I have previously posted about figuring out and labeling missing values.
A couple of list members asked me to elaborate so I am pasting from my reply
emails.
*/Understanding missing data is one key to understanding the uncertainty in
empirical statements./*
One way of explaining the importance of specifying why data is missing is to
do a mental exercise thinking about a dataset with one case per member of
the population of interest. We try to reason by analogy from the cases in
the dataset to the cases in the population of interest.
As used here population is the complete set of entities about which ones to
reason. Members of a population might be persons, households, families,
postal codes, counties, pieces of mail, trees in a forest, and so forth.
Sampling theory helps us get a perspective on part of the uncertainty.
However, that is just part of the uncertainty.
The dataset has variables that are measurements of various constructs/ideas.
We get a handle of measurement uncertainty by considering how well the
measurement represents the construct by looking at the validity, the
reliability, the coarseness of measurement, and so forth.
Complete the individual or focus group exercise about looking at elections
with something like this.
Consider a dataset with one case per person in a jurisdiction. Below are
some examples of the kinds of missing data values a case might have on
various variables --
NOT SELECTED by a scientific sampling method.
Not eligible to vote: underage
Not eligible to vote: not a citizen
Eligible to vote not registered.
Registered without specifying the party.
Did not vote at all
Did not vote for this office.
Marks not readable this ballot.
Marks not readable this office
Ballot mangled.
What are some other ways data could be missing in this context?
-----
Art Kendall
Social Research Consultants
--
Sent from:
http://spssx-discussion.1045642.n5.nabble.com/=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Art Kendall
Social Research Consultants
Locked
