Threshold I:

Introduction to the Threshold Model

The threshold model should really be called the multifactorial threshold model. It assumes (1) that the many factors (hence, the term multifactorial) contribute to a disorder; (2) that the effects of each single factor are small but the effects add up; and (3) once the additive effects of the factors pass some critical value (the threshold), one becomes affected with the disorder. In short, there is multifactorial causality but a dichotomous phenotype (affected vs. unaffected).

The figure above depicts the threshold model. The horizontal axis may be termed the liability, the vulnerability, the susceptibility, or the predisposition for a disorder. This is a very important concept and represents the sum of all the multifactorial effects. It is termed a latent variable because we can never directly measure the total liability. (We may be able to measure several individual causal effects that contribute to liability, but we cannot measure everything that contributes to it.)

The horizontal axis gives the frequency of individuals with a specific liability. You have probably noticed that the curve is a normal curve. Because we cannot directly measure the total liability, it is customary to scale the curve as a standard normal distribution (i.e., a normal distribution with a mean of 0.0 and a standard deviation of 1.0).

The vertical red line represents the threshold. Individuals whose liability exceeds this value will be affected and are depicted by the red area under the normal curve. Individuals whose liability is below the threshold will be unaffected.

Most psychopathology shows a variable age of onset, so behavioral geneticists are interested in the concept of lifetime risk or lifetime prevalence for a disorder. Lifetime risk is a numerical estimate of the answer to this question: if we took a population and followed them to a very old age, what proportion of the population would have developed the disorder at some point during their lives? Geneticists and epidemiologists begin estimating lifetime risk by questioning a large number of people in the general population about the symptoms for a disorder and at what age those symptoms began. Complicated mathematics (that need not concern us) is then used to compensate for the fact that some of the unaffected people are still within the age of risk and may develop the disorder at a future time.

After lifetime prevalence is estimated, the mathematics of the normal curve is used to find the threshold. In this exercise, you can enter the lifetime risk into the field labeled "Proportion affected", click on the "Find threshold" button, and the curve will adjust itself to show you the model. Note well that the proportion affected should be number between 0 and 1.0 and not a percentage affected. If 4.5% of the population were estimated to develop an anxiety disorder, then enter .045 and not 4.5. One could also enter a threshold value and find the lifetime prevalence (proportion affected).


  1. Schizophrenia has a lifetime risk a bit less than .01 in the general population, but it is convenient to think of the prevalence as the round number of 1%. Enter .01 for schizophrenia and examine the model.
  2. It is estimated that as many as 25% of males will meet diagnostic criteria for alcohol abuse or alcohol dependence at some point in their life, making it the most frequent form of psychopathology. What does the model look like for alcohol problems in males?