SOC202 MARCH 27, 2012
Regression
- Correlation oe ass and casuastion are not the same things
- String ass may be used as evidence of causal realitonships but they not prove variables are
causally releated
- Spurious relations but caused by 3rd
- Need to never prove anything
Intro
- The goal of research is to explain why variables vary
- The concept of regression builds on other techniques we have already learned
- Regression is closely allied with correlaiotn
- Rehression analsus provides the aboilty to quant the rel importance
- Regression how one vriable predicts another x predicts y outcome
- Real life not just two variables, regression take in a vairtey of variablkes
Regression model
- Dep and in
- Use model to predict value
- A y intercept when x is 0
- B is slope or regression coefeccticent for x
- Amount y changes for 1 unit of x
- E is called the error term not perfectly explain social reality things cant explain
Regression model
- Sentence lenth vs prior conviction
- Regression is casual relationship
- Mean us center of x and y
- Y increases 3 units for every of x slope is 3
Requirements for regression
- Linear
- Both variables at interval data
- Straight line realions
- Random sample
- Assumption of normality
- The realtion is homoscedastic, for rough cugar above line
Recap