Twin studies are often used in the study of the heritability of behaviour. In carrying out twin research, the goal is to see whether the rate of behaviour occurring in both twins is higher in monozygotic (identical) twins or in dizygotic (fraternal) twins. The rate of similarity is called the concordance rate.

Research by Stunkard et al (1986) showed that the concordance rate for weight was higher in MZ twins than in DZ twins, indicating a 77 - 85% rate of heritability; this was true across the spectrum of thin to obese. If weight may have a genetic root, then what about eating disorders?  Considering that eating disorders appear to run in families, the following study looks at both a twin study and a family study in order to determine the level of heritability.

Holland et al (1988) wanted to see whether monozygotic twins would have a higher concordance rate for anorexia nervosa than dizygotic twins. As MZ twins share 100% of the same genes, if there is a genetic link to anorexia nervosa, it should be more often the case that both MZ twins have the disorder than both DZ twins.  In other words, there should be higher concordance for MZ than DZ if there is a genetic basis for anorexia nervosa.

The research team studied 45 pairs of female twins. 18 of the pairs were recruited from a hospital clinic; 27 were volunteers that were recruited through magazine advertisements of the study. In all cases, at least one of the twins had to have anorexia in order to be in the study. There were 25 Monozygotic (MZ) and 20 dizygotic pairs of twins.

The twins were interviewed separately and at the same time in order to avoid contamination of the data. The interview was a structured diagnostic interview for anorexia nervosa, based on the DSM III. In addition, they were asked to complete a number of questionnaires that focused on eating habits and personality traits. 

Finally with each pair of twins a family pedigree was constructed for psychiatric disorders. This was done to determine the prevalence of eating disorders in first and second-degree relatives.

The researchers found a 56% concordance rate for the MZ (identical) twins and only 5% for the DZ (fraternal) twins. The MZ twins had a lower mean minimum BMI and higher scores "drive to thinness" and "body dissatisfaction."  The fact that the percentage for MZ twins was well below 100% indicates that genes are not wholly responsible.  Thus, genes can provide a predisposition, i.e. they make the individual vulnerable but do not directly trigger the disorder.

In addition, 4.9% of female first-degree relatives and 1.16% of female second-degree relatives had had anorexia nervosa.  These are figures are higher than the 0.1% prevalence of anorexia nervosa in the female population. The researchers argued that the rate of heritability is approximately 80%.

Zygosity was determined by blood testing, guaranteeing that the MZ twins were in fact, identical twins.

The twins were interviewed separately and simultaneously to avoid data contamination and to avoid bias. Standardized diagnostic interviews were used which should lower the risk of interviewer bias.

The prevalence of anorexia in families was confirmed by Strober et al (2000). This means that this part of the results is reliable.

Twin studies are natural experiments. Therefore, they cannot establish a cause-and-effect relationship between an IV and DV. The internal validity is low, as factors other than the IV may have resulted in anorexia. Confounding variables such as personal experience, environmental stimuli and diet may have played a role in the disorder.

The study suffers from ascertainment bias.  By only looking at families with twins in which eating disorders are identified, there is no study of other families to see the extent to which eating disorders may run in families. The study does attempt to overcome ascertainment bias by using not only participants who are going to a clinic for help but also asking for volunteers from the community.

A genetic etiology is seen as oversimplified and reductionist. It is also not clear what particular physiological process is controlled by the genes which then leads to the disorder.