Dear Sir, thank you so much. I was struggling to find a solution in Jasp. When i reverse code an item in Jasp for my study model, the cronbach alpha reduces to 0.43. However, if i dont reverse code that particular item, the alpha is 0.63. Both alpha values are low reliability as per literature. I have taken the questionnaire from well established research. I am not sure if i have reverse code the item as it reduces the reliability. Kindly advise.
I would only reverse code the item if that makes sense substantively based on the item wording. For example, if all other items are positively worded and the item in question is negatively worded, I would recode the item. Best, Christian Geiser
@@QuantFish Thank you for the reply. In my research there are 4 variables. for one of the variables, If there are 5 questions in my questionnaire and one of them is negatively worded, I have recoded that one question in Jasp as you suggested in this video. However, it shows a low cronbach's alpha at 0.43 for that variable. However, for other variables i get a high reliability. In your experience, would this qualify for a good ground for hypothesis testing? or i need to rectify the low alpha questionnaire? actually, i have 276 respondents and unfortunately i will not be able to repeat the survey. time and money.
Dear Sir, thank you so much. I was struggling to find a solution in Jasp. When i reverse code an item in Jasp for my study model, the cronbach alpha reduces to 0.43. However, if i dont reverse code that particular item, the alpha is 0.63. Both alpha values are low reliability as per literature. I have taken the questionnaire from well established research. I am not sure if i have reverse code the item as it reduces the reliability. Kindly advise.
I would only reverse code the item if that makes sense substantively based on the item wording. For example, if all other items are positively worded and the item in question is negatively worded, I would recode the item.
Best, Christian Geiser
@@QuantFish Thank you for the reply. In my research there are 4 variables. for one of the variables, If there are 5 questions in my questionnaire and one of them is negatively worded, I have recoded that one question in Jasp as you suggested in this video. However, it shows a low cronbach's alpha at 0.43 for that variable. However, for other variables i get a high reliability. In your experience, would this qualify for a good ground for hypothesis testing? or i need to rectify the low alpha questionnaire?
actually, i have 276 respondents and unfortunately i will not be able to repeat the survey. time and money.