Class Notes (1,100,000)
CA (650,000)
McMaster (50,000)
SOCIOL (2,000)
Lecture

# SOCIOL 2P06 Lecture Notes - Confounding, Randomized Experiment, Regression Analysis

Department
Sociology
Course Code
SOCIOL 2P06
Professor
Mark Mckerrow

This preview shows page 1. to view the full 4 pages of the document.
September 16, 2011
1. Classroom Observation
can be part of randomized experiment
some topics
pedagogical approaches
student interaction
teacher-student interaction
usually done in elementary school
Useful
richness of direct observation
self reports are suspect
Causation and Correlation
if we think that one thing (x) causes another (y) we think that they will vary
together
eg. if we thinks that caloric intake (x) increases body mass (y) we would
expect that higher caloric intakes is associated with higher body mass
Manipulating X
if we manipulate the cause, the outcome will change
we often can’t manipulate x
so if we see x or y vary together without our having influenced x
eg How much does education increase wage?
cannot manipulate: examine existing relationship
Correlation does not Imply Causation
If X causes Y, then we think that X and Y will (almost) always vary together
We are looking for relationships between X and Y to estimate causal effects
But, we don’t think that when two things vary together, one is always
causing/influencing the other
Even when it is, it is often not the only influence
wage doubles for a bachelor’s degree