In order to establish
an association among a set of variables, regression technique is being used by
statisticians, researchers, and academicians across the world and across
several academic domains over the years. This is the foundation of all kinds of
sophisticated econometric methodologies being used in the field research at the
present time. With a view to finding out the explanation of variance in one
variable by mean of variances in another group of variables, ordinary least
square methodology is used, which is the most common form of linear regression.
Among various statistical softwares, which are available in the market, SPSS is
the oldest and the foremost one in carrying out a linear regression technique
in the most effective manner.
There are few simple
steps to be followed to perform a linear regression in SPSS. Following are
those steps:
·
Import the data into
SPSS editor.
·
Choose linear
regression from ‘Analyze’ section.
·
Choose dependent
variable and the independent variable(s).
·
Choose the method as
‘Enter’. This will demonstrate both of the significant and insignificant
coefficients of the independent variables, along with their probability values,
and the steps, at which they became insignificant.
·
In the ‘Statistics’
section, choose ‘Estimates’ for regression coefficients, choose ‘Model fit’,
and in the residuals section, choose ‘Durbin-Watson’.
·
Continue with the
regression.
·
Once the results have
come out, some indicators need to be checked among all of those. First, if the
Durbin-Watson index is more than the regression coefficient, i.e. value of
R-square, then the model is perfect. Else, the model is spurious and
redesigning of the same is required. Second, while checking coefficients,
un-standardized values are preferred over standardized values, as the former
one is based on the true units of the independent variables, and the latter one
is equalized across all independent variables.