Online advertising development and their economic effectiveness

2.2 Second Test 
The second question analyzing the variables that determine the preference to click on either a banner ad or a 
pop-up advertisement. To answer the question, we shall characterize the decision-makers according to their 
socio–economic variables. 
This is a binary decision to be handled with a binary logistic model based on the following utility function: 
(2) 
ij
i
j
ij
Z
U




'
Where
β
j
is the vector of coefficients to be estimated, 
i
Z
is a vector of decision makers 
i
’s socio-economic 
characteristics and stated attitudes, and 
ij

is an error term linked to the fact that we do not observe actual 
utility but only the fact of belonging to a specific sub-group. 
Using the assumption that the error terms are independent, identically distributed and follow the Extreme Value 
Distribution, the probability that an individual 
i 
belong to category 
j
is: 
(3) 


j
z
z
ni
ni
ni
e
e
P
'
'


And the probability lies between 0 (pop-up advertisement preferred) and 1 (banner advertisement preferred). 
This could be written as: 
N
SD
d
t
av
/


---

Australian Journal of Business and Management Research Vol.1 No.6 [121-133] | September-2011
125 
(4)
.....
1
ln
2
2
1
1












Z
z
p
P
LogitP
ni
ni
ni



Logistic regression measures the model estimation fit using -2log of the likelihood value (-2LL), where 0 
corresponds to a perfect fit.