Internet Banking Service

Appendix III

Logistic Regression Equation

The interpretation of the results of a logistic regression equation is difficult since the equation estimates the logarithm of the odds of a favorable answer to the intent question. What is really more useful is the probability that a respondent evaluating the statements for six significant variables would actually give a favorable response to that question. In order to do this, first the anti-logarithm of the result is taken. This provides the odds that someone would provide a response of ‘yes’ to a question when converted into a probability. In general, if the odds of event A happening are ‘a:b’, then the probability of A happening is

a/(a+b)

The following table presents some examples of possible responses to the six significant variables and the probabilities are calculated. Response 1, given by a person who has a negative intent. Responses 2 and 3 provide the value for a one-unit increase in Savings and response 4 is an average response.

Note that this suggests that if the banks can establish just a slightly more favorable response, the probability of a positive intent will increase. What is left to be done is to determine whether the positive response to intent question will translate into positive usage of the Internet.

Variable

B

R1

R2

R3

R4

Constant

-5.644

Saving

0.645

2

3

4

3

Complex

-0.459

4

4

4

3

Selfuse

0.425

2

2

2

3

Triable

0.471

2

2

2

3

Techadv

0.66

2

2

2

3

Telebank

0.552

2

2

2

3

ABAC Journal Vol. 22, No.3 (September - December, 2002), pp. 63 - 80