STAT3613

Tutorial 6 Solution

Q1
a)
> summary(logit)

Call:
glm(formula = switch ~ appvtime + branch + odprice + guanxi +
sow, family = "binomial", data = bank)

Deviance Residuals:
Min 1Q Median 3Q Max
-2.418 -0.699 -0.275 0.688 2.348

Coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 3.926 1.370 2.87 0.0042 **
appvtime -1.043 0.340 -3.07 0.0022 **
branch 0.587 0.317 1.85 0.0644 .
odprice -1.314 0.626 -2.10 0.0359 *
guanxi -0.731 0.153 -4.79 0.0000017 ***
sow 0.393 0.235 1.67 0.0943 .

b)
H1. Perceived service quality (assessed by the time required to process applications) is associated with a
business customer’s intention to stay with a service provider.

An increase in the approval time for loan applications of the competing bank (coeff. = -1.043, p <
0.01) is significantly associated with a decrease in the odds of switching. This finding supports H1
and reinforces the positive association between perceived service quality and customer retention.

H2. The higher the number of bank branches, the less likely that a customer will intend to switch to
other service providers.

The number of bank branches of another bank relative to the number of the respondents’ main bank
(coeff. = 0.587, p = 0.0644) is not significantly related to the odds of the SME switching from the
main bank to the other bank. This does not support H2.

H3. For a given level of service quality, the product price differential relative to competitors is
negatively associated with customer loyalty.

An increase in the overdraft price by competing banks (coeff. = -1.314, p < 0.05) results in an
increase of the odds of an SME remaining with the main bank. This finding supports H3.

H4. A customer’s intention to switch from a service provider is negatively associated with their
perception of guanxi with that provider.

1
 The perceived guanxi (coeff. = -0.731, p < 0.001) between the SME’s decision maker and the main
bank’s personnel resulted in a decrease in the odds of in switching intention. This finding supports
H4 and indicates that the perception of having better guanxi with the main bank makes an SME less
likely to switch to another bank.

H5. The higher the extent of the existing SOW, the less likely a customer will intend to switch service
providers.

The number of banks the SMEs currently using (coeff. = 0.393, p = 0.0943) is not significantly
related to the likelihood of switching. This does not support H5.

c)
factor levels Part-worth range Relative.
importance
appvtime 1 week -1.043 2.087 20.57%

2 – 3 weeks -3.13
branch Less than 0 1.173 11.57%

More than 1.173

1.5% -1.971
Guanxi Strongly -0.7315 4.389 43.27%
disagree

Strongly agree -5.12

>5 1.574

d)
> predict(logit,newdata=nd)
1
-2.394
> predict(logit,newdata=nd,type="response")
1
0.08364

2
Q2
a)
> summary(fit)
Coefficients :
Estimate Std. Error t-value Pr(>|t|)
companyABB -0.154130 0.321208 -0.4798 0.6313
companyGeneral Electric 0.090972 0.301823 0.3014 0.7631
companyWestinghouse 0.213562 0.293395 0.7279 0.4667

> cbind(new,predict(fit,newld))
company ID altid choice predict(fit, newld)
1 ABB 999 1 TRUE 0.2045496
2 General Electric 999 2 TRUE 0.2613635
3 Westinghouse 999 3 TRUE 0.2954508
4 McGraw-Edison 999 4 TRUE 0.2386361

b)
> summary(fit)
Coefficients :
Estimate Std. Error t-value Pr(>|t|)
companyABB -0.12379 0.67855 -0.1824 0.8552410
companyGeneral Electric -0.67122 0.71941 -0.9330 0.3508145
companyWestinghouse -0.68723 0.71505 -0.9611 0.3364991
price 2.18058 0.58658 3.7175 0.0002012 ***
energyloss 2.65561 0.67371 3.9418 8.087e-05 ***
maintenance 0.59369 0.43703 1.3585 0.1743132
warranty 1.14070 0.33099 3.4463 0.0005684 ***
spareparts -0.13262 0.21757 -0.6096 0.5421585
installation 0.52002 0.17288 3.0081 0.0026290 **
problem 2.03218 0.54968 3.6971 0.0002181 ***
quality 2.63941 0.68775 3.8378 0.0001242 ***

Price, energy loss, warranty, ease of installation, problem solving and quality are significant at  = 5%.
They are all positively related the choice probability.

c)
> range/sum(range)
price energyloss maintenance warranty spareparts installation
0.18332 0.22326 0.04991 0.09590 0.01115 0.04372
problem quality
0.17085 0.22190

Energy loss > quality > price > problem solving > warranty > maintenance > ease of installation > spare
parts

3
d)
> cbind(new,pred=predict(fit,newdata=newld))
price energyloss maintenance warranty spareparts installation problem
s1 4 5 6 1 3 3 3
s2 2 2 7 2 5 2 6
s3 7 3 4 5 1 1 6
s4 2 4 3 1 3 5 4
quality company ID altid choice pred
s1 7 ABB 999 1 TRUE 1.904829e-01
s2 4 General Electric 999 2 TRUE 2.037904e-07
s3 4 Westinghouse 999 3 TRUE 8.088138e-01
s4 7 McGraw-Edison 999 4 TRUE 7.030916e-04