## What is a Lurking Variable?

A lurking variable is a variable that is unknown and not controlled for; It has an important, significant effect on the variables of interest. They are extraneous variables, but may make the relationship between dependent variables and independent variables seem other than it actually is. In other words, the variables will cause your results to be biased. In addition, any correlation or regression analysis you perform will be misleading. *How *misleading these analyses are will depend on how severely the lurking variables affect the dependent variable.

## Example

Recently, there has been a strong correlation between diet soda consumption and traffic accidents. While you might take this a step further and think that it could be caffeine consumption that’s the cause of the accidents (or perhaps a caffeine hangover), a more likely cause is simply an increasing population (Brase & Brase, 2006).

## Lurking Variables and Accidental Bias

The introduction of one or more lurking variables into your experiment results in accidental bias, the more technical term for the errors that creep into your results. Lurking variables are intrinsic to your study; They aren’t actually caused by “accidents” at all, which is why Soares (1985) suggested that “lurking-variable bias” is a more suitable name for accidental bias.

## Identification

For research results to be valid lurking variables must be identified and then either eliminated, held constant, or included in the study. They can be identified with regression analysis: plot the residuals, and if you see a trend (either linear or non-linear), this is evidence that particular variable is affecting the response variable (Fligner et. al).

## References

Brase, C. & Brase, C. (2006). Understandable Statistics: Concepts and Methods. Cengage Learning.

Fligner, M. et al. (2003) The Practice of Business Statistics Student Solutions Manual.

Soares, J. (1985) Optimality of random allocation design for the control of accidental bias in sequential experiments. Journal of Statistical Planning and Inference 11 81-87.