Table 4

Discriminant Function Analysis Results.

A

Classification Functions

Classification Matrix

CES-D item

controls

depressed

D-C

Rows: Observed classifications

No

Constant

-1,07

-10,10

-9,03

Columns: Predicted classifications

Item No 2

0,05

1,49

1,43

% Correct

controls

depressed

Item No 5

0,39

1,41

1,01

Controls

98.33

118

2

Item No 6

-0,23

0,89

1,13

depressed

87.5

5

35

Item No 7

0,57

1,19

0,63

Total

95.62

123

37

Item No 9

-0,38

0,56

0,94

Item No 10

0,29

0,94

0,65

Item No 11

0,12

1,09

0,96

Item No 13

0,70

1,77

1,07

Item No 14

0,46

-0,14

-0,61

Item No 17

-0,11

1,21

1,32

Item No 19

0,21

-0,99

-1,20

Item No 20

-0,20

-1,03

-0,83

B

Classification Functions

Classification Matrix

Factor No

controls

depressed

D-C

Rows: Observed classifications

Constant

-0,28

-1,45

-1,16

Columns: Predicted classifications

Factor 3

-0,09

0,26

0,35

% Correct

controls

depressed

Controls

100

120

0

depressed

0

40

0

Total

75.62

160

0

A: Analysis with individual CES-D items entering the procedure. When the equation: -9,03+1,43*(Item No 2)+1,01*(Item No 5)+1,13 *(Item No 6)+0,63*(Item No 7)+0,94*(Item No 9)+0,65*(Item No 10)+0,96*(Item No 11)+1,07*(Item No 13)-0,61*(Item No 14)+1,32*(Item No 17)-1,20*(Item No 19)-0,83*(Item No 20)>0 is true, then the subject is a depressed patient. This method correctly classified 98.33% of controls and 87.5% of depressed patients B: Analysis with factor scores entering the procedure. When the equation: -1.16 +0.35*(factor 3 score)>0 is true, then the subject is a depressed patient. This method correctly classified all controls but no one depressed patient

Fountoulakis et al. BMC Psychiatry 2001 1:3   doi:10.1186/1471-244X-1-3