Independent thesis Advanced level (degree of Master (Two Years)), 10 credits / 15 HE credits
The aim of this study is to introduce a new S-shaped hazard function, derive its corre-
sponding likelihood function and then apply it to a real dataset to show advantages of
such method. We dened an inovative S-shaped hazard function based on arctangent
with three additional parameters. Our hazard function opens new possibilities for the
measurement of the lock-in e ect directly using one of its parameters. Its signicance is
further tested by a standard Wald test. We derive a log-likelihood function and apply
it on a data from Volvo Cars Project. According to our results we conclude that the
program was unsuccessful. Surprising shapes of nal hazard functions lead us to closer
examination of possible limitations. We conclude that our results could be a ected
by several aspects such as data quality, computational methods and possible violation
of independency assumption. Nevertheless, this study works as a useful summary of
the newly-dened S-shaped hazard function and highlights its innovative possibility to
estimate lock-in e ect together with its signicance testing.
2014.