Primary Data and Secondary Data: Reliability Test?
Primary Data and Secondary Data: Reliability Test?
Internal consistency
Excellent
Good
Acceptable
Questionable
Poor
Unacceptable
George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and
reference. 11.0 update (4th ed.). Boston: Allyn & Bacon.
How to write the results of your Cronbachs alpha values in your article of thesis?
Table 1.Reliability of measurement
Constructs
Valid number
Number of items
Cronbachs alpha
Turnover intension
200
.778
Trust
200
.913
Justice
200
.920
200
.740
Table 1.Represent the estimated values of Cronbachs coefficient alpha examined the reliability
and internal consistency of the measures. For the present sample, value of Cronbachs alpha vary
from0.74 to 0.92 which indicates that each multi- item construct possesses high reliability
turnover intension (alpha = .778), trust (alpha = .913) justice (alpha = .920) employee job
satisfaction (alpha = .740) the high Cronbachs alpha value for each construct implies that they
are internally consistent. In brief, the higher the Cronbachs value of the construct the higher the
reliability is of measuring.
If you find Cronbachs alpha acceptable then you can go further analysis otherwise not.
Like PCA, CORRELATION, REGRESSION ETC.
Steps for reliability test .analyze------scale----reliability and ok bellow is resulted file of
reliability test.
Step 2. Drag all items of your first variable (means drag all questions of your first variable
into right side box and ok(repeat same procedure for all your variables )
Step 3.resutls of reliability test for first variable. Check reliability of variables and write
results in above table form.
PCA
Now we are going toward PCA (principal component analysis), let suppose we have observe
reliability of all the items and we find reliability values of all items acceptable in this case
can conduct principle component analysis.
We will find two types of values in PCA, KMO and Bartlett's Test AND Eigen values.
Now why we use KMO?
The KMO measure of sampling adequate indicates the suitability of employing factor analysis.
The value of KMO varies between 0 and 1 a value of 0 indicates that there is larger dispersion in
the pattern of correlations hence, application of factor analysis become inappropriate. A value of
1 indicates that the patterns of correlation are relatively compact so the application of factor
analysis becomes appropriate. It is a general rule of thumb that a KMO value of 0.5 is poor and
0.6 is acceptable and value closer to 1 is better and more desire able further more values between
0.5 and 0.7 are mediocre, values between 0.7 and 0.8 are good values and values between 0.8
and 0.9 are great values above 0.9 are superb
Bartletts test sphericity
is conducted to check the significance of the relationship between the items of a construct. If
there is no relationship among the items of a construct then it will be pointless to go ahead with
the factors analysis. Bartletts test assumes a null hypothesis of no correlation. Generally, a pvalue <0.05 confirms the significance of the relationship among variables. In my table the pvalues of Bartletts test in the case of all constructs is less than 0.001 which provides evidence
against the null hypothesis of no correlation. So we can continue with factors analysis.
What values we should write for analysis in our thesis or article?
KMO and Bartlett's Test
Constructs
Number Kaiser-Meyer-Olkin
of items
Bartlett's
Bartlett's
of Test
of
Adequacy.
Sphericity
Sphericity sig
Turnover intension
.690
771.529
.000
Trust
.685
674.495
.000
Justice
.656
634.317
.000
.858
1384.0
.000
Step 2.drag all items of your first variable and note, dont include informational variables
like gender, age etc, and below is detail about what u must click in descriptive, extraction
rotation etc. and after click all required options do ok.
We have need of
KMO and Bartletts
test values, and
significance values
Step 4. You can see all five items of first variables merge and adopt a form of single
variable now go to your spss file and give the original name of your variable to new
variables , from this method convert all the variables in single variables.
Components
Total
% of variance
Cumulative %
Turnover
Comp1
3.220
64.404
64.404
Trust
Comp1
2.734
91.126
91.126
Justice
Comp1
2.964
89.791
89.791
Job satisfaction
Comp1
6.245
56.774
56.774
Note: Constructs means our variables, component means when we run PCA
let suppose when I run PCA for first variable we saw only one variable
extract or develop , its not necessary that all the items produce one
component ,let suppose justice items produce 2 component means two
variable then we will write like this and remaining items same
Construct
Justice
Comp1
Comp2
If I interpret above results, turnover having one component which have 64% information
of all the items of justice variables. And remain same interpretation, and if two component,
then let suppose justice have two component and one component have information of all
items x% while component 2 having information of all items of justice x%.
Correlation
Let suppose we have run PCA and summarized the items of all variables into only four
variable
The correlation analysis has been used to confirm mutual association among the items of each
construct. Then main purpose of correlation is to check the relationship strength means strong
relation or weak or moderate etc.
1. If value comes less than 0..34 we will say weak relationship among variables
2. If value comes 0.34-----.74 moderate relation
3. If value comes more than .74 strong relation
In the following window we can see 23 items are summarize into four variables , now we
will check correlation among variables , now u have no need to examine all the items u
can use only these four variables
Employees
Turnover
intentions
satisfaction
turnover 1
intentions
Trust
.134**
Justice
.129*
.798** 1
Employees
satisfaction
job
.891
**
.524
**
.241*
.
1
job
In the following window write your dependent variable and independent variable
Your dependent
variables and all
independent
variable
To see
autocorrelation in
your data you can
click on statistics and
D.W VALUE
Turnover Intension
1
2
3
4
5
Justice
3
4
5
6
7
8
9
10
11
12
13
14
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
5
5