ST201

General Stat Notes/Wrong Answer Notes

Topics:

Simple Linear Regression

Multiple Regression

Omitted Relevant Variables, Multicollinearity and Dummy Variables (and more?)

Stepwise modeling, Missing Values, Model Validation (and more)

Logistic Regression

Time Series

Intro stuff

Statistical ethics: Don't use misleading diagrams, notation, etc when presenting statistics

Two types of models: Mathematical and Statistical

Mathematical is theoretical and exact

Statistical is empirical and allows uncertainty

Elements of a statistical model: Regression coefficients, Constant, Error Term(residual), Dependent variable

Model has to be uncomplicated, include important factors, describe data well and have realistic assumptions.

Modelling steps: Confounding variables: Variables we don't account for in statistical models that screw up the accuracy of models, related to both the independent and dependent variable
 * Formulate Research Question
 * Choose relevant variables (use influence diagrams)
 * Find suitable Functional Form(linear, quadratic, exponential etc)
 * Analysis(using currently possessed empirical data)
 * Model Checking(against other empirical data, multicollinearity, homoscedasticity, normality indepence),
 * Interpretation (what does it mean)