Path: Janda's Home Page > Workshop > Factor Analysis > Outline
Inter-correlation matrix

Although rectangles could be drawn so that long rectangles have short widths and wide rectangles have short lengths, we expect to find measures of height and width to correlate positively, and for this set of 100, they do:

Intercorrelations
2
3
4
5
6
7
8
1
Length
0.14
0.99
0.17
0.93
0.80
0.60
0.98
2
Width

0.16
0.93
0.49
0.69
0.88
0.33
3
10Length + e

0.19
0.93
0.81
0.61
0.97
4
10Width +e

0.49
0.67
0.84
0.35
5
20L + 10Width + e

0.96
0.85
0.98
6
20L + 20Width + e

0.95
0.90
7
10L + 20Width + e

0.74
8
40L + 10Width + e

What are the "principal components" of the variation in this set of correlations?

Go to the next page, # 4: The principal components of 100 rectangles