Missing data

[Lena Gan]

I am looking to analyze missing data so I tried with command mdesc and summ but it came back with no missing data. When I tried codebook it came back with 98 and 99. These are not defined as missing data but rather "don't know" and "prefer not to say." Is this correct? Is this why the mdesc and summ did not show them as missing? Command misstable only works if I conduct multiple impution but does not list out the missing values. Please advise. Thank you.

[Jared Greathouse]

The syntax for mdesc escapes me now, but that's likely why it didn't come up in the mdesc table, as system values (I think they're called) for missing survey variables

[Rich Goldstein]

yes, that is why your commands did not show these as missing; use -mvdecode- if you want to change this; see

help mvdecode

[William Lisowski]

Stata's knowledge of missing values is limited to those special values it defines as missing - not to the special values that your dataset uses to denote missing values. See

help missing

for a comprehensive discussion, and as Rich notes, you can use mvdecode to convert special values like 98 and 99 to Stata missing values - and not only that, but you can convert each to a different one of Stata's 26 "expanded missing values" so you can distinguish between them.