Lurking variable is a variable which "is camouflaged by another variable that is usually deceptively appealing" (Borradaile G. J. 2003, s. 136). Lurking variable may have an effect on other variables that are taken into consideration. It is however, "hiding in the background and it affects the data". So before we are able to make any assumptions, we need to take into consideration the effect of lurking variable on other data. This is especially difficult as this variable is often not even recognized to be relevant for the association (Bolstad W. M. 2004, s. 2).
Features of lurking variables
Lurking variables have the following features (Brase C. H., Brase C. P. 2010, s.129; Mathews P. G. 2005, s.173):
- may be known or unknown;
- may affect other variables in an unpredictable way;
- "may cause erroneous biases to appear" in the calculation.
Randomization helps protect the calculation from lurking variables although the effect of lurking variables is still not completely remunerated because we are not able to foresee all of the lurking variables and their effect. This conclusion is widely accepted therefore without randomizing a calculation involving lurking variables, your calculation will not be positively perceived by the people that have knowledge in the area (Mathews P. G. 2005, s.173-175).
The effect of lurking variables
Lurking variables affect two types of variables: the predictor variable and the response variable (Tamhane A. C. 2012).
An example of this may be a situation when a teacher does not take into account the possibility that introverted students may react in a different way to a positive reinforcement than extraverted students. Extraverted students tend to react more positively when positively reinforced whereas introverted students tend to react more positively when negatively reinforced. If the teacher misses this lurking variable and comes up with a training program based mainly on positive reinforcement, the effect will be visible only on some of the students.
Therefore, here a reaction to positive reinforcement (predicted variable) as well as the actual end effect (response variable) get affected by lurking variable, which is student's introverted or extroverted personality (Lerner R. M. 2015, s. 791).
- Bolstad W. M.(2004), Introduction to Bayesian Statistics Wiley Interscience, s. 25-27.
- Borradaile G. J. (2003), Statistics of earth science data: their distribution in time, space and orientation Springer Science & Business Media, s. 136.
- Brase C. H., Brase C. P. (2010), Understanding basic statistics Brooks/Cole, Cengage learning, s. 129-131.
- Lerner R. M. (2015), Handbook of child psychology and developmental science, theory and method John Wiley & Sons, s. 791,807.
- Mathews P. G. (2005), Design of experiments with MINITAB ASQ Quality Press, s. 172-174.
- Tamhane A. C. (2012), Statistical analysis of designed experiments: theory and applications John Wiley & Sons, chapter 1.1.
Author: Justyna Szczepaniec