQ July 7, 2014 at 2:51 pm

You are right, it is good to fight prejudice with new facts. What they show is that there is no doomsday scenario. I’m not convinced about cause & effect as I wrote above, but we can at least agree the kids are unharmed. And that’s important for people to know.

59 Z July 7, 2014 at 2:52 pm

If the study showed that children raised by homosexuals were worse off than the alternatives, would you reconsider having children or adopting children?

60 Jan July 7, 2014 at 3:40 pm

Regardless, we know you’re not going to change your tune. I think I have literally never seen you convinced that your prior beliefs could be wrong, based on information anyone shared with you on this blog.

The Regnerus study has been eviscerated as bunk, but if you had any good research pointing the other way, I’d be interested to see it.

61 Z July 7, 2014 at 3:59 pm

You projecting your priors onto me may make you feel better, but it makes you look ridiculous.

62 Jan July 7, 2014 at 4:02 pm

Waiting for the data, my man. Bring it on.

63 Z July 7, 2014 at 4:10 pm

@Jan: Data for what? I asked a question.

64 Jan July 7, 2014 at 4:19 pm

My points is: if there is good data pointing the other way, we want to see it. Are hypotheticals all you have? If so, you’re dismissed.

65 Scott Cunningham July 7, 2014 at 4:49 pm

This isn’t the study to take to battle though. Its a convenience sample first of all. And the selection problems related to treatment selection aren’t even articulated let alone addressed in any serious way. I appreciate your frustration, but I think this study is not the one to say positive things about except that the authors are at least attempting to answer the question. That’s admirable, but it seems like the goal should always be better data, but also, better design. Its not merely problems with the data iow — homosexual parentage is non-random and if anyone wants to know the effect it has on child outcomes they will require either an explicitly randomized treatment assignment (unlikely) or they will require modeling the treatment selection explicitly so as to block any backdoor paths between the same sex variable and child outcomes.

66 lesbian wife & aspiring parent July 8, 2014 at 10:30 am

Agreed, this falls far short of being a valid study, but I hope that it lays the groundwork for a more rigorous study in the near future.

As I mentioned below, this kind of thing is not at all useful for convincing LGBT parents or the religious to change their minds. However, it can be useful for convincing the convincible and short-circuiting future moral panics.

67 Boonton July 7, 2014 at 3:42 pm

Interestingly the results stated “correcting for socioeconomic factors” children of same sex couples did well. So what does that mean? That means if you have lower income or are lower class you chances of having kids who have problems are much higher. The study just demonstrates that being gay doesn’t appear to increase the risk of those problems anymore.

This is much like many clinical studies which immediately adjust for age. Why? Because for almost all health problems, being older means you are more likely to have the problem. So studies adjust for age to see if factors like smoking, diet, exercise, supplements etc. raise or lower the risks.

So does this mean that people who are lower ‘socioeconomic class’ should feel guilty about having kids? How about the middle class? If class is such a huge factor in bad outcomes do you have a duty to remain childless if your income can’t break $100K?

68 Z July 7, 2014 at 4:00 pm

Instead of posing, try answering the question, even though it was not aimed at you.

69 Boonton July 7, 2014 at 4:04 pm

I’d say the answer is no. Averages are helpful but they ultimately are not that helpful for one’s individual circumstances.

70 Z July 7, 2014 at 4:21 pm

@Boonton: Exactly. The very tiny number of homosexuals looking to adopt are never going to be persuaded otherwise. Similarly, those opposing homosexual adoption will never be convinced it is just as good. That’s because it really has nothing to do with what’s bets for children. This is a proxy fight over larger cultural issues.

71 Luciom July 8, 2014 at 6:56 am

Unfortunately yes, that’s the conclusion. Based on a ton of evidence, poor people ethically should have less children than rich people.

This is pretty well understood in europe where young couples postpone becoming parents until they reach a decent income (and if they don’t, they usually avoid making babies). At least, young couples that are not immigrants, that is.

72 Boonton July 8, 2014 at 11:53 am

Kind of assuming non-existence is better than below average or even average existence.

73 lesbian wife & aspiring parent July 8, 2014 at 10:16 am

Which alternatives? Are you asking if I would abort my child, who is currently at 20 weeks? If so, the answer is no. Putting him/her up for adoption is a possibility, I suppose, but you would need rather tremendous evidence to get me to consider that, as you would for any straight parent.

The fact is that no amount of studies are going to convince me or anyone else with a dog in this race to change our minds. The purpose of a study like this is to make some small step toward convincing the average persuadable person in the middle that gay parenting is alright–and accordingly, to convince legislators and their panic machines that it’s more beneficial for them to move on to the next social panic.

74 The Other Jim July 7, 2014 at 3:27 pm

>the apocalyptic predictions of people against same sex parenting

Citation needed.

75 Chris S July 7, 2014 at 5:28 pm

I agree with lesbian wife (and her possibly attractive wife) that the absence of evidence here is indeed a positive for her claim of being able to reasonably raise a child.

Often not addressed in this debate: Gay parents harm kids…compared to what? Even if (stipulating) the absolute best environment to raise a child is a male and female married hetero couple of upper middle class means, college educated, open minded but not gullible, surrounded by lifelong married heterosexual grandparents with a compliments of aunts, uncles, cousins in a solid school district, clean drinking water and 1.3 brothers/sisters, does that mean that anything less than that standard should somehow be outlawed?

So much for the struggling single minority parents… take away their kids! (And give them to whom??)

I’ll let you guys fill in the remainder of the parade of horribles.

76 Luciom July 8, 2014 at 6:59 am

It would be easier to say that it’s 100% sure that a baby can have a better life with an homosexual couple anywhere in the rich world compared to what his life would be if he grew up in an orphanage in a third world country. That’s 100% clear to anybody with a IQ higher than 60.

77 lesbian wife & aspiring parent July 8, 2014 at 10:25 am

Even supposing I were straight, I would still be 1) a sub-optimal mother, thanks to genetics, psychological factors, etc, and 2) not significantly worse or better off than any parents I have known. Optimal parenting is not an option for the vast majority of humanity.

The decision to have a kid has been an expression of hope for me. I expect it is for most parents.

And yep, my wife is a cutie.

78 y81 July 7, 2014 at 10:34 pm

I guess it depends on what you mean by “doomsday scenario.” I could gather a sample of reasonably healthy 80-year-old cigarette smokers, and that would definitely prove that cigarettes aren’t poison like cyanide is poison. Still, plenty of relatives, at least in most families, would start “concern trolling” if they heard that you had started smoking.

79 Boonton July 8, 2014 at 9:00 am

Unlikely, if you gathered some 80 yr old cigarette smokers and some non-smokers you’d likely find the non-smokers were healthier than the smokers…even at 80 years old.

Selection bias would kick in if you did something really biased like sample your 80 yr old smokers from a cruiseship specializing in retirees and sampled your non-smokers from a hospice center.

80 y81 July 9, 2014 at 1:31 pm

So all of a sudden you have become a devotee of random samples? I found it convenient to gather the 80 year olds on a cruiseship and the others at a hospital. What’s wrong with that?

81 Marie July 7, 2014 at 11:16 pm


As one of the folks in the other camp, I can tell you this study doesn’t budge me a millimeter, and that’s because they don’t define good the same way I do. So if they believe non-apocalyptic means no increase in asthma, and you do also, then that’s a non-apocalypse for you, but not for me.

If they do a study that is well conducted using what I use to determine for myself whether I’m raising my kids well, and still get good results, then I might budge.

I’ve met a ton of sociopaths who learn well. I know some real horrible mean girls with great teeth. I know a bunch of kids with speech problems that are beautiful human beings that any parent would be thrilled to have raised, and a whole ton of kids with “behavioral problems” that actually just have actual minds that they use independently.

As a potential mom, you have to decide for yourself what you think makes for good parenting, and decide whether you think you can do *that* in the context of your life. Just like I had to. That’s the only criteria, not some study. We may hold differing opinions on the generality, but personally I don’t look to some study to tell me whether I’m a good mom (not even in on the margins) and in my not-asked-for opinion, neither should other women, even the ones who also are lesbians.

82 Art Deco July 7, 2014 at 2:29 pm

From the Abstract

A cross-sectional survey, the Australian Study of Child Health in Same-Sex Families, was distributed in 2012 to a convenience sample

As in Ann Landers’ mailbag. It was a cross-sectional study, not a longitudinal or panel study. Why would this interest an academic economist at a research university?

83 Boonton July 7, 2014 at 3:04 pm

Why are cross-sectional studies of no interest to academics? Who are interested in cross-sectional studies? Biker gangs?

I’m not clear why a ‘convenience sample’ would make a study invalid. Say I mailed out a bunch of surveys asking people their income and what state they live in. I from the responses I notice NYC veers towards higher income, Alabama towards lower.

Sure you could say that since it was a ‘mailbag’ of responses maybe Alabama has a lot of shy rich people who don’t like responding to surveys so in reality the two states have the same income. But that’s a difficult arguement to really make.

84 y81 July 7, 2014 at 3:20 pm

Who is interested in statistically invalid studies? Politicians, activists, reporters who either have an agenda to advance and/or don’t understand statistical methodology. There’s actually a pretty big market for such studies.

As for your hypothetical study, you seem to be defending it on the grounds that it shows what we already knew or thought we knew. Statistically invalid studies which tell us what we already believed–without increasing our true knowledge–are absolutely worthless.

85 Boonton July 7, 2014 at 3:32 pm

You forgot to mention why it is statistically invalid. Yes mailing out a bunch of surveys and reling on people to respond to them *may* introduce biases in the results, but that in itself doesn’t make the study invalid. You’ll have to do better.

86 Blanchard July 7, 2014 at 5:06 pm

A convenience sample is taken from a group that is either self selected or easily accessible.

If a researcher were to conduct interviews on a college campus, which is the most convenient sample, they would get very different result compared to conducting the same interviews at a mall, a park, or a church. Further more, any convenient sample likely has several additional similarities which may lead to a bias in results; in a college campus most interviewees will be under 25, unemployed and politically left leaning. Due to the underlying bias, a convenience sample can not be representative of the general population.

87 y81 July 7, 2014 at 10:27 pm

I thought that in a generally sophisticated audience like this one, everyone understood that convenience samples such as the one you propose provide no reliable information about the population sampled–only about the people who happen to return the questionnaires. So your hypothetical results provide no reliable information about relative incomes in NYC and Atlanta. If you truly don’t understand that, then this blog is much too sophisticated for you, and you probably don’t understand even the parts of it you think you understand. You should take an undergraduate statistics course. The issues aren’t that complicated, but they cannot be rigorously examined in a blog comment.

88 Boonton July 8, 2014 at 9:12 am

In my hypothetical, Alabama results in lower income than NY. Some possibilities:

1. The samples are valid, Alabma indeed has less average income per person than NY.

2. They aren’t. Alabama either equals NY or is greater than NY. But for unknown reasons rich people in NY responded while rich people in Alabama were shy about responding.

The sampling itself is not the problem. It’s not like I sent my survey to Trump Tower residents in NY and a trailor park in Alabama. There is indeed some other factor that causes Alabama to be different than NY causing the sampling to be a problem.

So I agree this type of survey doesn’t prove NY>Alabama. But it isn’t useless. If I’m seeking to open a new luxury car lot and the Alabama Chamber of Commerce has been telling me Alabama=NY in income, then such a survey would cause me to lower the probability that is a true statement.

89 Scott Cunningham July 7, 2014 at 4:53 pm

Because of selection bias. Read about the contemporary criticisms of the Kinsey report which also used convenience sampling. The purpose of random sampling is to ensure that the sample analogs for the population parameters is valid.

For small populations, like lgbt parents, if you can’t get a large enough sample through representative randomized sampling then you need to look into the modern methods people now use for studying smaller populations. There are Bayesian methods like capture-recapture as well as ones that use network theory such as snowball sampling and respondent driven sampling.

I speak from experience. I have a convenience sample of 700 sex workers, and only afterwards did I learn all the stuff i should’ve learned first. Convenience samples at minimum need weights to make them representative but I’m not sure they actually did that even.

90 Boonton July 7, 2014 at 5:00 pm

So you’re saying if they didn’t get enough lgbt parents in the responses that may not be a good sample of what the ‘typical’ lgbt household looks like? I think that’s a valid criticism but the criticism I saw here was centered around responses..which begs the question if parents are more likely to respond if results are positive (or they are more likely to shade responses to make themselves look positive) why would this source of bias be different for different types of parents?

91 Scott Cunningham July 7, 2014 at 7:59 pm

It’s not that convenience sampling doesn’t have enough lgbt persons. It’s that the sample will contain a distribution of characteristics both observed and unobserved, and if the sample is generated through random sampling, then that sample distribution will converge (by central limit theorems) to the unknown population distribution.

Let’s take the worst case we can imagine. Say that someone sampled interracial marriages but used only interracial marriages of people in prison. They then compared the child outcomes of those “interracial households” to some other group — say same race marriage households. Well surprise surprise — the interracial group has kids with more behavior problems. And you don’t need to have a PhD in statistics to see what the problem is. It’s not that there are not enough interracial parents in the sample either — it’s that interracial marriage is not homogenous with regards to observed and unobserved factors that determine or are associated with child outcomes. So if all you do is focus on sample size, and not the data generating process, you don’t make matters better — with selection bias you make them worse.

I don’t know enough about their design to know how severe the problem is here. BUT, I also don’t have any priors. So I don’t really know what is and is not normal or weird in the first place. That is why randomized sampling though is so crucial — bc you know how the randomization works, the units selected are selected at random and not for reasons related to the potential outcomes.

92 Boonton July 8, 2014 at 8:56 am

It doesn’t seem that convenience sampling in itself would make a study invalid. The question is does the “convenience population” differ from the actual population in important ways.

As was pointed out, if you use college students as your sampling population then stats like employment and income are not going to look anything like the general population. But what if you’re taking blood samples to see how common a certain gene is? Assuming the gene is not correlated with making it to college, the convenience sample could very well be close enough to a true random sample to provide useful data.

The abstract provided doesn’t go into detail about how the researchers found their convenience sample, only that they had a list of self-identified same sex households and sent them a survey. I’m not clear what they are comparing the results too…did they also have a sample of non-same sex families or were they comparing the results to some other larger studies on non-same sex families?

If they sampled non-same sex families as well, it strikes me that convenience samples could cancel each other out. If their convenience sample of same-sex families was biased towards positive results, why wouldn’t their sample of non-SS families be likewise biased? In that case the failure to find negative results, in fact find positive results for the SS families would be important…unless you can show tht the sample somehow captured exceptionally bad non-SS families and exceptionally good SS families.

93 Art Deco July 7, 2014 at 6:32 pm

1. If the probability of the respondent making it into the sampling frame is not known, you cannot generalize from the results.

2. People do not live in a discrete moment. If you wish to assess effects which emerge over time, a longitudinal study or panel study is the most telling.

3. Dr. Cowen is aware of this and with scarcely a doubt would not submit a study so conducted to a journal editor and would not expect it to pass peer review.

94 Jarosław Hirny July 7, 2014 at 2:49 pm

Hey, don’t mislead!

This is not about *ADOPTED* children. This is about any children of homosexual parents. In table 1 it is clearly stated (roughly, without going into details of parent indexing) that 310 childrens were biological kids of their parents and only 2 (TWO!) — adopted.

95 Das July 7, 2014 at 3:48 pm

Well, it is obvious now and the evil conservatives should not be allowed to speak on this topic anymore. This study is unrefutable scientific proof that only gay people should be allowed to have kids.

96 Chris S July 7, 2014 at 5:30 pm

Just say “Hitler” so we can invoke Godwin’s law and be done with it.

97 Steve Sailer July 7, 2014 at 4:35 pm

Because Science!

98 The Other Jim July 7, 2014 at 8:58 pm

“Science is whatever we want it to be.” — Al Gore

99 join July 7, 2014 at 4:43 pm

these studies are too hard. Even if the result is negative, it may just be caused by the community where the child grows up. Where he might be an outcast just for having gay parents. Then is the bad result caused by the parents or by the community (probably anti-gay republicans)?

Thus everybody will just pick his side and stick with it,

100 Scott Cunningham July 7, 2014 at 4:54 pm

^ Precisely why quasi experimental methods are needed for the question.

101 andrew' July 7, 2014 at 9:29 pm

I think the question has always been how do the children view family? Kind of weird to assume that would be unaffected.

102 Nikki July 7, 2014 at 10:32 pm

You are being uncharacteristically unintelligible. Who assumes that would be unaffected? And what’s the problem with its being affected?

103 Scott Cunningham July 8, 2014 at 12:46 am

It’s got to be an economically interesting effect if we are going to make policy. If having lgbt parents makes me like dogs as pets over cats — well, bad example as maybe that would be bad since obviously cats are the superior animal. But whatever, old Elvis vs young Elvis. Point is, if that’s all it does is shift preferences around then who cares. The question should be whether it affects the economically relevant variables.

104 A B July 7, 2014 at 11:29 pm

To me, one interesting thing is the definition of ‘psychologically healthy.’ I think it makes more sense to look at it in the aggregate and not only include psychological state but the resultant society that they create. As an example, we need most children to grow into adults who choose to enter into a long-term monogamous relationship and succeed in raising their own children– lots of them. If you have a bunch of children in a group and they all grow up and don’t have kids, each one individually may measure as ‘healthy’ but as a group something must have gone seriously amiss.

My instinct is that there is probably a small measurable drop-off with homosexual couples just because the modeling can’t be as good, and some kids just might need a strong father or strong mother figure.
I guess my point is that the science here is inherently politicized at the definitions.

Comments on this entry are closed.

Previous post:

Next post: