Sociology 3306A/B Lecture Notes - Lecture 5: Logistic Regression, Sidney Crosby, Spurious Relationship
Document Summary
Interaction terms: we always work with this sort of model in sociology. In sum: through the introduction of control variables there is evidence for, spuriousness, chain relationships (very common, suppressor variables, statistical interaction. It must satisfy the condition: the difference between the observed values on y and predicted values (regression line) A large slope doesn"t necessarily imply a stronger relationship between. One variable may have a big slope, and the other a smaller slope it"s not easy to interpret which has a bigger impact. Its value is intrinsically linked to the units of measurement of the variables involved. It is impossible to look at the slope of exercise and of smoking, and determine which has the larger impact (they are different units of measurement) How do we get around this: standardized slope, there is a standardized slope associated with each independent variable that ranges from -1 through +1.