Researchers can choose correlation analysis to examine the relationship between variables. The selection of correlation analysis techniques depends on the scale of measurement used for the data. In statistics, the data measurement scale of a variable consists of nominal, ordinal, interval, and ratio scales.
Variables measured using the Likert scale are essentially measured on an ordinal scale. So, what is the appropriate analysis for variables measured using the Likert scale approach?
Variables measured with the Likert scale are essentially similar to data measured on an ordinal scale. Therefore, variables measured using the Likert scale fall into the category of non-parametric variables.
Variables measured using the Likert scale cannot directly obtain numerical values. Hence, a method is needed to analyze these non-parametric variables. Examples of such variables include motivation, competence, performance, etc.
When measuring a variable like motivation, direct numerical values cannot be obtained. To measure variables (e.g., motivation), statements representing the variable being measured can be created. The number of statement items is not limited and should be considered representative of the behavioral variable.
Scoring Technique for Variables with Likert Scale
In Likert scale response items, the choices can be scored on a scale of 1-5, with the answer options and scores detailed as follows:
- Strongly Agree (Score 5)
- Agree (Score 4)
- Neither Agree nor Disagree (Score 3)
- Disagree (Score 2)
- Strongly Disagree (Score 1)
Referring to the example above, a behavioral variable is created with 12 statement items. If respondents answer all the statement items, the scores are calculated as follows:
- Minimum Score = Lowest Score x Number of Statement Items
- Minimum Score = Score 1 x 12 statement items
- Minimum Score = 12
- Maximum Score = Highest Score x Number of Statement Items
- Maximum Score = Score 5 x 12 statement items
- Maximum Score = 60
Based on the calculations above, we can conclude that for the motivation variable, the lowest score is 12, and the highest score is 60. The total score from each statement item represents the score of the behavioral variable.
Testing the Validity and Reliability of Likert Scale Statements
Validity and reliability testing are essential to ensure that the questionnaire/instrument created is valid and reliable. For validity and reliability testing, researchers can choose respondents outside the main sample. The data obtained is then evaluated based on the output of validity and reliability tests.
If the statement items of the variable are valid and reliable, data collection can proceed from the main sample. If the results of the statement items are not valid, researchers need to make corrections to the items until all items are valid and reliable.
How to Analyze Correlation for Variables Measured with the Likert Scale
As I mentioned in the previous paragraph, correlation analysis for variables measured on an ordinal scale, which is a non-parametric variable, can consider Spearman rank correlation analysis.
Based on the output of Spearman rank correlation analysis, researchers can determine whether the relationship between variables is significant or not. Additionally, we can obtain information about the strength and direction of the relationship between competence and performance.
Spearman rank correlation analysis can be conducted using statistical software or calculated manually. In this instance, I will provide an example of how to perform Spearman rank correlation analysis using SPSS.
The first step before entering data is to set up the “variable view” window in SPSS. In this step, the columns that need to be filled are Name, Label, and Measure. Other columns can follow SPSS’s default settings.
In the Name column, enter the variable names. Then, in the Measure column, select “Ordinal” because both variables being correlated are assumed to be measured on an ordinal scale.
Next, you can input each data point one by one or copy-paste the data from Excel if you have already entered it there.
To start the analysis in SPSS, click Analyze -> Correlate -> Bivariate. It will open the “Bivariate Correlations” window. Next, move the variables to be correlated from the left box to the right box.
Under the correlation coefficient options, select Spearman. Then, for the test of significance, choose Two-Tailed. You can activate the flag significant correlations option, so if the correlation is significant, there will be an asterisk symbol in the SPSS output.
How to Interpret Spearman Rank Correlation Analysis Output
As an example, the attached output represents the correlation analysis between competence and performance:
Based on the above output, statistical hypotheses can be tested. In addition to hypothesis testing, the correlation coefficient values can reveal the strength and direction of the relationship.
Aligned with the research objective to understand the relationship between competence and performance, the statistical hypotheses created consist of null and alternative hypotheses.
For instance, the created hypotheses are as follows:
H0: Employee competence does not have a significant relationship with employee performance.
Ha: Employee competence has a significant relationship with employee performance.
From the SPSS output, it is observed that the sig. value is .000, indicating that the p-value is < 0.05. Therefore, the null hypothesis is rejected in favor of the alternative hypothesis. Thus, it can be concluded that employee competence has a significant relationship with employee performance.
Furthermore, the correlation coefficient of 0.813 indicates a strong relationship. The larger the correlation coefficient value (approaching one), the stronger the relationship between the variables. Hence, it can also be inferred that employee competence and performance have a strong relationship.
Moreover, the positive sign of the correlation coefficient indicates a positive correlation between competence and performance. It can be analogized as improving employee competence will enhance employee performance.
Well, this concludes the article I can write for now. I hope it is beneficial and provides new insights for those in need. Stay tuned for educational updates from Kanda Data next week.
