Object Scores

You can also request a listing and plot of object scores. The plot of the object scores can be useful for detecting outliers, detecting typical groups of objects, or revealing some special patterns.

The object scores table shows the listing of object scores labeled by social group for the Guttman-Bell data. By examining the values for the object points, you can identify specific objects in the plot.

Figure 1. Object scores
Object scores

The first dimension appears to separate CROWDS and PUBLIC, which have relatively large negative scores, from MOBS and PRIMARY GROUPS, which have relatively large positive scores. The second dimension has three clumps: PUBLIC and SECONDARY GROUPS with large negative values, CROWDS with large positive values, and the other social groups in between. This is easier to see by inspecting the plot of the object scores.

Figure 2. Object scores plot
Object scores plot

In the plot, you see PUBLIC and SECONDARY GROUPS at the bottom, CROWDS at the top, and the other social groups in the middle. Examining patterns among individual objects depends on the additional information available for the units of analysis. In this case, you know the classification of the objects. In other cases, you can use supplementary variables to label the objects. You can also see that the categorical principal components analysis does not separate MOBS from PRIMARY GROUPS. Although most people usually don’t think of their families as mobs, on the variables used, these two groups received the same score on four of the five variables! Obviously, you might want to explore possible shortcomings of the variables and categories used. For example, high intensity of interaction and informal relationships probably mean different things to these two groups. Alternatively, you might consider a higher dimensional solution.

Next