Stop missing data!

Missing data is not ideal, but is a reality with most research.  This is especially true of social science research where data often comes from subjects who are not forced to complete a survey or question they do not want to.

Researchers get scared by the idea of “imputing” data for missing values, but they need not be.  When handled correctly, missing data techniques simply correct for bias and decrease the likelihood of committing a Type II error.  No need to worry.  They are better than the alternative–pairwise or listwise deletion.

Researchers fright with implementing “state-of-the-art” missing data techniques suggests missing data techniques are (apparently) not well understood.  If only a handy guide existed–something you could cite to help determine and substantiate what types of missing data analysis you should (and did) engage in.  And code… if only someone gave you the syntax to make these missing data problems (relatively) easy to solve.

Good news!  The guide (and code) is now here!  As author Dan Newman mentions in Guideline #0 (yes, there is a Guideline 0),  “Abstinence is not an option” (p.383)–deleting responses with missing data is a missing data technique (and a sub-par one to be sure).  With a Guideline 0 and titles like these, you might just want some missing data be the end of your read.