Mehmetoglu, Jakobsen. Applied Statistics Using Stata. by Mehmet Mehmetoglu & Tor Georg Jakobsen

Author:Mehmet Mehmetoglu & Tor Georg Jakobsen
Language: eng
Format: epub
Published: 2016-11-02T13:40:16.413771+00:00

In addition to these assumptions, influential cases also lead to problems for logit regression. The same is relevant for cases with unusual combinations of X-variable values. In large-N samples, this is usually not a major concern. Potentially influential cases can be tested for in logistic regression using the Pregibon (1981) influence statistic. This measures the standardized change in estimated parameters that results from deleting the observation along with the other observations that share the same covariate pattern. A large value of indicates that the given pattern exerts substantial influence (values over 1 are considered large).

Discrimination is also a problem associated with logistic regression. It refers to our ability to predict the dependent variable. It appears when, for a given X-value, we get a perfect or an almost perfect prediction of the value of Y. This can occur when we have a zero cell count (complete separation), that is, when all or close to all with a given X-value have the same value on Y. If we were investigating the effect of political ideology on the likelihood on being in favour of revolution, we might be faced with the data shown in Table 8.2.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.