I am researching on employee engagement and the questionnaire is set in Likert scale. I have employed judgement sampling technique. I would like to know what kind of statistical tools can be used to analyze the data?
Hey Nandini,
the sample technique does not make a difference for the statistical tool, just to the generalizablity. If you would tell me more about your research question I could tell you more about the right tool. What is the question you want an anser to.
All the best
Timo
Nandini, you may be limited to descriptive statistical tools like measures of centrality, measures of spread, measures of location, measures of relationship, and difference.
Hi, Nandini,
Actually the majority of research studies in social and behavioural sciences rely on non-probabilistic sampling techniques. However, the researchers widely use inferential statistics in their analyses (such as correlation tests, regression tests, structural equation modeling, etc.). Timo Lorenz is right in advising that it is the matter of generalizability. So the non-probabilistic sample does not necessarily mean that you can not run statistical tests.
If you can successfully argue that your sample shares the characteristics with the wider (or intended) population then you can proceed with with the statistical analysis of the data. However, if your sample does not share such characteristics with the whole population (e.g., the sample is imbalanced in terms of age or gender or some other demographic characteristics) then it is not advisable to run statistical tests as the generalizability of the findings would be obviously questionable. Just use descriptive statistics of the findings in such cases.
Hope this helps.
For a discussion on type of sample and inferential analysis try
https://www.researchgate.net/post/Is_a_probability_sample_required_in_inferential_statistical_analysis
First, with regard to the Likert scale items, you might take a look at at thread I have started here on how use those kinds of measures:
https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items?
Second, I agree with Larisa that many people use inferential statistics on non-prabability samples, and this can be quite useful.
Finally, like everyone else, I agree that the main thing you are sacrificing is true generalizability. But I again agree with Larisa that you can make substantive arguments about where and when your findings might apply in other settings. One label for this kind of thinking is "transferability." I personally think of statistical generalizability as a special case of the larger question about when the results from one study have implications for settings outside what was covered in that particular study.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire