Confirmation Bias Run Amok

By Rod Dreher • June 12, 2012, 10:30 AM

A reader writes:

The response to the Regnerus study on gay parenting is proof that *everything* is evidence for why we need same-sex marriage. If gay couples are high-functioning and healthy, well, obviously they deserve to be married. If they’re unstable and bad for kids, well, obviously they need marriage. What the world needs, my friend, is GAY MARRIAGE.

Just so we’re clear.

Yes. In fact, if we had gay marriage, the eurozone currency crisis would resolve itself, global warming would recede, the lion would lay down with the lamb, and they would open a free Disney theme park on the summit of Big Rock Candy Mountain. If people could just get past their bigotry, they would see that it only stands to reason.

I kid. The point is, the idea of that we can have or ever will have a rational debate about any of this is a sham. I’m writing a piece now about an episode of American imperialism in our distant past, and it’s amazing to me, even at my advanced age, how readily our forebears decided what they wanted to believe, and arranged the intelligence to fit the preferred narrative. I agree with Niall Gooch:

V annoying that “rational” has ceased to be a useful descriptive term & instead become a tribal identifier for a social/cultural viewpoint.

— Niall Gooch (@niall_gooch) June 12, 2012

Posted in Gay Marriage. Tagged confirmation bias, gay marriage.

47 Responses to Confirmation Bias Run Amok

Justin St. Giles Payne says:

June 12, 2012 at 10:38 am

Yeah, this is confirmation bias, Rod – your own. You saw a headline study that “proved” something you already believed was true, and you ran with it despite the fact that the “study” had methodology problems that wouldn’t pass muster in a first-year sociology class.

Before you can complain about the debate not being rational, you actually have to be rational. You weren’t; you saw something that you thought was good for your position and you jumped on it before applying any sort of rational skepticism.
Austin says:

June 12, 2012 at 10:42 am

Re: Niall Gooch’s point:

isn’t this sort of the postmodern critique? That there has NEVER been a time where “rational” has been a useful descriptive point? Isn’t that the basic critique of the enlightenment? That the term “rational” has been a cover for whatever cultural/imperial ambitions people have had?
James says:

June 12, 2012 at 10:47 am

“If they’re unstable and bad for kids, well, obviously they need marriage. What the world needs, my friend, is GAY MARRIAGE.”

That’s not just confirmation bias, it’s outright foolishness. No two people, gay, straight, or Martian, who are unstable and bad for kids should marry. Marriage does not, in and of itself, heal people’s deficiencies, it magnifies them, challenges them, and if the two people aren’t up to the growth required to overcome, it implodes…hence the current divorce rates.

If you’re in my office for pre-marital counseling and I see you are unstable and going to be bad for each other and your potential children, the last thing I do is tell you to move forward onto the marriage, full steam ahead. That’s when I strongly encourage you to do some real soul-searching and reconsider your plans. To otherwise would be the height of irresponsibility.
Jon says:

June 12, 2012 at 11:01 am

That comment is actually beneath you. Will Saletan and Ta Nehisi Coates, among others, are actually making rational points about what is wrong with the design of the survey. Are they motivated by a predisposition to favor gay marriage? Of course. However, neither you nor Ross has actually defended the study on the merits. Rather, you quote the study, bypass the substantive criticisms Saletan makes, and slap a label of “Traditional Families Best for Kids.” Who’s biased again?
Jon says:

June 12, 2012 at 11:04 am

To amend my previous comment, In fact Ross does actually seem to acknowledge and understand the problem with that study.
Mike says:

June 12, 2012 at 11:16 am

I agree that it is getting difficult to have a rational conversation about same-sex marriage, but this isn’t just a problem with the “gay marriage will solve everything” crowd. It’s also a problem with the “gay marriage will result in Kristallnacht for Christians and result in babies roaming the streets” crowd also. Both sides are so entrenched and, arguably, hysterical that it’s difficult to have anything but tribal warfare.
Saint Andeol says:

June 12, 2012 at 11:22 am

The point is, the idea of that we can have or ever will have a rational debate about any of this is a sham.

This kind of sounds like a cop-out. While there are many on both sides of the debate prone to histrionics and mudslinging, the presence of emotion doesn’t automatically disqualify any points from being “rational”.

It’s always the loudest and most controversial that get the most attention, because the first and foremost goal of news outlets is to stay in business. Then when the rest of us try to talk about it, we end up addressing whatever controversy is on the other side.

Just because you believe you’re losing the fight doesn’t mean you get to declare rational debate impossible. There have been plenty of rational, well reasoned counterpoints to your assertions about how society will collapse if we redefine marriage out of existence and whatnot.

You make accusations of irrationality, and yet almost all of your points rest on accepting the Christian viewpoint of homosexuality as disordered and sinful. Our unwillingness to accept this assumption is in no way irrational.
Sam M says:

June 12, 2012 at 11:39 am

Of course, this cuts both ways. Opposnents of SSM do the same thing.

“Gay couples high-functioning and stable? No need for same-sex marriage.”

“Gay couples not high-functioning and stable? Better not have same-sex marriage.”
Tyro says:

June 12, 2012 at 11:43 am

I agree that it is getting difficult to have a rational conversation about same-sex marriage, but this isn’t just a problem with the “gay marriage will solve everything” crowd. It’s also a problem with the “gay marriage will result in Kristallnacht for Christians

This is a false equivalence, because it is the gay marriage opponents who are the ones claiming that gay marriage will undermine civilization itself, while the proponents only claim it will result in a society that is slightly better and more just, in a small way, then what we have now.

Unless you can point me to a person who actually said something close to “gay marriage will solve everything”, I have to suspect you’re just trying to occupy a cautious middle ground for fear of offending.

Ironically, of course, the “gay marriage will destroy Christianity and society” are the same ones claiming we are also on the cusp of adopting muslim sharia law.
Franklin Evans says:

June 12, 2012 at 11:48 am

Broad-brush generalization alert: It’s not that the arguments are not rational — anyone who reads with comprehension but without emotional prejudice will get the rationality on the first read — it’s that the people making the arguments are appealing to people who cannot be bothered with rationality.

In short: We have become a culture of emotion, abandonment of impulse control, and it looks right for me therefore it must be right.

Commiseration statement: Those of us who see the methodological holes in these things have long since lost that battle to the mass marketing “psychologists” whose confirmation bias waxes and wanes with the checks they receive from their clients. Money is truth, and anything that isn’t money is a lie.
JohnE_o says:

June 12, 2012 at 11:53 am

Before you can complain about the debate not being rational, you actually have to be rational.

No you don’t – as Rod just demonstrated…
Bubba says:

June 12, 2012 at 12:45 pm

Yeah, things have “gone tribal”.

Back in the mid 70s, there were lots of moderate Republicans in favor of gay rights. Today, you could count them on two hands and a foot.

Who’s “gone tribal” again?
JohnE_o says:

June 12, 2012 at 12:45 pm

“Yeah, this is confirmation bias, Rod – your own. You saw a headline study that “proved” something you already believed was true, and you ran with it despite the fact that the “study” had methodology problems that wouldn’t pass muster in a first-year sociology class.”

See also – the recent post on Denmark and its State Church.
Jay says:

June 12, 2012 at 12:47 pm

he point is, the idea of that we can have or ever will have a rational debate about any of this is a sham.

Who says we can’t have a rational debate about this, unless you are the one who can’t.

As many others have pointed out, this study says nothing about the effects of gay marriage on children, but rather the effects of divorce. Which, I would agree, is negative. And probably especially so if the divorce came about at least in part because one of the partners was living in the closet.

Of course, allowing gays a more mainstream role in society is, as Saletan pointed out, supposed to help prevent the sort of consequences that follow from feeling compelled to live a lie.

I’m not sure how any of this appeals to emotion and I would honestly welcome your rational response to this.
Fitzgerald says:

June 12, 2012 at 12:58 pm

Really? I mean, really?

You’ve written a lot about failed opposite sex romantic relationships which fail to provide children with stable, loving environments to grow up in, and give them little positive to emulate. You’ve also written about incredible, opposite sex unions that give children a loving, nurturing base of support and morals that will lead them into fulfilling adult life

Both of these things seem to make you think that what the world needs is MORE STRAIGHT MARRIAGE (of the second type).
alcogito says:

June 12, 2012 at 1:47 pm

Every time another of these stories appears, I am reminded of “defining deviency down”, and wonder which practice that was once taboo will next become a “right”. Bigamy, polygamy, anyone?
Mr. Patrick says:

June 12, 2012 at 2:03 pm

The left would be willing to release “rational” back into the public domain if the right could be persuaded to give up “faithful”, “moral”, and “family” in exchange.
Dave Dutcher says:

June 12, 2012 at 2:03 pm

We can’t have a rational debate easily about it because the right to gay marriage is one of those things believed by the majority of people not because of formal argumentation, but because it’s simply one of those parts of the mental furniture that it is assumed an educated person will possess.

Often the same people who are for SSM now when asked about straights living together will cheerfully insist that marriage is only a piece of paper. Or they will believe that SSM will civilize gays, while they are having wild and frequent sex during their college years and acting uncivilized despite marriage having been available to them from the beginning of time.

It’s only a formal system of thought that has been reflected on that can be debated. For people holding an unpopular viewpoint, the resistance from the “mental furniture” people can force reflection, but many people go their entire lives without hearing a challenging word against the common ideas of their circle or class.
Andrea says:

June 12, 2012 at 2:05 pm

Is the argument irrational or do you simply disagree? You write a lot on the topic but I haven’t seen a convincing argument for not legalizing gay marriage in what is, after all, a secular society.
John says:

June 12, 2012 at 2:13 pm

Would love to hear Rod actually comment on the identified major flaw with the study, which is that it is basically comparing divorced, single, and/or adulterous gay parents who had partners drifting in and out of the home with straight parents with intact marriages. I mean, that is a glaring problem of methodology.
CapeCodder says:

June 12, 2012 at 2:16 pm

Rod, it is becoming quite obvious that few of the common sense, well-documented, even ‘rational’ pro-gay, pro-ssm, etc comments from your readers are even being read by you, much less understood.

This study has been debunked by almost everyone, except NOM.

Yet, it comes up with an answer you like.

Is that how you do your own journalism? Have an automatic thesis without doing the actual hypothesis background work?

Do you see the sense of exasperation your readers are beginning to pile up with your rants and ravings.

People used to follow leaders (kings, popes, etc) because they were the only ‘educated’ ones. Then there was the Renaissance, and universities were founded, and colleges, and the peasants began to actually *think* and have *independent* ideas. All of a sudden the infallible was discovered to be fallible.

And now, those who cry for ‘rational’ thought (you) wind up being the ones who are shown to be thoroughly irrational.

You really ought to sit down, read what we write, and really ponder if we are really the rational ones.

Thank You.
J says:

June 12, 2012 at 2:16 pm

As I’ve said before on this blog, the gay marriage argument is at bottom argument about sanity.

The way it is set up is that we have to decide whether gay people are sane and properly to be socially normalized via legalizing their marriages. Or whether heterosexual people with lesser mental disorders like OCD, ADHD, etc who marry and have children are to be kept privileged and normalized by retaining “traditional marriage” in which healthy social functioning is not a criterion so long as children- mentally healthy or not, socially well functioning or not- are produced. (And even when none are, because penile-vaginal contact is an evident social good in of itself.)

Every argument for “traditional marriage” is then at some level an argument about, and subtly in favor of, insanity and social dysfunction. Which must be imputed to homosexual people in any way possible, and distracted from in heterosexual people in any possible as well. The necessity to strain the portrait of American society without proper evidence, without adducing adequate warrants, without sincere or proper arguments to justify it is the sad hallmark of the ‘secular’ conservative argument in my opinion.

The supposed Divine fiat as argument is in the end the most honest conservative stance, imho. But that of course dangles in air as well and relies on a strained interpretation of human nature based in Ancient World conditions and assumptions. As Rod points out via the Celsus critique of Christianity, and the David Frankfurter study of Christianity’s rise suggests, Christianity is socially successful where it is successful in good part because it creates a social haven for the mentally unstable and unlearned and those in need of metaphysical certainties. So in some sense The Church has been something of an insane asylum from the start, diluting the insanity with mysticism- the religious root of hope, which theism per se is not (as demonstrated by Calvinism and various paganisms).

The question the gay marriage argument raises in earnest is whether conservative Christianity has punted on being a preservation of mysticism (though the lip service is copious) and is becoming an insane asylum outright. The increasing politization and glaring lack of production of mystics and real moral leaders isn’t exactly subtle about the long term trend.
Turmarion says:

June 12, 2012 at 2:19 pm

Franklin: Those of us who see the methodological holes in these things have long since lost that battle to the mass marketing “psychologists” whose confirmation bias waxes and wanes with the checks they receive from their clients.

Which just goes to demonstrate what I said a few threads back, i.e. that the ad agency is intrinsically evil….
Fred says:

June 12, 2012 at 2:28 pm

The study is fatally flawed:

http://andrewsullivan.thedailybeast.com/2012/06/are-gay-parents-worse-than-straight-parents.html
Church Lady says:

June 12, 2012 at 2:48 pm

Not sure why the conclusion, “the world needs gay marriage” doesn’t follow from both the evidence and from conservative principles. Isn’t one of the primary arguments of social conservatives that most people should marry, and that it would be good both for very well off people to marry, and for poor people to marry, that in virtually all cases, marriage is good for people? That no matter what one’s skills or place in society, marriage is a positive structure that most should engage. Why would that not be the case for gay people also?

It’s not illogical for gay marriage advocates to make this claim, if marriage really is as valuable a social good as conservatives claim. And isn’t it illogical for social conservatives to extoll the universal good of marriage for straight people, but to heap scorn on the idea that it’s a universal good for gay people? Not to mention grossly unfair.
Liam says:

June 12, 2012 at 3:00 pm

“defining deviency down”

Well, for the most spectacular example of that in US policy in the past decade or so, it would be the approval of the use of torture in the War on Terror.
Kristoofus says:

June 12, 2012 at 3:12 pm

>>The idea of that we can have or ever will have a rational debate about any of this is a sham.

Rod, it doesn’t look like you’re up for it today, at any rate.
Saint Andeol says:

June 12, 2012 at 3:19 pm

Sorry Rod, but as more and more of the internet and blogosphere get into the details of this study, it’s becoming increasingly clear that it doesn’t mean what you claim it means or what the anti-SSM crowd is trying to spin it as.

Confirmation bias has nothing to do with it, it’s just sloppy “science”. And to claim this proves that rational debate is impossible is just ridiculous.
Jamie O'Neill says:

June 12, 2012 at 4:53 pm

… “the lion would lay down with the lamb” …

Actually, it would lie down.
matt says:

June 12, 2012 at 5:47 pm

If you’re interested in the topic, you might read the study, and then read the several devastating critiques of the study, and then make some judgment about the study’s meaning, and then compose a thoughtful post that informs your readers. If you’re not that interested, you could just do meta-level snark.
Colin Vollebergh says:

June 12, 2012 at 6:33 pm

“Every time another of these stories appears, I am reminded of “defining deviency down”, and wonder which practice that was once taboo will next become a “right”. Bigamy, polygamy, anyone?”

Latent Appliance Fetishist.
Mitchell Young says:

June 12, 2012 at 8:35 pm

“Regnerus defended his study to HealthPop, saying he set out to do a population-based study, which is considered the “gold standard” in his field. Other study’s samples, he said, interviewed “convenient samples” of people researchers knew, friends, or groups that are linked together somehow, but he wanted a totally random sampling.” (from ‘HealthPop’

Now, I haven’t read the study. But I do know this.

1) T-N Coates is not a professional social scientist, and neither is Will Saletan. Nearest I can figure, both were English majors (open to correction)

2) Regnerus is a professional social scientist, at a flagship state university. Further, he is an associate professor. That means he still doesn’t have the security of tenure. Academia is liberal, esp. on gay marriage. Even in Texas, Austin. He exposed himself to a lot of retribution. Given these risks, it is doubtful he would be shoddy in his study.

3) The study was published in ‘Social Science Research’. It is published by a leading scientific journal publisher — Elsvier. here is their peer review policy.

Given all that, those accusing Dr. Regnerus of shoddy work, especially well-known writers like Saletan, had better have their ducks in a row.
Jan Hus says:

June 12, 2012 at 8:37 pm

Comments prove the point of Rod’s post. Rationality has little to do with anything. It’s about emotions, sophistry and power. Little else.

One doesn’t need studies to see that SS arrangements lack the essential ingredients that marriages have in their very nature: sexual difference and blood kinship. The most ideal -made for TV- SS arrangement is lacking those. As a result, as a basic level, it is automatically an inferior context for raising children compared to an actual marriage.

The fact that the outcome for children raised among SS partners is inferior, shouldn’t be surprising. Then again, the whole strum and drang over this non-issue has never been about what is best for children has it?
CapeCodder says:

June 12, 2012 at 9:11 pm

Mitchell: an Associate professor is one who has received tenure, and is no longer an assistant professor. Just an FYI.

Interestingly, Rod has yet to make any comments on the deeply flawed aspects of the study, and one needn’t be a sociology or other major to see the convoluted reasoning in this study as pointed out by English majors (who know how to read, and make a convincing argument).
sdb says:

June 12, 2012 at 10:33 pm

CapeCodder,
Associate professor is a rank. One may be an associate with or without tenure. We recently hired a senior faculty member from outside of academia at the rank of Full, and he is on the standard 7year probationary tenure track. Just an FYI.

It is true that most profs follow the assistant-associate(with tenure)-full path, so Regnerus probably has tenure, but he hasn’t been promoted and such promotion (and the accompanying raises) depend on peer evaluation of his work, ability to continue to procure funding, etc… He is certainly taking a risk here. That doesn’t mean he is right. But I think Mitchell’s caution against dismissing this work out of hand because it challenges your preconceived position and has supposedly been debunked by your favorite blogger(s) is well taken.

As I said in the previous post, I don’t think we can say that we have any positive knowledge of psychology (and would be very, very skeptical about sociology as well) based on the recent work by Leslie John et al. and Wicherts et al. If these studies are at all accurate, I don’t see how we can do anything other than start from scratch.
Peter H says:

June 12, 2012 at 10:35 pm

David Blankenhorn makes three quick points about the Regnerus study. His second point: The study provides little if any insight on the topic of family structure.
Mitchell Young says:

June 13, 2012 at 12:47 am

“Back in the mid 70s, there were lots of moderate Republicans in favor of gay rights. ”

Back in the mid-1970s homosexuals were pretty much just asking to be left alone, not to radically alter one of the most basic institutions of our society.

Oh, and there was a little epidemic in between then and now.
TTT says:

June 13, 2012 at 12:56 am

Jan Hus:
Comments prove the point of Rod’s post. Rationality has little to do with anything. It’s about emotions, sophistry… One doesn’t need studies to see that SS arrangements lack the essential ingredients that marriages have in their very nature

Correct. You “don’t need studies” because you have your emotive response to gays that reaches your conclusions for you. Well-said.
Mitchell Young says:

June 13, 2012 at 1:07 am

I wonder how many of those claiming this study has been ‘debunked’ have actually read it, let alone understood it.

I’ve just read through the entire first paper — one of a series which will be springing from this nationally representative sample, I can say that it is solid. Scholars from five universities participated in the design of the questionaire, after a two-day meeting to discuss the project. Nearly 3000 respondents, uh, responses were analyzed.

On the complaint that Regnerus is comparing apples and oranges (intact biological families with gay families where divorce, etc might have taken place), there are two responses.

1) first, believe it or not, the intact biological is the modal form, even today, with 58% of the total. It is logical to compare with this type of family. It serves as sort of a baseline

2) Means/Proportions on a variety of variables are presented for a variety of family types. Make your own comparisons if you wish. Just scanning through, it seems that homosexual families of both varieties usually come off ‘less optimally’ (as Regnerus would put it) than most family types, though those differences may not be statistically significant. (And remember, just because a T-Test or Logit doesn’t reach statistical significance, doesn’t mean there is no real difference. The small number of homosexual families in the survey mean that getting that p<.05 means drastic differences are present).

3)Remember, there is no way that homosexuals can form an intact biological family. This has implication, as 'evolutionary biology' has shown.
Justin St. Giles Payne says:

June 13, 2012 at 7:31 am

I wonder how many of those claiming this study has been ‘debunked’ have actually read it, let alone understood it.

We’re wondering if any of it’s defenders did, actually. The only way you could defend a study that defines a divorced heterosexual couple as “gay” is if you just fundamentally did not read or understand the study – or, are just so dishonest in your opposition to SSM that, like Rod, you’ll seize on anything that appears to lend even tenuous support.

You’re not even getting the basic statistics right. It’s easier, not harder, to get a Type 2 significance error when your sample size is small; the small sample size, in other words, means that p < .05 is less significant, not more, because it's more likely to be the result of chance. Recall that if you flip an unweighted coin once, the distribution will be 100% one face and 0% the other. That deviation from the expected mean is fundamentally insignificant.

Elementary stuff.
Mitchell Young says:

June 13, 2012 at 12:04 pm

“Elementary stuff.”

Uh, no Mr. St. Giles Payne.

It is indeed true that type 2 errors are easier to get with small sample sizes, which is precisely my, and Regnerus’s point. Type 2 errors mean that the researcher fails to find a difference — at a given ‘p’ level — between two groups’ means or proportions of some variable. The implication is that given the ‘statistically significant’ decision standard (for social science research usually p<.05) and a small sample size, the t-test will only reach significance when there are large real differences between groups. Smaller, but none the less real differences won't reach that p<.05 criteria. Given the tiny percentage of homosexual families out there, getting a group with an 'n' big enough to conduct meaningful t-tests (i.e. ones not prone to large type 2 errors) is virtually impossible.

Your example of the coin makes no sense as is, because a T-test (the test which constitutes the bulk of Regnerus's analysis) is a test between two groups. However, with a little tweaking I can try to make some sense of your coin-flipping. We'd have to come up with a hypothesis. Let's say that is 'the coin is double-headed', with the null hypothesis being 'the coin is standard'. Flipping the coin once and getting a heads would indeed tell me nothing about the truth of one of the first hypothesis.

And guess what, it doesn't reach the p<.05 level of significance either. (It's at the p.05.

The above example also shows that , p<.05 is neither 'less or more' significant (contrary to your language), it is simply an arbitrary (though standard) level. A statistical test either meets that level or doesn't. It is more difficult to meet that test (indeed impossible with a too small sample size e.g. saying, 'yes, this is a two headed coin with only four flips and a significance criteria of p< .05 .

Regnerus and indeed all researchers into homosexual families are confronted with this small sample size problem, because despite media images, homosexuality is rare, and stable homosexual couples raising children are rarer still. The simply don't turn up in great numbers in random samples of the population (and the NFSS is about 3 times the size of the standard Gallup sample), so it is difficult to get significant (p<.05) when comparing them with other arrangements. This is the reason Regnerus gives for his rather maximal definition of a male or female homosexual family.
Fred says:

June 13, 2012 at 2:06 pm

Mitchell Young,

Associate professors almost always have tenure. That is all.
Justin St. Giles Payne says:

June 13, 2012 at 3:43 pm

It is indeed true that type 2 errors are easier to get with small sample sizes, which is precisely my, and Regnerus’s point.

No, it’s certainly neither of your points; your point is that if you get P < .5 in a small sample size, that "proves" significance because the sample size was so small; Regnerus's point is the exact opposite of yours, that his P is significant because his sample size is large. The flaw in his study is that his sample size is large only because he has too broad a definition of who counts as a “same-sex parent.”

The implication is that given the ‘statistically significant’ decision standard (for social science research usually p<.05) and a small sample size, the t-test will only reach significance when there are large real differences between groups.

No, that’s exactly backwards (and, compounding your error, you seem to think that a P value comes out of a t test. This is incorrect – t tests produce t values. Additionally you can perform a t test on a single group.) In a small sample size, it’s easier to get a statistical property to reach false significance because random variation plays such a larger role in the statistical properties of the group. That’s why procedures such as statistical boostrapping exist, to deal with small sample sizes by exploring the variation in the mean or median values by artificially resampling the group.

Again, this is elementary stuff. You just don’t know what you’re talking about. Small sample sizes produce less significant results, not more significant results, because it’s easier to get a P > .5 deviation between X and X-bar simply by random assortment the smaller your sample is. It’s easy to see why – what is the contribution to mean by an individual in your sample when the sample size is one? (It’s 100%.)
Mitchell Young says:

June 13, 2012 at 4:11 pm

Wow, left off a critical phrase. Type 2 errors are when a real difference exists but the researcher fails to get to statistical significance with his or her sample. Also looks like I cut a whole paragraph almost.

If you flip the coin four times, and it comes up heads all four times, it *still* doesn’t meet the p<.05 level of significance. So even with a double headed, coin, a *drastic* real difference from a standard coin, you need at least an 'n' of 5. Regnerus's sample — reflecting the actual US population — didn't even have that many stable, raised birth to 18 in a homosexual household, respondents.
Justin St. Giles Payne says:

June 14, 2012 at 7:13 am

Type 2 errors are when a real difference exists but the researcher fails to get to statistical significance with his or her sample.

No, Type 2 is when you mistakenly reject the null hypothesis (the null hypothesis is always that there is no statistical difference between the sample and the population, or between the treatment and the control.) In other words it’s finding statistical significance when there is, in fact, no significance. And because individual variation is a larger contributor to the mean in a smaller sample size than in a larger one, it’s easier to commit Type 2 error with a small sample than with a large one – not the reverse, as you insist.

If you flip the coin four times, and it comes up heads all four times, it *still* doesn’t meet the p<.05 level of significance. So even with a double headed, coin, a *drastic* real difference from a standard coin, you need at least an 'n' of 5.

…no, Mitchell. Look, this is elementary stuff. Your local community college probably has a stats course that can set you straight on this.

Regnerus’s sample — reflecting the actual US population — didn’t even have that many stable, raised birth to 18 in a homosexual household, respondents.

Which is why it’s meaningless – finding significance in a sample that small is Type 2 error.
Mitchell Young says:

June 15, 2012 at 1:10 am

No, Type 2 is when you mistakenly reject the null hypothesis JSGP

BZZZ -wrong.

“A type II error occurs when one rejects the alternative hypothesis (fails to reject the null hypothesis) when the alternative hypothesis is true.” [my emphasis -my]

Don’t believe that? Here’s another way of phrasing it.

A statistical term used within the context of hypothesis testing that describes the error that occurs when one accepts a null hypothesis that is actually false.[my emphasis--MY]

With small sample sizes, type 2 errors — failing to reject the null hypothesis when it is in fact false — are easy to get. That is, it is easy to not detect real differences between groups. To go back to the coin example.

Hypothesis — coin is double head
Null hypothesis — it is a standard, heas/tails coin

Suppose the hypothesis is true (unknown to the experimenter). He or she flips the coin 4 times. Four heads (obviously) result. With null hypothesis testing the research would ‘fail to reject’ the null hypothesis, because the probability (p) of getting the ‘four heads result’ by chance if the null hypothesis (flipping a standard coin) is true (the experimenter is in the dark, remember) is 1/16, or .0625 and .0625>.05 (the largest standard decision criteria),. The experimenter would ‘fail to reject the null hypothesis’ which was, in fact false — a type 2 error.

Just about the only thing you’ve been correct on JSGP is the existence of a ‘one sample’ T-test. I’d forgotten about it is virtually never used in social research, and at any rate it doesn’t apply to Regnerus’s paper. And — like I said — you’d have stipulate a hypothesis, actually a hypothetical mean or proportion for it to have any meaning.

I’d done schooling you for free, JSGP.
Siarlys Jenkins says:

June 16, 2012 at 3:30 pm

I’m going to continue to speak rationally whether anyone else does or not.