This preview shows page 1. Sign up to view the full content.
Unformatted text preview: he standardized factor score of fum i on variable j . This
transformation provides interaction terms that have a lower bound of zero
and an upper bound of infinity.
Multinomial logistic regressionwas then used to obtain the maximum
models were evaluated : (1) using “no involvement” asthe base case from
which deviations are interpreted (Model I (2) excluding the “ no involvement” option and using bbexpolting” t e b case (Model I and (3) excluding
as h ase
“no involvement” and “exporting” options and using joint venture as the
base case (Model III). All three of these models fit the data very well. A
total of 62.5%, 68.0% and 7 0.3% of the observations are classified correctly
in the three models respectively, compared 34.076, 8.0% and 56.6% that
would have been expected to chance. In addition, two logit models,
including interaction terms and the other excluding these terms, were estimated to determine the overall relevanceof the interaction terms. Inclusion
of the six interaction terms helped the log likelihood ratio to de...
View Full Document
This document was uploaded on 03/22/2014.
- Summer '14