A measurement model for the Rorschach

Graphs presenting the relations between protocol level number of answers, R, protocol level number of M answers, the factor score based on a card level Poisson model (M1), and the factor score corrected for number of Responses (M28):

Graph with histograms and scatterplots for the four variables

Graphs with highlights of different groups of persons. Each group are identified by stars in the scatterplots and the corresponding bars in the histograms inverted to the 4. quadrant.

Highlight1, highlight2, highlight3, highlight4

The graphs shows a moderate correlation between M and R.

The scatterplot between M and M1 shows that the same number of M may correspond to different factor scores, because the scorings are based on different patterns of cards which have different intercepts and factor loadings. Still, there is correlation between R and the factor score

The factor score M28 depends both on the patterning of M answers on the cards and on the total R. The scatterplot between R and M28 shows no correlation, and the starred points indicates that persons with the same M28 score may have R scores over the total spectrum.

The scatterplot between M1 and M28 shows a correlation, but the starred dots indicates the the work of the correction for R. Persons with the same M score but higher R, have lower M28 factor score (similar to the principle of a M% calculation)




The graphs are made with S-plus..