You are on page 1of 7

6.7 Regression Analysis – A simplified version!

Regression analysis is a statistical procedure appropriate for analysing the association between two or more
variables.

6.7.1 Bivariate Regression

As the name implies, here we analyse the relationship between an independent variable (predictor variable) and a
dependent variable (criterion variable). For example, advertising expenditure (Predictor) and sales (Criterion).

Let’s see how we do this:

1. Click on Analyze  Regression  Linear. You will see the Linear Regression dialog box as follows:

2. Click on LUBS is very professional and click ► to move this variable to the Dependent Variable
box. This is our criterion variable.
3. Click on I had high expectations of LUBS before coming to LUBS and click on ► to move this
variable to the Independents (s) box. This is our predictor variable.
4. Click on OK, and you will see the partial results of the analysis as follows:

Variables Entered/Removedb

Model Variables Entered Variables Removed Method

1 I had high expectations of LUBS before coming to LUBSa . Enter

a. All requested variables entered.

b. Dependent Variable: LUBS is very professional

Model Summary

Std. Error of the


Model R R Square Adjusted R Square Estimate

1 .844a .712 .704 .77458

ANOVAb

Model Sum of Squares df Mean Square F Sig.

1 Regression 56.301 1 56.301 93.840 .000a

Residual 22.799 38 .600

Total 79.100 39

a. Predictors: (Constant), I had high expectations of LUBS before coming to LUBS

b. Dependent Variable: LUBS is very professional

Coefficientsa

Unstandardized Coefficients Standardized Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) .170 .380 .447 .657

I had high expectations of


LUBS before coming to .947 .098 .844 9.687 .000
LUBS

a. Dependent Variable: LUBS is very professional

2
This is how you report your results:

Say our hypothesis was:

H5: There is a strong correlation between previously held expectations of LUBS and students
evaluations of LUBS professionalism

R Square of .712 shows a high correlation between the predictor variable and the criterion
variable. (Maximum we could get is 1. An R Sq value of 0 represents no correlation between the
two variables). Beta value of .844 indicates a positive correlation between the predictor and the
criterion. This means that higher the previous expectation level held, higher the evaluation of
professionalism of LUBS. A significant model emerged (F1, 38 = 93.840, p < . 0005. Results support
the hypothesis as the significance score of .000, shows that the association is significant at p < .
01.

9.2 Multiple Regression Analysis

This is where you wish to measure the degree of association between more than one independent
variable (Predictors), and a dependent variable (Criterion). The analytical model we use in this case is
called Multiple Regression analysis.

Let’s do an example and see.

1. Click on Analyze  Regression  Linear.

2. Click on LUBS is very professional, followed by ► and move it to the Dependent


box.

3. Click on the last two variables in your original Data Editor and move them into
independent (s) box:

3
I had high expectations of LUBS before coming to LUBS, AND

I had good reviews of LUBS before I joined the course

4. Click on OK. It will then produce these results.

Variables Entered/Removedb

Variables
Model Variables Entered Removed Method

1 I had very good reviews of LUBS before I joined the course, I . Enter
had high expectations of LUBS before coming to LUBS

a. All requested variables entered.


b. Dependent Variable: LUBS is very professional

Model Summary

Adjusted R Std. Error of the


Model R R Square Square Estimate

1 .877a
.769 .756 .70346

a. Predictors: (Constant), I had very good reviews of LUBS before I


joined the course, I had high expectations of LUBS before coming to
LUBS

ANOVAb

Model Sum of Squares df Mean Square F Sig.

1 Regression 60.790 2 30.395 61.422 .000a


Residual 18.310 37 .495

Total 79.100 39

a. Predictors: (Constant), I had very good reviews of LUBS before I joined the course, I had high
expectations of LUBS before coming to LUBS
b. Dependent Variable: LUBS is very professional

4
5
Coefficientsa

Standardized
Unstandardized Coefficients Coefficients

Model B Std. Error Beta t Sig.

1 (Constant) -.107 .357 -.299 .767

I had high expectations of .628 .138 .560 4.550 .000


LUBS before coming to
LUBS

I had very good reviews of .397 .132 .371 3.012 .005


LUBS before I joined the
course

a. Dependent Variable: LUBS is very professional

6
This is how you can report the results:
Hn = Student expectations, and reviews they had prior to embarking on their course at LUBS are
positively correlated with Students perception of LUBS professionalism.

In order to test the above hypothesis, a multiple regression analysis using the Enter Method was used to
measure the correlations between the two independent variables (‘I had high expectations of LUBS
before coming to LUBS’ and ‘I had very good reviews of LUBS before I joined the course’ and the
dependent variable; ‘LUBS is very professional’. A significant model emerged (F2, 37 = 61.422, p < .
0005. Adjusted R square = .756. Significant variables emerged from the results are shown below:

Predictor variable Beta p

I had high expectations of LUBS before coming to LUBS .560 p< .0005

I had very good reviews of LUBS before I joined the course .371 p = .005

The results indicate that both independent variables are strong predictors of the criterion variable (LUBS
is very professional) in the model, hence supporting the hypothesis. However, ‘I had high
expectations….’ has the largest beta coefficient (.56). This means that this independent variable is
making the strongest unique contribution to explaining the dependent variable. The Beta value for ‘I had
very good reviews…, was less (.371) indicating that it made less of a contribution to the variation in the
dependent variable. The p values show that both independent variables (predictors) are making a
statistically significant contribution to the variation in the dependent variable (criterion variable) at
p<.005 level and p = .005 level.

(Difficult? Just go through the steps again. You will understand it. Also, whenever you run the
procedure for your own research, just follow the same steps. If you do not lose this booklet, I don’t
think that you will have a problem. Honest!!)

You might also like