Logistic Regression: A Potentially Useful Tool for Researchers

The potential usefulness of logistic regression is illustrated using data from 112 harvested timber tracts. The binary (0 or 1) dependent variable is indicative of either a profit or loss on a particular logged tract. Explanatory variables describing tract characteristics as well as attributes of the logging firm were used in developing a model. A classification model was developed on a randomly selected subset of the 112 tracts. A second subset of the data was used to cross-validate the development model. The development model contained five explanatory variables (four continuous and one categorical variable) and had a classification accuracy of 78.7 percent, while the validation sample had a classification accuracy of 70.3 percent. By comparison, random classification based on group size (i.e., profit or loss) would yield a 58.5 percent classification rate.

You must be logged in to download any documents. Please login (login accounts are free) or learn how to Become a Member