I'm working on dissertation research which examines risk perception about the hot environment among migrant construction workers using 5-points Likert survey questionnaire. At this point, I'm puzzling out the methods to be considered for statistical analysis of the ordinal data. I'm considering, chi-square, t-test. Kindly every single piece of advice would be acknowledged.
Abdulahi A Akanji - The data used in calculating a chi-square statistic must be random, raw, mutually exclusive, drawn from independent variables, and drawn from a large enough sample. The most suitable statistical tests for ordinal data (e.g., Likert scale) are non-parametric tests, such as Mann-Whitney U test (one variable, no assumption on distribution), Wilcoxon signed rank sum test (two variables, normal distribution), Kruskal Wallis test (two or more groups, no assumption on distribution).
Can you say more about the measures you have, and especially what you want to use as independent and dependent variables? It might be possible to combine your Likert-scored items into a scale that you could use as a continuous rather than ordinal variable.
Hi Abdulahi, yes the 5 points Likert scale will be the most appropriate in terms of perceptions. You may use "words or elements" such as: Totally agree, agree, neutral, disagree, or totally disagree, depending on the type of question you want to answer. I utilized SPSS (Statistical Package for Social Sciences) which was very user-friendly. Firstly you will have to check the reliability of your test results by using the Cronbach Alpha. Checking the mean value of the results, for example, if its 4.5 then the perceptions are more towards disagree or totally disagree. You may also utilize cross-tabulations for the different categories to see the percentage (%) of people agreeing on disagreeing on your statement. Good luck
You may use SPSS. This is a powerful statistical analysis package with inbuilt functionalities. However, you need to clean and validate the dataset, removing the outliers or any other weird numbers.
I am sure that the interview is one of the options for your statistical research because we don't know what they feel for us.
Hello Abdulahi, as Dr Manzoor has rightly stated the most suitable test statistical test for Ordinal data are nonparametric, but Ordinal variables with any number of categories can be analysed using parametric test is the data is approximately normally distributed.
Chi squared test is perfectly OK.
Look at this article for more clarification. https://www.preprints.org/manuscript/202001.0291/v1
If it's perception... You can use a scale and rate responses quantitatively... Or make it a qualitative study and analyse responses thematically
I will go with the opinion of Dr Manzoor, however, if the data can be converted to interval scale and you can show that is normal, independent and continuous, you may adopt t-test. (You can use SPSS to do the tests).
Please note that chi-square may also be adopted as suggested by Daniel John, even without proving the continuity, but t-test is more powerful.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques