ABSTRACT
Statistical method for analysing the respective contributions of variables (usually 2 or 3) to the variance in the data. Given two sets of scores, e.g. (a) 10, 10, 10, 10, 10, 10 and (b) 12, 12, 12, 12, 12, 12, it is intuitively obvious that all the variance is due to the sets (of which there may of course be more than two), if they were: 10, 12, 10, 12, 10, 12 and 10, 12, 10, 12, 10, 12, it equally obvious that it is all ‘internal’, if they were 10, 11, 10, 11, 10, 11 and 11, 12, 11, 12, 11, 12, then half is obviously due to the sets and half internal. More subtly, if they were 10, 11, 10, 11, 10, 11 and 12, 15, 11, 14, 12, 10, there would be an additional ‘interaction’ effect since the amount of internal variance itself also differs. In real life, however, it is invariably more complicated than that, requiring statistical analysis to calculate. This would involve a ‘2-way’ ANOVA, ‘3-way’ ANOVA being used when there is an additional variable (e.g. you might be comparing male and female performance in two or more conditions). The ANOVA enables one to unravel the various sources of variance and determine which are statistically significant. In cases where the number of variables exceeds 3, it is usually more appropriate to use factor analysis.
