Juha Alho, professor

This course covers the statistical modelling of binary outcomes using re= gression techniques. Binary outcome refers to YES/NO answers to such questions as is the respondent ill o= r not, did the respondent vote for a particular political party or not, did= the respondent buy a particular product or not etc.

Typical background characteristics on which the probability of the two p= ossible responses may depend include age, sex, socio-economic status, life = style characteristics etc.

In this setting the application of ordinary regression techniques is onl= y partly justified. Among the many models that could be entertained, logist= ic regression is favored becau= se of its mathematical tractability. A price one has to pay for adopting it= (or some other variant) is that the maximum likelihood theory = becomes more complicated and the in= terpretation of the parameters is more complex than in ordinary linear regr= ession.

The course emphasises intuitive understanding rather than mathematical p= recision. It should be accessible for all doctoral students in social scien= ces, for example.

R will be used for computation, but no previous familiarity with the pro= gram is assumed.

In order to pass the course, the students are expected to complete a set= of home work assignments that will be graded, and a small final exam.

Did you forget to register? Please contact tilasto-info[at]helsinki.fi.<= /p>