Genetic similarities – twin and kinship studies on antisocial behaviour and aggression

Travis DixonBiological Psychology, Research Methodology

Twin and kinship studies are important methods used in biological psychology as they allow researchers to see links between genes and behaviour.

Clarifications to the new IB Psychology guide (first exams May 2019) has said that students might be asked specifically about twin and/or kinship studies to discuss “genetic similarities.” This blog post can be used as supplementary information if required.

How and why twin studies are used in psychology

Twin studies gather groups of monozygotic (MZ – identical) and dizygotic (DZ – non-identical) twins and they measure their behaviours using a range of tests. For example, they may use questionnaires and interviews to measure their levels of antisocial or aggressive behaviour.

After gathering behavioural data, the researchers compare the similarities between MZ twins and DZ twins by calculating the concordance rate. The concordance rate of a behaviour is a measure of how similar (on average) the twins are. More specifically, it measures the probability that if one twin has a particular characteristic the other twin will also have it. For instance, if a concordance rate is 1.0, it means that there is 100% chance that if one twin has a characteristic (e.g. they commited a crime) then the other will as well.

If a behaviour is 100% genetically based, then we would expect the concordance rate for MZ twins to be 1.0 (or 100%) because they are genetically identical.


Gottesman and Goldsmith (1994, as cited in McGue and Bouchard 1998),  – Link) studied 85 sets of young twins and found that in children, the concordance rate of juvenile delinquency² in MZ twins was 0.91 (91%) when compared to 0.73 (73%) for DZ twins. This means that if one identical twin broke the law, there was a 91% chance the other twin had as well. For DZ twins, this dropped to 73%. The difference suggests genetics is a factor, because both types of twins share 100% of the environment and so if behaviour was because of environmental factors, we would expect closer or identical rates of concordance between MZ and DZ twins. Because they’re higher for MZ twins, we know genetics is a factor but since it’s not 100% for MZ twins, it’s not the only factor.

In this same study, studied over 500 adult twins and the researchers found a criminal conviction concordance rate of 0.52 for adult MZ twins and 0.23 for DZ twins. In other words, if one MZ twin broke the law, there was a 52% chance that the other one had as well. Once again, we can see that genetics is a factor because the likelihood that a person commits a crime is almost double if their identical twin commits a crime, compared to if it’s their DZ twin. However, we can’t say that this behaviour is 100% genetics because the MZ twins still are only 50% concordance – if criminal behaviour was 100% genetic we would expect the concordance rate to be 100%.

Kinship Studies

MZ and DZ twins aren’t the only participants in these kinds of studies. Some researchers compare MZ and DZ twins but also include normal siblings, half siblings, step and adoptive siblings. Making comparisons between these individuals based on genetic relatedness can also help researchers draw conclusions about the extent to which behaviour is a product of genes or the environment.

In a really interesting study by Natsuaki et al. (2009, link), the researchers compared 390 sibling pairs (780 participants), including MZ, DZ, full, half and genetically unrelated siblings (e.g. step-siblings). The kids were measured for “externalizing problems.” These are examples of antisocial behaviours that are directed towards the external environment (e.g. towards others). Examples of externalizing problems include physical aggression, vandalism and stealing.

The table shows the results and how similar the sibling pairs of various genetic relatedness were in terms of externalized antisocial problems. You can see that MZ twins were most similar, followed by DZ and so on (the higher the correlation score the more similar the twins in their rates of externalized antisocial behaviour). Interestingly, unrelated siblings scored higher than full siblings, but a further statistical analysis found that these results were not statistically significant.

Genes and Behaviour (Natsuaki et al. 2009)

This table shows how similar sibling pairs of various genetic relatedness were in terms of externalized antisocial problems. The intra-class correlation coefficient shows how closely the siblings are in their rates of externalized (antisocial) behaviour.

This is a good example of a kinship study as it’s not just studying twins, but is also testing a range of genetic relatedness and provides further evidence for the role of genetics in antisocial behaviour. However, just like with concordance rates if behaviour was 100% genetic we would expect a correlation closer to 1.0 (the highest score).

One really interesting (but seemingly common sense) finding from this study was that when they controlled for other factors, they found that the behaviour of one’s sibling has an effect on behaviour. This has been called “deviancy training” (a type of social learning) and while the effect was small, it could be an important consideration in explaining behaviour – genes can affect behaviour, but the environmental factor of how siblings interact with one-another on a day-to-day basis that can have an impact on behaviour.

Critical Thinking Considerations

  • Why is it problematic to assume that 100% of twins’ environments are shared? What environmental factors could still be different for MZ and/or DZ twins?
  • If MZ twins don’t have 100% concordance for a behaviour, we know the behaviour is not purely genetic and other factors must be involved. However, if there is 100% concordance, we can’t necessarily conclude that the behaviour is 100% genetics. Why not?
  • One common limitation given to correlation studies is that there is a problem with bidirectionality – is it possible for this to be an issue in twin studies?
  • Twin and kinship studies show links between genes and behaviour. But is this useful? Why is it limited? How could identifying specific genes and experimenting on these (for example in animal experiments or studies on the MAOA gene) be more beneficial?


1: the extent to which variations in behaviour can be attributed to genetics.

2: a crime committed by a minor (e.g. <18 years old)