Dear all, I gave 116 respondents 18 translated sentences and asked them to indicate their levels of acceptance of these translations on a five-point scale. Some translations result from strategies aiming at popularization while others result from strategies aiming at specialization. I find that a respondent who likes accessible translation often gives higher ratings to the first category of translations and lower ratings to the second category. In contrast, a respondent who prefers technicality does the opposite. I am very confident of the data quality because respondents were very cooperative, but the reliability test yielded a very low value of Cronbach’s Alpha, making it difficult for me to continue the survey. Please tell me whether a reliability test is necessary or useful in my case. Thanks a lot!
The respondents are all volunteers. I cannot say they are representative of the target readers of the translated work, but I am sure they were very cooperative because the questionnaire took at least 30 minutes, I offered them no pay and they were free to quit evaluating the sentences.
In light of the characteristics of the survey and the data set under examination, it would appear that a reliability test such as Cronbach's alpha is not entirely appropriate or meaningful in this case.
The data can be described as follows:
Categorical data are defined as data that are not quantitative but are instead grouped into categories. The data set comprises two distinct categories of translations: popularized and specialized. Cronbach's alpha is predicated on the assumption that the items in question are measuring a single latent construct and are unidimensional. However, given that the items in question represent two distinct strategies, they may be inherently multidimensional.
If the objective is to assess individual preferences for different translation strategies rather than to measure a single underlying construct, Cronbach's alpha may not be the most appropriate method. Instead, it may be more appropriate to examine how respondents' ratings differ between the two categories.
Did you make certain that all the items were scored in the same direction? Coefficient alpha assumes that all correlations are positive, so you may need to "reverse" some of the items.
Yes, a reliability test is necessary to ensure that the translations of your survey are consistent and accurate across different respondents. It verifies that the translated items measure the same construct as the original survey.
Souad Djegham Any time you cite a source such as ChatGPT or other AI,, you need to supply that reference -- otherwise, people will think that these are your own ideas.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA