Missing data

[Researcher-4]

Hello experts

Since SPSS assigns the value "system missing" to missing data, what is the advantage of assigning 99 or any other value to missing data over not assigning any value?

Thank you

Matar

[Albert-Jan Roskam]

Hi Matar,

The answer is: the meaning of missingness can differ.
Hmmm, that sounds kinda profound doesn't it? ;-)
Does it mean 'asked, but not answered', or 'never
asked' or 'don't know' or 'not filled in' or ...
You would throw away such information if you just code
everything as sysmis. The user missing definition
enables you to ignore categories, but still retain the
information. It's very important to ask yourself who
those missing cases could be or what they could
represent, for example because it might give you clues
about biases. Try running the following, on a variable
I call 'myvar':
missing values myvar (99).
fre myvar.
missing values myvar ().
fre myvar.

Cheers,
Albert-Jan

--- matara <[hidden email]> wrote:

> Hello experts
>
> Since SPSS assigns the value "system missing" to
> missing data, what is the advantage of assigning 99
> or any other value to missing data over not
> assigning any value?
>
> Thank you
>
> Matar
>

____________________________________________________________________________________
Any questions? Get answers on any topic at www.Answers.yahoo.com.  Try it now.

[statisticsdoc]

Stephen Brand
www.statisticsdoc.com

Dr. Matar,

One can assign various numbers to missing data in order to keep track of reasons why the data are missing.  For example, it may be desirable to differentiate between such reasons as:

99 Coded as missing because of a skip pattern (subject was instructed to skip this question because of an answer to a previous question).

99999 Coded as missing because subject gave an out of range answer

98 Question was asked but not answered.

FREQUENCIES will display how often different types of missing data occur.   You might also decide that some types of missing data cause concern for the validity of the data (e.g., lots of out of range answers on open-ended questions), while other types of missing data are not a cause for concern (e.g., following skip patterns).

HTH,

Stephen Brand



Dr. Matar wrote:

Hello experts

Since SPSS assigns the value "system missing" to missing data, what is the advantage of assigning 99 or any other value to missing data over not assigning any value?

Thank you

Matar

--
For personalized and experienced consulting in statistics and research design, visit www.statisticsdoc.com

[Art Kendall-2]

The distinction between user missing and system missing is a major
strength of SPSS.
Although there may be some rare occasions where it does not make a
difference why the value is missing, it usually does.

There are two ways that data should have system missing as a value:
1) the input data was not compatible with the way it was defined. e.g.,
"aa" could not be read as a number, "nevermore" could not be read as a date.
2) a transformation could not produce a valid result because of its
logic or the particular values that variables have.
When these occur they may indicate how to clean your data or debug your
syntax.

We all make mistakes in data entry and in data transformation, keeping
the distinction, makes it much easier to debug our own work.

Also it makes a difference why the data is missing.
"How many children have you had?" might legitimately be answered zero by
a woman. But would be missing for all males, and for females who are
very young, or in their golden years., etc. A woman might refuse to
answer.  An interviewer might forget to ask the question.

These should be given different user missing values because the user
knows why the values are missing.  The missing status is not a function
of the program.  Why the data is missing influences the analysis and
what you might do to remedy the situation.

Art Kendall
Social Research Consultants

Albert-jan Roskam wrote:

>Hi Matar,
>
>The answer is: the meaning of missingness can differ.
>Hmmm, that sounds kinda profound doesn't it? ;-)
>Does it mean 'asked, but not answered', or 'never
>asked' or 'don't know' or 'not filled in' or ...
>You would throw away such information if you just code
>everything as sysmis. The user missing definition
>enables you to ignore categories, but still retain the
>information. It's very important to ask yourself who
>those missing cases could be or what they could
>represent, for example because it might give you clues
>about biases. Try running the following, on a variable
>I call 'myvar':
>missing values myvar (99).
>fre myvar.
>missing values myvar ().
>fre myvar.
>
>Cheers,
>Albert-Jan
>
>--- matara <[hidden email]> wrote:
>
>
>
>>Hello experts
>>
>>Since SPSS assigns the value "system missing" to
>>missing data, what is the advantage of assigning 99
>>or any other value to missing data over not
>>assigning any value?
>>
>>Thank you
>>
>>Matar
>>
>>
>>
>
>
>
>
>____________________________________________________________________________________
>Any questions? Get answers on any topic at www.Answers.yahoo.com.  Try it now.
>
>
>
>