Thanks for getting us back on track, Igor.
I agree generally with your thrust, but let me focus on what we may disagree about.
There's no reliable method for estimating the heredity of a trait, even a Mendelian one. For complex polygenic traits, there's only the most well-understood, often difficult to arrange, but gold-standard-when-you're-lucky-enough-to-get-them methods which are available at present, and which are responsible for the current progress of the brave young science of genetics, such as it is.
One of those is the twin study, involving the contrasting of MZ and DZ twins. It's not perfect, and it's often very hard to find enough twins (the homophobia and homosexuality studies are quite fortunate, in that regard), but it's valuable when you're lucky enough to have the data. Imperfect as they are, the best methods of genetics have proven themselves to be among the very most reliable of methods in "the imperfect science" of biology.
As the study I mentioned cites:
There is considerable support for the existence of generalized prejudice (see Ekehammar et al. 2004). Different types of prejudice (e.g. racism, sexism and prejudice toward homosexuals) have been shown to be highly correlated (e.g. Bierly 1985; Ekehammar and Akrami 2003), implying that people who reject one out-group will also tend to reject others. As Ekehammar et al. (2004) describe in their article, two major theories have risen to explain why some individuals are more prejudiced than others. In the first theory, individual differences in prejudice are considered to be due to stable factors within the individual (their personality characteristics); in this case prejudice is seen as an expression of personality (Ekehammar and Akrami 2003; Heaven and St. Quintin 2003). The second theory implies that individual differences in prejudice are caused by factors linked to the outside world, like intergroup relationships and social life (see e.g. Guimond et al. 2003; Guimond 2000; Reynolds et al. 2001).
Earlier twin studies have demonstrated that individual differences in personality are substantially heritable (e.g. Jang et al. 1996; Loehlin et al. 1998), and therefore, it is possible that there are genetic influences on homophobia as well.
...
The MZ twin pair correlations are significant higher than the DZ twin pair correlations (χ12 = 15.64, P < 0.001), suggesting that genetic effects are a source of familial aggregation in attitudes toward homosexuality.
...
Subsequently, a common effects sex-limitation model is fitted, by fixing the genetic correlation for DZ opposite sex twins at 0.5. Results show no significant deterioration of model fitting (χ12 = 1.43, P = 0.23), consistent with the same sets of genes influencing homophobia scores for males and females.
...
Under the general ACE model, additive genetic and common environmental effects together account for between 48% and 59% of the variance in homophobia (see Table 5). It is impossible to distinguish between AE and CE models, as removing the effects of either A or C causes significant deterioration in model fit.
The same can be said of MSM itself. The genetic and environmental factors
can be roughly estimated.
http://en.wikipedia.org/wiki/Biology_an ... in_studiesNot only are such studies inconclusive, there isn't a definition of "conclusive" out there. Only progressively less rough, less inconclusive estimates. Even a 100% gene/trait correlation isn't "proof" of causation...
... on the other hand, our best scientists, doctors, and agriculturalists 100% sure do treat sufficiently high quality correlation as proof in everyday, life-and-death reality. Because it's all science can do. And it's not bad.
Confounding does not reduce our best methodology to dismissable, know-nothing guesswork, either. Especially not when they are performed with the highest standards of double-blinding and anonymity-shielding, as in this case.
To the extent that any such knowledge must be taken seriously, and we are not to abandon empiricism in favour of solipsism, this type of data
should be accorded due respect, unless one notices error in the study design (?).
Properly controlled association research is "merely" estimation, yet also some of the hardest scientific knowledge in biology, and "merely" the best of what science can realistically deliver, at present. Life-or-death medical treatment decisions, and agricultural economics policies governing billions of dollars, are made on much less every day, because genetics is - though ostentatiously humble - still exceedingly substantial, and unlikely to get conservative numbers like P < 0.001 wrong without a glaring methodological error. In the cases of both homosexuality and homophobia, the data
does speak substantively about their genetic heritability, as the authors of the studies to which I link are bold enough to state outright, under peer review by the greatest minds of their field.
Genetics is "just correlation" in somewhat the same sense that evolution is "just a theory".