Contrast Analysis

Contrasts as a Unique Analysis

A contrast analysis is useful when a researcher wishes to be more specific when analyzing differences among groups. For example, a researcher might have six groups with predicted differences. The researcher predicts that X1, X2, and X3 will be equivalent, whereas X4 will be twice as strong as X1, X2, and X3, whereas X5 and X6 will be twice as weak as X1, X2, and X3. Written out, the following predictions could be given the following coefficients:

This shows that X1, X2, and X3 have a coefficient weight of 2; X4 has a coefficient weight of 4; and X5 and X6 have a coefficient weight of 1. To determine the contrast coefficients for the predicted differences, the researcher would take the mean of the coefficients and subtract that from each individual coefficient. In this example, the mean of the coefficients in these predictions is 2. Subtracting the mean from each coefficient would render the following rankings of each group:

The resulting contrast coefficients, when summed, should equal 0. In this example, 0+0+0+2−1−1 = 0, thus the rankings can be considered the contrast coefficients for proposed contrast. The purpose of this is to rank each individual group to be representative of the predicted contrast proposed. This example illustrates this concept through integers, but when using contrast analysis researchers will often use more complex values to represent the contrast coefficients of each group.

...