The Quantitative Literacy Gap in Sociology Undergraduate Education

Thomas Linneman wrote an article appearing earlier this year in Teaching Sociology that documents a continual upgrading in the statistical methods used in sociology articles.  He asks the reader to ponder whether or not sociological statistics courses are preparing undergraduate students to read most published sociological quantitative investigations (they are not) and for statistics instructors to think about covering interaction and nonlinear terms in regressions and logistic regression so undergraduate students.  This way, our undergraduate students will at least be able to understand the bulk of published quantitative sociological work.

I am sympathetic to Linneman’s aims here.  In my own upperclassmen statistics course I have the students read David Brady et al.’s “Rethinking the Risks of Poverty” which uses multi-level linear probability models with interaction terms.  I try mightily to get the students to understand interaction terms so they can understand the statistics that Brady and his co-authors present.  Interactions come at the end of the term; in some semesters this is more rushed than it should be and when that happens I wonder if my students would have been better off if I hadn’t covered interactions at all.

I am having a hard time envisioning how I could cover the other “advanced” topics that Linneman advocates for (nonlinear effects and logistic regression).  The main issue is I need to review some basic quantitative literacy concepts which take up a third of the semester before I can even dive into regression.

The only way I could see this working is if I dispensed with spending class time on basic quantitative literacy (e.g. percentaging tables, comparing quantities, measures of central tendency, levels of measurement).  Instead, I would either (a) offload those topics to readings students would do on their own time, or (b) hope the General Education quantitative literacy course required of undergraduate (who may take it before, during, or after they take my statistics course) covers those things.  I am not sure either option is that appealing.

I don’t really have a solution here; I think Linneman’s goal could be met more easily if (a) the department requires majors take multiple statistics courses and (b) the instructors of the two courses work closely together to coherently sequence the content covered.