12.1 Coding, data entry, editing, data cleaning 12.2 Descriptive analysis
13.1 Data Interpretation
DATA PROCESSING, ANALYSIS AND
INTERPRETATION
Getting data ready for analysis
• Data coding
Coding the variables
Coding the response/items for each variable
Eg. Variable for gender = sex
Coding item 1=male, 2=female
The numerical scale can be coded by using the actual number circled by the respondents
(question 6 to 21)
Random checks should be conducted to ensure
data are coded correctly
Table 12.1
Coding of Serakan Co. Questionnaires
___________________________________________________________________________________________________________
1. Age (years) 2. Education 3. Job level 4. Sex [1] Under 25 [1] High school [1] Manager [1] M
[2] 25-35 [2] Some college [2] Supervisor [2] F [3] 36-45 [3] Bachelor’s degree [3] Clerk 5. Work shif [4] 46-55 [4] Master’s degree [4] Secretary [1] First [5] Over 55 [5] Doctoral degree [5] Technician [2] Second
[6] Other (specify) [6] Other (specify) [3] Third
5a. Employment Status
[1] Part time [2] Full time
___________________________________________________________________________________________________
Here are some questions that ask you to tell us how you experience your work life in general.
Please circle the appropriate number on the scales below.
__________________________________________________________________________________________________
To what extent would you agree with the following statements, on a scale of 1 to 7, 1 denoting very low agreement and 7 denoting very high agreement?
___________________________________________________________________________________________________________
6. The major happiness of my life comes from my job. 1 2 3 4 5 6 7
7. Time at work flies by quickly. 1 2 3 4 5 6 7
8. I live, eat and breathe my job. 1 2 3 4 5 6 7
9. My work is fascinating. 1 2 3 4 5 6 7
10. My work gives me a sense of accomplishment. 1 2 3 4 5 6 7
11. My supervisor praises good work. 1 2 3 4 5 6 7
12. The opportunities for advancement are very good here. 1 2 3 4 5 6 7
13. My coworkers are very stimulating. 1 2 3 4 5 6 7
14. People can live comfortably with their pay in this organization. 1 2 3 4 5 6 7
15. I get a lot of cooperation at the workplace. 1 2 3 4 5 6 7
16. My supervisor is not very capable. 1 2 3 4 5 6 7
17. Most things in life are more important than work. 1 2 3 4 5 6 7
18. Working here is a drag. 1 2 3 4 5 6 7
19. The promotion policies here are very unfair. 1 2 3 4 5 6 7
20. My pay is barely adequate to take care of my expenses. 1 2 3 4 5 6 7
21. My work is not the most important part of my life. 1 2 3 4 5 6 7 __________________________________________________________________________________________________________
Getting data ready for analysis
• Categorization
Responses to negatively worded questions should be reversed and recoded
Can be done easily on the computer using Transform and Recode command
• Entering Data
Care should be taken if data are entered manually
SPSS Data Editor can be used to enter and edit data
Rows represent cases while columns represents a variable
After all data been entered, the data are ready for analysis
Data analysis
• Basic objectives:
Getting a feel for the data
Testing the goodness of data
Testing the hypotheses
• Feel for the data
Checking for the central tendency and the dispersion
If there is less variability, the questions could be not properly worded
Check for similar response for every questions
Remember, if there is no variability in the data, then no variance can be explained
Data analysis
• It is always prudent to obtain:
Frequency distributions for the demographic variables
The mean, standard deviation, range and variance on the other dependent and independent variables
An inter-correlation matrix of the variables, regardless whether hypotheses are related to the these analyses. If the correlation between two variables is high, say over . 75, we should wonder whether they are really two
different concepts or we are measuring the same concepts.
Data analysis
• Testing goodness of data
Reliability
Cronbach’s alpha. The closer Cronbach’s alpha is to 1, the higher the internal consistency reliability
Split-half reliability coefficient
Stability measures include:
• Parallel from reliability
• Test-retest reliability
Validity
Criterion-related validity
Convergent validity
Discriminant validity
•Questionnaire was designed and administered based on disproportionate stratified random sampling
•Usable responses equals 174 (sample size)
•Data were analyzed using SPSS version 11.0
Data Analysis and Interpretation
Data Analysis and Interpretation
Steps in data analysis and interpretation
(1) Checking the reliability of measures : Cronbach’s alpha.
Result for ITL is shown below :
•(2) Obtaining descriptive statistics: Frequency distribution
The objective is to get the profile of the respondents
These profiles can also be supported using graph and other visual
presentations.
Example: See output 12.2, p.312
Steps in data analysis and
interpretation
Output 12.2 Frequencies
From the menus, choose:
Analyze
Descriptive statistics Frequencies…
(Select the relevant variables) Choose needed:
Statistics…
Charts…
Format (for the order in which the results are to be displayed) Frequencies Output
Respondent’s Department
Frequency
Frequency PercentPercent Valid Valid Percent
Percent Cumulative Cumulative Percent Percent Marketing
Marketing Production Production Sales Sales Finance Finance Servicing Servicing Maintenance Maintenance Personnel Personnel Public Relations Public Relations Accounting Accounting Total Total
1313 4949 4444 55 34 34 55 1616
33 55 174174
7.57.5 28.128.1 25.325.3 2.92.9 19.5 19.5 2.92.9 9.29.2 1.71.7 2.92.9 100.0 100.0
7.57.5 28.128.1 25.325.3 2.92.9 19.5 19.5 2.92.9 9.29.2 1.71.7 2.92.9 100.0 100.0
7.57.5 35.635.6 60.960.9 63.863.8 83.3 83.3 86.286.2 95.495.4 97.197.1 100.0 100.0 100.0 100.0
•(3) Descriptive statistics: Measures of central tendencies and dispersion
Examine the means, standard deviations, variance, maximum and
minimum for the interval-scaled independent (5-scale) and dependent (4- scale) variables
The variance for job satisfaction, burnout and job characteristics are not very high.
Sec output 12.3, p 313
Steps in data analysis and
interpretation
Output 12.3
Descriptive Statistics: Central Tendencies and Dispersions From the menus, choose:
Analyze
Descriptive statistics Descriptive…
(Select the variables) Options…
(Choose the relevant statistics needed) Descriptive Output
Descriptive Statistics
NN MinimumMinimum MaximumMaximum MeanMean Std Std Deviation
Deviation VarianceVariance Dist Justice
Dist Justice Burnout Burnout Job Sat Job Sat Job Char Job Char ITLITL
173173 173173 170170 167167 174174
1.001.00 1.001.00 1.611.61 2.312.31 1.001.00
5.005.00 4.334.33 4.284.28 4.694.69 4.004.00
2.379 2.379 2.671 2.671 3.117 3.117 3.474 3.474 2.212 2.212
.756.756 .521.521 .507.507 .518.518 .673.673
.570.570 .271.271 .257.257 .268.268 .453.453
•(4)Inferential statistics: Pearson Correlation Matrix
See Output12.4,p.315
ITL is significantly and negatively correlated to:
-Perceived distributive justice -Job satisfaction
-Job characteristics
ITL was found to be positively correlated to job burnout.
Job satisfaction is negatively correlated to burnout and ITL but positively correlated to perceived equity and characteristics.
Steps in data analysis and
interpretation
Output 12.4
Pearson Correlations Matrix
From the menus, choose:
Analyze Correlate Bivariate…
( Select relevant variables ) Option…
Select:
a. Type of correlation coefficient: select relevant one ( e.g. Pearson, Kendall’s Tau,Spearman )
b. Test of significance – two tailed, one-tailed.
•Pearson correlation is appropriate for interval and ratio-scaled variables
•Spearman Rank or Kendull’s Tau coefficients are appropriate for nominal and ordinal scale variables.
•Lower coefficient indicates that the variables are distinct or different while higher coefficient, say .75, it could mean that the variables we are
measuring might have the same properties or characteristics.
Steps in data analysis and
interpretation
•(5)Hypothesis testing
H1- Use t-test to determine whether there is a significant mean difference between men and women in terms of perceived inequities
H2 & H3-uses ANOVA (analysis of variance) to check whether is there any difference in the means between shifts in term of job
satisfaction
Steps in data analysis and
interpretation
H4-Use chi-square test to test whether there is a
relationship between the shifts (1,2,3) and the status of employment (part-time vs full time)
H5- Use multiple regression analysis to measure the extent, the four independent variables being able to explain the variance in the dependent variable.
Steps in data analysis and
interpretation
•Testing hypothesis 1:
From the output 12.5, the means are 2.43 and 2.34 with standard deviations of .75 and .76 for the women and men respectively in terms of perceived equity (distributive justice). The difference in the means is not significant,
therefore we can reject hypothesis 1.
Steps in data analysis and
interpretation
Output 12.5
t-Test for differences between two groups (independent sample test)
choose:
analyze
compare means
independent-samples t Test…
select a single grouping variables and click define groups to specify the two codes to be compared.
Options…
(specify confidence level required-05,.01,etc.)
T-Test output Group statistics
Independent Samples Test
NN meanmean Std deviationStd deviation Std error meanStd error mean Dist justice
Dist justice treatmenttreatment Male femaleMale female 149149 2525
2.432.43 2.342.34
.75.75 .76.76
.052.052 .154.154
levene’s test for equality of variance
levene’s test for equality of variance T-test for equality of meansT-test for equality of means
95% confidence 95% confidence interval of the mean interval of the mean FF SignificanceSignificance tt dfdf Significance (2-Significance (2-
tailed)
tailed) Mean differenceMean difference Std. error Std. error difference
difference lowerlower upperupper Equal
Equal
varianceassumed
varianceassumed 1.311.31 .352.352 .74.74 171171 .461.461 .03.03 .10.10 .30.30 .91.91 Equal variance not
Equal variance not assumed
assumed .67.67 2929 .506.506 .03.03 .09.09 .29.29 .89.89
Steps in data analysis and interpretation
• Testing hypothesis 2: use of ANOVA
– Since there are more than two groups within the sample and job satisfaction is measured on an interval scale,
ANOVA is most appropriate.
– The results are shown in the output 12.6 p.317
– The degree of freedom for between groups is K-1, where K is the total number of groups equal 2(3-1) while within
groups is N-K, where N is the total number of respondents equal 159 because there are 12 missing responses.
(otherwise it is 162)
– Hypothesis 2 is substantiated because the F value is significant at p<0.04
Output 12.6 ANOVA choose:
Analyze compare means
one-way ANOVA…
(select the dependent variables and one independent factor variables) oneway ANOVA…output
ANOVA
Sums of Sums of squares
squares dfdf Mean Mean square
square FF Sig.Sig.
Job sat
Job sat Between Between groups groups Within Within groups groups total total
1.659 1.659 39.645 39.645 41.304 41.304
22 159159 161161
.831.831 .249.249
3.327
3.327 .038.038
Steps in data analysis and interpretation
• Testing hypothesis 3: use of ANOVA
– The results are shown in output 12.7, p.318
– The degree of freedom of between groups is K-1, where K is the total no. of groups equals 4(5-1) while within groups is N-K, where N is the total number of respondents equal 163
– Hypothesis 3 is not substantiated because the F value is not significant (p=.288)
Output 12.7
ANOVA with ITL as the dependent variables one-way ANOVA output
ANOVA
Sums Sums of of square square
dfdf Mean Mean square
square FF Sig.Sig.
ITLITL between between groups groups within within groups groups total total
2.312 2.312 75.143 75.143 77.455 77.455
44 163163 167167
.578.578 .461.461
1.254
1.254 .288.288
Steps in data analysis and interpretation
• Testing hypothesis 4: use of chi-square test
– Since both variables are nominal, a chi-square(x2) test is most appropriate
– The output can be seen in output 12.8, p.319
– The chi-square value of 2.31 within 2 degree of freedom seems to be not significant
– No relationship between employment status and shifts – Hypothesis 4 cannot be substaintiated
Output 12.8 chi-square test
choose:
analyze
descriptive statistics crosstabs…
(enter variables in the rows and columns boxes) statistics…
select chi-square
crosstabs output
employment status*shift cross-tabulation Employment status
Employment status shiftshift
First
First SecondSecond ThirdThird totaltotal Full time
Full time Part time Part time total total
103103 1616 119 119
2525 88 33 33
1818 44 22 22
146146 2828 174 174
value
value dfdf Asymp. sigAsymp. sig Pearson
Pearson Chi-square Chi-square Likelihood ratio Likelihood ratio Linear-by-linear Linear-by-linear Association Association N of valid cases N of valid cases
2.312 2.312 2.163 2.163 1.103 1.103 174174
2 2 22 11
.314 .314 .339.339 .294.294
Steps in data analysis and interpretation
• Testing hypothesis 5: use of multiple regression analysis.
– The four independent variables were regressed against the dependent variables to measure how much variance in the dependent variable can be explained by the four
independent variables correctively.
– R (.548) is the correlation coefficient of the four
independent variables with the dependent variables after all inter correlations among the four variables are taken into account.
Steps I data analysis and interpretation
• R
2,which is the explained variance is actually the square of the multiple R (.548)
2• The ANOVA table shows the F value of 16.72 is significant at the p =.0001 level
• The results indicate that 30% of the variance (R- square) in the dependent variables has been significantly explained by the four independent
variable, therefore hypothesis 5 can be substantiated
Steps in data analysis and interpretation
• If we look at the coefficient table, column Beta under standardized coefficients, we find that the highest
beta is job satisfaction (-.371) and is significant at the .ooo1 level
• This indicates that this could be the only independent variable that is significant
• Therefore the conclusion is that, if we want to reduce ITL, it would be necessary for us to enhance job
satisfaction of employees
Steps in data analysis and interpretation
Overall interpretations:
Only two out five hypothesis were substantiated.
Job satisfaction is the most significant variable that influence ITL.
ITL does not differ with job level
Job satisfaction is found to be significantly lower for those who work evening shifts, further corrective action should be recommended.