Hello everyone, I did a survey with a 4 item 3 point-likert scales ( for example; not true, sometimes and true) and a 5 item 4 point-likert scales ( for example; strongly disagree, disagree, agree, strongly agree). I conducted a cronbach's alpha, but it was too low due to a low number of items.
Would there be an alternative to report the reliability analysis that would result in a higher score? Ideally, I would like to create scales from these items.
I also thought of reporting only the correlation coefficient of these scales.
Could you help me?
In the second question, it could be 5 point likert scale, you can add neutral in between
What I would recommend is conducting item-level confirmatory factor analysis for ordinal outcomes based on polychoric correlations. This would be more appropriate given the ordinal scale level of your items and also more informative than calculating Cronbach's alpha for ordinal data. In the factor analysis you could study in more detail whether the items represent unidimensional scales and estimate the item and scale reliability in a more appropriate way. You could also directly relate the factors to external variables, thereby addressing the unreliability issue in the items. Software programs for confirmatory factor analysis and structural equation modeling such as Mplus and lavaan allow you to conduct ordinal item factor analysis based on polychoric correlations.
Based on your explanation, it might be great, if you run factor analysis before employing reliability. Further you should adjust your scales before employing reliability analysis on software platforms e. g., SPSS, Smart-PLS. However, you could adjust the scales, following the higher-lower or lower to higher. For example, let's say, not true-1 , sometimes-2 and true-3, then you should check the frequency distribution on excel. Let's say, Variable-Apple has four items, where, one respondent answers like this- item-A 1; item-B 3, item-C 1 and item-D 1. This is not healthy because the highest and lowest mixing. If most of the respondents answer like the aforementioned example then reliability score is expected to be poor. All the best!
Thank you so much for your answers! Amitab Bhattacharjee and Meghna Jain
Cronbach's Alpha in SPSS Statistics - procedure, output and interpretation of the output using a relevant example | Laerd Statistics.
https://statistics.laerd.com/spss-tutorials/cronbachs-alpha-using-spss-statistics.php
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology
- Fixatives