Exam 16: Logistic and Time Series Regression

The same principles that apply to multiple regression, such as full model specification, also apply to time series regression.

Logistic regression fits a U-shaped curve to the data.

Logistic regression can be used to predict the likelihood of an event occurring.

Logistic regression is the same as multiple regression with dummy variables.

Logistic regression deals with situations in which the dependent variable is dichotomous.

Classification tables should have 50% or fewer correctly predicted values.

With time series data, the assumption of random distribution of error terms is usually violated.

The odds ratio compares the probability of something occurring, as compared to it not occurring.

The Hosmer and Lemeshow test compares the observed and predicted values and should be statistically significant.

It is better to correct for autocorrelation by taking first differences than by examining relationships in levels form.

Values of the Durbin-Watson statistic around 2 indicate autocorrelation.

The dichotomous nature of the dependent variable violates an assumption of multiple regression.

The Nagelkerke R² is analogous to R² in multiple regression.

Autocorrelation is a problem, but serial correlation is not.

A step impact variable is similar to an increasing impact variable.

Lagged variables are variables whose effect becomes manifest at some future time.

Wald χ² is the test statistic used in logistic regression for testing the statistical significance of logistic regression coefficients.

Policies and program are sometimes evaluated by including dummy variables in the model.

Autocorrelation is detected by examining the variance inflation factor.

The Durbin-Watson statistic has a range of values for which test statistics are inconclusive.