Aggregated Data (CORRESPONDENCE command)

To analyze aggregated data, such as data from a crosstabulation where cell counts are available but the original raw data are not, you can use the WEIGHT command before CORRESPONDENCE.

Example

To analyze a 3×3 table, such as the one shown below, you could use these commands:

DATA LIST FREE/ BIRTHORD ANXIETY COUNT.
BEGIN DATA
1 1 48
1 2 27
1 3 22
2 1 33
2 2 20
2 3 39
3 1 29
3 2 42
3 3 47
END DATA.
WEIGHT BY COUNT.
CORRESPONDENCE TABLE=BIRTHORD (1,3) BY ANXIETY (1,3).
  • The WEIGHT command weights each case by the value of COUNT, as if there are 48 subjects with BIRTHORD=1 and ANXIETY=1, 27 subjects with BIRTHORD=1 and ANXIETY=2, and so on.
  • CORRESPONDENCE can then be used to analyze the data.
  • If any of the table cell values (the values of the WEIGHT variable) equals 0, the WEIGHT command issues a warning, but the CORRESPONDENCE analysis is done correctly.
  • The table cell values (the values of the WEIGHT variable) cannot be negative.
Table 1. 3 x 3 table
Birth Order Anxiety High Anxiety Med Anxiety Low
First 48 27 22
Second 33 20 39
Other 29 42 47