| Missing Data | Various codes are used to describe missing data and special values. The general strategy is as follows:
Any question that is skipped due to the flow of the questionnaire (not applicable) is coded as a blank in the dataset.
Any question that should have been answered according to the flow of the questionnaire, but no response was recorded on the questionnaire (missing) is coded with a field full of 9s for a numeric field, or with a field full of question marks for an alphabetic field.
Other special codes are used in a standard manner throughout the file. The codes used are recorded with a field full of 9s, but with the final digit being one of the following below:
Don't know 8
Inconsistent 7
Other 6
For example, if the month of birth was unknown, as the field requires two digits, the value would be 98.
For alphabetic fields, the following codes were used:
Don't know Z
None/no one Y
Other X
In summary
1 digit 2 digits 3 digits 4 digits Alpha fields
Not applicable b b b b b
Missing 9 99 999 9999 ?
Don't know 8 98 998 9998 Z
Inconsistent 7 97 997 9997
None/no one 0 00 000 0000 Y
Other 6 96 996 9996 X
Special responses 95,94... 995,994... 9995,9994... W
where b indicates a blank space.
Fields requiring more than 5 values (excluding special values above) and less than 96 values are given at two digits with leading zeros for codes below 10. All fields requiring more than 95 values and less than 995 values are given 3 digits with leading zeros for codes below 100.
|
