) there is no method of detecting GxE methods besides actually bothering to measure a gene and an environment, which behavioral geneticists rarely do these days
2) the issue here is at least threefold with respect to measurement
a) the environment is measured poorly. we have no clue what environmental measures are going to mediate/moderate/interact with genotypes, and we can only measure a limited amount. especially given that our constructs/instruments are never true representations of environments, there is a lot to worry about here
b) genotypes are irreplicably associated with genotypes, and weakly if at all. we all know the replication crisis applies to behavioral genetics perhaps moreso than any other field. GWAS is the epitome of this with poor corrections for multiple comparisons, population structure, etc
subsection i) even when we find a reliable association for a single SNP, it is typically not prevalent enough among a large enough section of the population, *and* because genes are correlated with environments, there is a restriction of range and non-random sampling of the population creating more statistical issues
c) underlying developmental heterogeneity makes it difficult to detect reliable gene-environment interactions, as they often emerge from stochasticity, and can induce violations of methodology assumptions (like the independence of the covariates wrt the outcome, especially for behavioral variables)
3) the sample size required to detect gene-environment interactions is absolutely massive, moreso than detecting both environmental and genetic correlations separately
4) models for gene-gene interactions and gene-environment interactions are still being developed (i linked a few the other day), but they have severe issues with respect to
a) population structure
b) confounds (both genetic and environmental)
c) their statistical properties
d) computability
e) requiring intractable sample sizes
or a primer on detecting gene-environment interactions see Wahlsten, G uo and Moore https://libres.uncg.edu/ir/uncg/f/D_Wahlsten_Intelligence_1994.pdf …
http://pzacad.pitzer.edu/~dmoore/publications/2018_moore_gxe-interaction_.pdf …
https://www.ncbi.nlm.nih.gov/pubmed/10878473
Wahlsten reports examples from experiments and associational research (in animals and humans), Guo develops a statistical model and proves that gene-environment interactions can make polygenic assumptions incorrect in their presence, and Moore talks about the methodological and theoretical assumptions that go into talking about gene-environment interaction