In my study I want to see whether 3 items in a questionnaire relate to autonomy through a reliability analysis. The following three items are relevant I think:
- Availability of task enrichment (yes/no)
- Use of autonomous task groups (yes/no)
- Standardized and detailed work instructions (yes/no). With regard to this last item, I'm looking at additional information about the extent to which it is applied (minimal, medium, high). This is because I think that if an organization has applied a high amount of standardized and detailed work instructions, this implies that employees experience no or little autonomy within their job. Therefore, the variable was reversed by recoding it into a different variable, in which the answer possibility ‘high’ was regarded as ‘no autonomy’ (0), whereas the values ‘minimal’, ‘middle’ and the answer possibility ‘no’ were labelled as ‘autonomy’ (1).
I was thinking that I was recoding it in the right way, because now the highest value (1) means autonomy, which means the same if the other two items have (1) as an answer.
However, my Cronbach's Alpha is -,144 right now and SPSS says: "The value is negative due to a negative average covariance among items. This violates reliability model assumptions. You may want to check item codings."
If I use the reversed version of the recoded variable, no problems arise with regard to Cronbach's alpha. However, the reversed version means that 1 = no autonomy and 0 = autonomy, which means that the highest option here does not mean the same as the highest options for the other two items.
What can I do to ensure that only option 3 (high) of answer possibility Yes is perceived as 'no autonomy', and the others are perceived as 'autonomy' and that it can be used in the reliability analysis?
Dear Anne-Marie,
In addition to Ajit's advice, please consider:
Please see my study guide for more information.
Hello Anne
1. As mentioned above, it is the average of output of correlation coefficients.
2. The acceptable value should be >= 0.7.
3. As a measure of consistency of your questionnaires, you have a serious problem and needs to be observed.
4. In order to discover the problems of your questionnaires, the measures of Item alpha reliability if item deleted and and split-half reliability need to be developed.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS