topic Re: Zero Vs missing values in Statistical Procedures

Zero Vs missing values

yael — Sun, 28 Aug 2016 13:04:50 GMT

Hello

I have a question regarding 0 (=zero) vs missing data.

If I have an observation that have many variables but one of the variable is not relevant to this observation (but relevant to other observations), should the value of this variable be 0 (=zero) or missing data (as I found there is a big difference between them in correlation matrix

Thanks a lot for any help!

Re: Zero Vs missing values

Reeza — Sun, 28 Aug 2016 13:12:09 GMT

Missing typically referers to an unknown or unobserved value where 0 refers to a value that is observed to be 0.

Also, is it categorical or continuous?

Given what information you've provided it's difficult to make a suggestion.

Please provide more background information regarding your data.

Re: Zero Vs missing values

yael — Sun, 28 Aug 2016 13:23:24 GMT

Dear @Reeza

Thanks a lot for your fast reply,

The varibale is FD ehich equal to family directors/total directors, this variable is continous.

For the observations that relate to family firms, this variable has any value. But, my data also has observations which are non-family firms. I wonder if under this variable it should be zero or missing value?

Re: Zero Vs missing values

Reeza — Sun, 28 Aug 2016 14:34:59 GMT

If this is proportion on family directors it should be zero.

Note that SAS excludes records with missing values so you need a 0. I'm assuming you have another variable that indicates if the firm is a family firm.

Re: Zero Vs missing values

yael — Sun, 28 Aug 2016 14:51:59 GMT

Thanks a lot, it really helps me 🙂