24 September 2021 3 8K Report

Dear all,

I collected human perception data on a Likert scale (1-5) from a total of 35 participants (5 participants with each belonging to 7 stakeholder groups). I asked participants about their perception of animal welfare using statements (scale items) for each construct (main variable). I had 6 main constructs with varying 4-9 items for those constructs. I am testing if their perceptions vary across the different stakeholder groups.

  • While measuring reliability, the Cronbach alpha was negative for most constructs except one. Those other constructs had a mix of negative scale items. I transformed the negative scales to positive, and the value for Cronbach alpha became positive. However, it was not the acceptable value (0.7 or above). How can I adjust this value to make it acceptable? Would you suggest Revelle’s beta or McDonald’s omega instead? If so, how do I analyse these other reliability measures in SPSS?
  • Lastly, to test the differences in perceptions across the different groups, we thought a Discriminant function analysis would be appropriate. Do you have suggestions to analyse a sample size of 35 (with 5 in each group) to test similarities or differences in Likert scale perceptions?
  • Any suggestions would be greatly appreciated.

    More Janice Vaz's questions See All
    Similar questions and discussions