Twin, family, and adoption studies all find concordance rates to determine the potential rate of heritability (vs the influence of the environment).  Association and GWAS studies are looking at specific genes and do not address potential environmental factors.

In family studies, it is often "family history" that is shared to determine if a trait runs in the family.  But even in some twin and adoption studies, it has been the practice to ask the twin about the behaviour of the other twin (who is not part of the study) or to ask an adoptee about their adoptive parents' behaviour. Stronger studies of each type would use medical records to draw conclusions.

Although we can study genes, there are no known behaviours that are the result of a single gene.  We do not yet know how the genes interact.  In addition, the idea that genes can be turned "on" and "off" makes it even more complex.

The two most common strategies for adoption studies are to either use a retrospective approach or a prospective approach.  In a retrospective approach, an adoptee develops a behaviour.  We then want to know whether the biological parents also show that behaviour. In a prospective approach, we know from birth that the parent of the child has a certain behaviour (schizophrenia, antisocial personality disorder) and we want to see if it develops in the child while being raised in a different environment.  The other examples above may be interesting, but they are not true adoption studies.  Comparing different adopted children is looking at the effects of adoption, rather than of the genetic origins of behaviour. If you only compare to a child's behaviour to the adoptive parents, then we cannot know if it was genetic or environmental without doing more genetic testing.  Finally, a child's favourite foods is most likely the result of learning and development, not genetics.

A Manhattan plot looks at the each gene for SNPs - or single nucleotide polymorphisms.  So, when looking at the third gene on chromosome one, we might find that people with depression have one nucleotide (A) and those without depression have the other (G). If this difference is highly frequent, then the statistical difference between the two groups is quite high. The y-axis is the overall statistical analysis for the difference between the control and the target group.  The red line is the level of statistical chance that would indicate that a gene would qualify as a candidate gene.

We are not yet sure what role genetics plays in obesity.  It could be simply that environmental factors play a role in Jack's obesity (e.g. his wife likes to bake and he likes to make his wife happy).  It could also be that stress has led to the expression of specific genes.

When we see numbers like this, it is difficult to know what conclusions to draw. It will partly depend on the size of a sample.  Remember that although DZ twins are not identical, they may share a lot of the same genes.  They could, for example, both have the genes for bipolar disorder.  That could explain why the concordance rates are similar for MZ, DZ and even siblings.  It doesn't mean that we can clearly rule out that bipolar disorder is genetic.

Caspi found that there were people who reported depression even though they did not have the short alleles.  The study is correlational, so a cause and effect relationship cannot be determined. Caspi also makes an assumption that serotonin causes depression, which may not actually be the case.  The study does, however, look at the interaction of genetic mutation and environmental stressors, so the study is not reductionist.