- Joseph et al. (2015) dub this "Argument B". It is often made as part of argument 1, as twin researchers typically acknowledge, to their credit, that the EEA is wrong insofar as MZ twins are generally treated more similarly than are their DZ counterparts. So the way this argument (#6) works is generally that researchers conduct studies aimed specifically at testing the validity of the EEA with regard to a specific trait, and they typically find no evidence that the EEA is invalid for trait-relevant environmental features.
How do critics of twin studies respond to these responses? They may respond to argument 1 by saying that the studies purporting to find support for the EEA are seriously methodologically flawed (e.g. Beckwith & Morris 2008, Charney 2017, Pam et al. 1996), or argue that limiting assessments of the EEA to "trait-relevant" factors obscures other important environmental factors for which MZ twins' environments are more similar (Joseph et al. 2015). They may even point out that two-thirds of MZ twins are "monochorionic", meaning that they share a placenta, so their environments are more equal than those of all DZ twins even before birth--something it is not possible to control for in twin studies (Charney 2017).
With regard to this last point, recent evidence presented by behavior geneticists indicates that the biasing effects of chorionicity on heritability estimates in twin studies are small and do not affect most phenotypes anyway (van Beijesterveldt et al. 2016). Other BG researchers argue that "... monochorionic MZ twins [are] more likely to differ in birth weight (result of the ‘transfusion syndrome’ and the following differences in availability of placental blood). To the extent that data are available, chorionicity and its consequences do not seem to threaten the logic of the twin method"\cite{Viding_2008}.
What about point 2? In the case of Barnes et al. (2014) and Felson (2014), this point is supported by arguing that heritability estimates are not significantly reduced by controlling for environmental similarity. For instance, Barnes et al. (2014) write, "...numerous studies examining the potential moderating effects of environmental similarity on h2 and c2 estimates have found that violations of the EEA result in statistically nonsignificant parameter deviations". They also cite both previous studies on the EEA and its effects on heritability estimates, as well as their own simulations, to conclude that "...when the EEA is violated, estimates are inflated by between 1 and 5 percentage points." [Emphasis in original.]
Point 3: Often made along with point 2, this is an important and common argument made by BGists to dismiss arguments by critics about the EEA. Often this will involve citing studies such as that of Kendler et al. (1994), which found that "... mothers' and fathers' approach to raising twins had no significant influence on twin resemblance for the four examined psychiatric disorders."\cite{Kendler1994} One example of such a citation is in a recent paper by Gillespie et al. (2017), which states that "A number of studies have reported that violations [of the EEA] do not significantly predict twin pair resemblance for a number of complex behaviours and phenotypes." There's a lot of citations at the end of this sentence in the original paper, and needless to say, Kendler et al. (1994) is one of these citations.\cite{Gillespie_2017}
Criticism 2: But they aren't identical!
Twin studies assume that MZ twins are 100% genetically identical, which is not true. For one, they have significant differences in their epigenetic profiles.\cite{pharmacy}\cite{Fraga_2005}\cite{Haque2009}\cite{Petronis2003} Yet another classical twin study assumption bites the dust. Or does it? Twin researchers can reply to this claim in a number of ways, but their preferred approach is to say that there are no better ways to control for genetic factors than by comparing MZ and DZ twins, because they are so close to being 100% and 50% identical, respectively.
For instance, Verhulst & Hatemi (2013) argue that "... there is no better control of gene expression than identical twins, and there are no better controls for both family environments and genotype than identical twins. Indeed, the best models of genetic expression for complex human traits involve identical twins."\cite{Verhulst_2013} Similarly, Segal et al. (2011), in their response to a Slate article slamming classical twin studies, describe genetic and other biological differences between each twin in a pair of MZ twins. They then argue that "Despite these biological differences, MZ twins (whether reared apart or together) are more alike than any other pair of individuals. Moreover, to the extent that MZ twins are not genetically identical, the true importance of genes is actually underestimated by the MZ-DZ comparison (my emphasis)."\cite{Segal_2011}
This argument (that identical twins not being identical makes heritability estimates from twin studies too low, not too high) is also a common one among twin study defenders, ranging from Segal et al. to Tabarrok's and Collins' responses to the same 2011 Slate article.\cite{revolution}\cite{heritabilitya} It has received empirical support from a recent study in Behavior Genetics, which found that if identical twins aren't quite identical, this leads to heritability estimates being higher than, and only slightly different from, what they should be, based on mathematical simulations.\cite{Liu_2017}
Criticism 3: What about epigenetics and gene-environment interactions?
A recent paper by Richard Lerner and Willis Overton, both longtime and outspoken critics of behavior genetics, argues that "epigenetics invalidates all models involving genetic reduction". They argue that this is the case because, contrary to genetic reductionist models such as that of the classical twin study, "Genes do not determine behavior" (Lerner & Overton 2017). Note that such models are supposed to include behavior genetic models, which they are accusing of being "genetic reductionist" because they (supposedly) grossly oversimplify the ways that genes affect human traits. Another recent paper reviewed the literature on the genetic contribution to aging traits and concluded: "...it is likely that epigenetics contributes to heritability because MZ twins (particularly dichorionic who have split earlier in development) have less DNA methylation difference than DZ twins."\cite{Steves_2012}
And then there is the issue of gene-environment interactions (often abbreviated GxE). As a 2004 Monitor on Psychology article noted, "Some researchers think that interactions between genes and environment, rather than genes and environment separately, may influence many traits. A recent study from Science (Vol. 297, No. 5582) by Avshalom Caspi, PhD, of King's College London, for example, suggests that a gene might moderate propensity for violence, particularly in people who are severely maltreated as children. Many twin study designs don't take this type of complication into account" (Winerman 2004).
But wait, you haven't heard behavior geneticists defend their beloved twin studies against these accusations yet. So what are their rebuttals?
You could take the approach of Barnes et al. (2014), who simply write that "...much of the GxE literature is under heavy criticism and epigenetics is in its infancy". They also use the argument that many (perhaps 90%!) of the findings of reported GxE interactions are false positives to cast doubt on whether such findings exist to a significant extent.
Alternatively, rather than questioning whether GxE effects occur, you could acknowledge that they do, but then say that behavior geneticists are already well aware of them and incorporate them into their models regularly. You could also argue that, sure, the CTD is seriously flawed because it ignores GxE interactions, but argue that behavior geneticists have other types of models they can use that do take into account such interactions. Consider the following quote from a response by two "genopolitics" researchers to a critique of the CTD:
"GE interplay is clearly ingrained within the theory and statistical model of modern twin analyses... the fact that the CTD does not estimate GE interplay is not a failure of the model. If GE interplay is suspected, then extensions of the CTD are available and should be utilized" (Verhulst & Hatemi 2013).
After citing this paper, Hatemi et al. (2014) argue that "Behavior genetic approaches have also inspired philosophical objections due to the firm belief that behavioral differences are entirely socialized. Several publications in the political science literature have begun to erode this view and explicate the theoretical and empirical justification for inclusion of genetic influences and biological mechanisms in general, including gene-environment interplay"\cite{Hatemi_2014}.
So what about epigenetics? Do BGists have any decent responses to the "epigenetics refutes the CTD" argument? Well, they could say this: "Most researchers consider political traits to be influenced by thousands of genetic markers both indirectly and through interactions with numerous environmental stimuli and other genes in complex genomic, epigenetic, and neural pathways. By contrast, many criticisms are developed as if responding to the view that political traits are simple Mendelian traits, governed by a single gene or a small set of genes."\cite{Hatemi_2012}
Another time-honored (but questionable) method is to do a simulation study estimating the effect of (be warned, this is a mouthful) differences in genetic similarity between MZ and/or DZ co-twins from assumed values on heritability estimates. In classical twin studies, genetic similarity between MZ and DZ co-twins is assumed to be 100% and 50%, respectively, and a recent simulation modeling study estimated that when the actual similarity is different from the assumed values, it leads to wrong estimates, but not that wrong: "Although estimates of genetic and nonshared environmental influences from the standard biometric model were found to deviate from “true” values, the bias was usually smaller than 10% points indicating that the interpretations of findings from previous twin studies are mostly correct."\cite{Liu2017}
In short, behavior geneticists argue that the evidence of GxE interactions and epigenetic effects is too preliminary and/or weak and/or inconsistent to cast much doubt on the results of twin studies. Unless they don't, in which case they say, "Uh, we already know about this shit, so you don't need to tell us, and we already take these phenomena into account", and "Don't attack a straw man, we don't think political beliefs are only determined by a single gene, of course other factors are important!"
Criticism 4: These studies assume random mating
Barnes et al. (2014) note that classical twin studies assume random (or assortative) mating. They state further that "...violation of the random mating assumption leads to inflated estimates of the shared environment effect and deflated estimates of heritability." Similarly, Verhulst & Hatemi (2013) write that "...CTD models assume random mating and due to the very large correlation between spouses (r ≈ 0.6) for attitudes, the CTD actually underestimates the magnitude of genetic influences".
Some researchers have used the "nuclear twin family model" to account for this assumption, and the assumption that dominant genetic effects and shared environment aren't present together, in the CTD. This has yielded somewhat lower estimates of heritability.\cite{Burt_2012}
Criticism 5: These studies assume MZ twins, DZ twins, and singletons all have the same prevalence for the trait under study
The CTD assumes that the trait being studied is equally prevalent in MZ twins, DZ twins, and singletons (this last word refers to people who aren't any kind of twin).\cite{Thomsen_2014} BGists respond by simply saying that this assumption is true, e.g.: "Twins are indeed representative of the general population for a wide range of traits and diseases..."\cite{Evans_2000} And in many other studies by BGists, the same claim has been made, namely that twin samples are representative of the general population:
- "...twins do not systematically differ from the general population of non-twins on many measures of behavior and development"\cite{Barnes2013}
- "The TEDS [Twins' Early Development Study] sample has been shown to be reasonably representative of the general population in terms of parental education, ethnicity, and employment status". \cite{Haworth_2008}
- "...results from twin samples generalize to singletons."\cite{Munn_Chernoff_2013}
- "...twins are representative of the general population of MS [Multiple Sclerosis] patients and are probably influenced by the same susceptibility genes and environment as nontwin cases."\cite{Willer_2003}
- "It would appear that, with respect to personality, twins are not systematically different from other people."\cite{Johnson2002}
- "...volunteer twins were not found to differ from age-matched singleton women in distribution or prevalence of: bone mineral density, osteoarthritis, blood pressure, hypertensive drug use, height, history of hysterectomy and ovariectomy, menopausal status and current alcohol and overall tobacco consumption."\cite{Andrew_2001}
But other researchers are more cautious, for example: "Twins are representative of the general population for some but not all measures of ocular biometry. Consequently, care should be taken when extrapolating twin data for these traits in heritability and other genetic studies."\cite{Sanfilippo_2011}
Criticism 6: Like, what even is heritability, anyway?
Critics of BG have long noted that the concept of "heritability" (at least in the context of human BG) is highly misleading because it seems to mean "genetic determination", but it really does not mean that at all. Instead, it just means the % of variation in a population in a given trait that is due to genes. "Genetic determination" is an individual-level characteristic, not a population-level one like heritability, so heritability says nothing at all about how much of any trait YOU have is due to YOUR genes.\cite{Sauce2018} As Ned Block noted more than two decades ago, it is therefore meaningless to ask, for example, "What's the heritability of my IQ?"\cite{Block_1995}
But there are even more problems: heritability refers only to the % of variation of genetic origin in a given population and in a given environment, so changes in the environment can always affect how heritable something is.\cite{Rose_2006} Not only that, but heritability also says nothing about the relative influence of genes or environment on a trait, and since genes definitely have to do with almost all human traits, knowing the percent "heritability" of any such trait is almost always useless.\cite{Moore_2016} Clearly, the common view that high heritability is indicative of current low environmental influence for any trait, or a lack of susceptibility of that trait to environmental changes, is entirely wrong.\cite{Tucker_Drob_2013}\cite{Hopper_2017}
Validated?
One of the key arguments used by BGists to defend the CTD doesn't really fit neatly into any of the categories above. Instead, it is used as though it by itself could answer all the criticisms above, and any other criticisms that could be made of the methodology. This is the argument that the results of CTD studies can be "validated" by producing similar estimates of heritability using different methodologies.
Thus, if this kind of replication of results is found when using non-CTD methods, it is cited by BGists as slam-dunk evidence that the CTD is valid for estimating heritability. One of the main reasons for this argument's strength (apparent or real) is that CTD studies are based on different assumptions than other genetic study designs (specifically, reared-apart twins, family, and adoption studies). One of the main reasons for this argument's strength (apparent or real) is that CTD studies are based on different assumptions than other genetic study designs (specifically, reared-apart twins, family, and adoption studies).
So here are some examples of BGists making this argument (all emphases:
- "Heritability estimates for CB (criminal behavior) from full- and half-siblings closely approximated those found from twins in the same population, validating the twin method."\cite{Kendler2015}
- "For the three externalizing syndromes examined, concerns that heritability estimates from twin studies are upwardly biased...were not supported...The heritability estimates for CB, AUD (alcohol use disorder) and DA (drug abuse) that we obtained from our sibling trios were very similar to those obtained from MZ and DZ twins from the same population using the same diagnostic methods."\cite{Kendler_2016}
- "Moving beyond the twin‐only design leads to the conclusion that for most political and social attitudes, genetic influences account for an even greater proportion of individual differences than reported by studies using more limited data and more elementary estimation techniques."\cite{Hatemi_2010}
- " The aim of the current study is to characterize more reliably than previously possible ECG [electrocardiogram] trait h2 using GREML estimation, and to compare these outcomes to those of the classical twin model...We found no evidence that the classical twin model leads to inflated h2 estimates."\cite{Nolte2017}
- " The aim of the current study is to characterize more reliably than previously possible ECG [electrocardiogram] trait h2 using GREML estimation, and to compare these outcomes to those of the classical twin model...We found no evidence that the classical twin model leads to inflated h2 estimates."\cite{Nolte2017}
- Not too long ago, I sent an email to prominent twin researcher Dorret Boomsma, asking her about how she would respond to classical twin studies' critics. In her reply (which I received July 17, 2018), Boomsma said that she would respond to them by "...pointing to the empirical evidence from extended twin-family designs and molecular studies, which all confirm findings from the classical twin design, as well as findings from twins reared apart studies."
The tactic of persecution
BGists love to construct a persecution complex that they supposedly are hindered by in their research, bravely challenging the orthodox view that all social behaviors are entirely determined by the environment. For instance, Hatemi et al. (2014) write that "The notion of a genetically informed model of attitudes and ideology, understood as a psychological disposition that guides behavior, independent of, and interacting with, social forces presented a fundamental challenge to the dominant rational choice and behaviorist social science paradigms." Thus they portray themselves as bravely going wherever the evidence leads, though they give their ideological genetic-determinist agenda away later in the same paper, when they cite criticism that BG studies of political behavior as "limiting the integration of biological factors into mainstream conceptualizations of political ideology."\cite{Hatemi_2014} Certainly, such a statement seems to indicate that whoever would write or say it would view biological factors as inherently important in political ideology, as an a priori belief, no matter what the evidence really says, no?
I am not the first to note this weird ad hominem tactic of BGists responding to scientific criticisms by accusing their critics of being political ideologues set to deny the potential for biological influences on human behavior. Panofsky even came up with a name for it: the "hitting-them-over-the-head style" (Misbehaving Science, page 142). Like global warming deniers, they will paint themselves as being persecuted for their "heretical" or "politically incorrect" beliefs that challenge the "dogma" of the scientific establishment. But while global warming deniers claim to be questioning the dogma of mainstream global warming science, BGists portray themselves as bravely fighting against the dogmatic beliefs of environmental determinism and the assumption of a total lack of biological or genetic influence on human behavior.
I will close this section by citing more examples of the "hitting-them-over-the head" approach used by BGists dismissing critics as ideologically motivated to deny the role of genes in human behavior completely. Enjoy! (Note: all emphases below are mine.)
- "A disconnect has developed between criticisms that focus on improving existing models and those that seek to abolish or eliminate the entire research agenda, oftentimes for ideological reasons, such as wholesale objections to biological work because of fear of past abuse, or threats to current dominant models. As a result of such largely unspoken existential divides, it has proven difficult for life and social scientists to enter an honest discussion about the limitations inherent in genetic work and still employ the methods in a progressive and useful manner."
- " Since its inception, [behavior genetics] has been a lightning rod of criticism, especially by scholars who are inalterably opposed to linking biology with behavior."
- " Beginning in the 1970s, politically motivated critics of behavioral genetics launched an all‐out crusade against such methods, the findings emanating from them, and even on the researchers themselves (Segerstråle, 2000). These critics called for an end to the idea that biology had anything to do with behavior, noting that sociobiology was a “dangerous idea.” "