Lecture 4 Flashcards
Dependent Variable - Non-metric scale
Independent Variable - Non-metric scale
Contingency Analysis/ Logistic Regression
Dependent Variable - Metric scale
Independent Variable - Non-metric scale
Variance Analysis, Regression analysis with dummy variables
Dependent Variable - Non-metric scale
Independent Variable - Metric scale
Logistic Regression
Dependent Variable - Metric scale
Independent Variable - Metric scale
Regression analysis (F-test and t-tests)
Non-metric scale?
0’s and 1’s
One group/sample statistical tests?
Scale type - Nominal/Metric
Two group/samples tests?
Scale type - Ordinal Metric
More than tow groups/samples
Metric
What does the hypothesis test check?
Whether the observed difference occured randomly as a result of sampling error or whether it indicates a difference between the two samples?
Type 1
Type 2
Errors
T1 - False positive
T2- False negative
R^2
Expresses the proportion the explained variance in the dependent variables (Y) that is explained by the regression line
Most important goodness of fit statistic
R Squared properties?
No rules for how high it must be
Doesn’t say anything about the importance of an influencing variable
It offers no information on how well the model performs outside the sample
Influenced by the properties of the sample: Decreases with lower variance of Y and X
Applications and limits of regression analysis?
Can be used for:
Explanation of relationships
Simulation of effects, Predictions and Identification of driving factors
works with classic metric data, 0/1 values, frequencies, etc.
It requires mathematical formulation of the mental model, good data and an appropriate specification of the regression model
Key assumptions of linear regressions
Require atleast two independent variables
There should be a linear relationship between the dep and ind. variable
The error term is normally distributed
No multicollinearity
Homoscedasticity
Sample size - 20 cases per ind. variables
