Authors   Abstract Two new reliability indices, ordinal coefficient alpha and ordinal coefficient theta, are introduced. A simulation study was conducted in order to compare the new ordinal reliability estimates to each other and to coefficient alpha with Likert data. Results indicate that ordinal coefficients alpha and theta are consistently suitable estimates of the theoretical reliability, regardless of the magnitude of the theoretical reliability, the number of scale points, and the skewness of the scale point distributions. In contrast, coefficient alpha is in general a negatively biased estimate of reliability. The use of ordinal coefficients alpha and theta as alternatives to coefficient alpha when estimating the reliability based on Likert response items are recommended. The choice between the two ordinal coefficients depends on whether one is assuming a factor analysis model (ordinal coefficient alpha) or a principal components analysis model (ordinal coefficient theta).     Recommended Citation

Your data are ordinal so you may benefit from this paper:

Ordinal Versions of Coefficients Alpha and Theta for Likert Rating Scales

Bruno D. Zumbo, University of British ColumbiaFollow

Anne M. Gadermann, University of British Columbia

Cornelia Zeisser, University of British Columbia

Zumbo, Bruno D.; Gadermann, Anne M.; and Zeisser, Cornelia (2007) "Ordinal Versions of Coefficients Alpha and Theta for Likert Rating Scales," Journal of Modern Applied Statistical Methods: Vol. 6: Iss. 1, Article 4.

Available at: http://digitalcommons.wayne.edu/jmasm/vol6/iss1/4

  Authors   Abstract Two new reliability indices, ordinal coefficient alpha and ordinal coefficient theta, are introduced. A simulation study was conducted in order to compare the new ordinal reliability estimates to each other and to coefficient alpha with Likert data. Results indicate that ordinal coefficients alpha and theta are consistently suitable estimates of the theoretical reliability, regardless of the magnitude of the theoretical reliability, the number of scale points, and the skewness of the scale point distributions. In contrast, coefficient alpha is in general a negatively biased estimate of reliability. The use of ordinal coefficients alpha and theta as alternatives to coefficient alpha when estimating the reliability based on Likert response items are recommended. The choice between the two ordinal coefficients depends on whether one is assuming a factor analysis model (ordinal coefficient alpha) or a principal components analysis model (ordinal coefficient theta).     Recommended Citation

Your data are ordinal so you may benefit from this paper:

Ordinal Versions of Coefficients Alpha and Theta for Likert Rating Scales

Bruno D. Zumbo, University of British ColumbiaFollow

Anne M. Gadermann, University of British Columbia

Cornelia Zeisser, University of British Columbia

Zumbo, Bruno D.; Gadermann, Anne M.; and Zeisser, Cornelia (2007) "Ordinal Versions of Coefficients Alpha and Theta for Likert Rating Scales," Journal of Modern Applied Statistical Methods: Vol. 6: Iss. 1, Article 4.

Available at: http://digitalcommons.wayne.edu/jmasm/vol6/iss1/4

Dear Nepoleon, First of all I want you to finalize what type of reliability you are investigating says stability aspect or equivalence aspect or both. The inte-class corelation coefficient can be calculated using Cronbach alpha for test-Retest. Apart from reliability testing i would suggest to include validity test also...

 Test-retest would require for example a 2 week lag.

Thank you so much Prof. Beatrice and Prof Khuman. I appreciate your prompt response. Yes I will look into the article and will also do a validity test. Just wanted to clear out the reliability aspect of the test. 

Thank you Prof. Swanenberg. Will look through it. I appreciate your input>

Dear Dr. Caballero:

I use SPSS, which has a direct function to calculate both inter-trial/item and test-retest reliability estimates. I would also suggest repeated measures-ANOVA between the 10 items for each participant's itemized scores to rule out any systematic error. The other way to do this is using the Bland-Altman plots.

Cronbach's alpha is a measure of internal consistency; therefore, the alpha coefficient that you will obtain will essentially inform you about how well the developed test measures the same construct, within the same testing session, i.e.between trials, or in some cases, within the same day, but not between sessions or days (Lj C., 1951 — http://psych.colorado.edu/~carey/courses/psyc5112/readings/alpha_cronbach.pdf )

An alpha coefficient closer to 1.0 indicates that the developed test is consistent within the testing session and is less erroneous. 

Hope this helps. Good luck!

In my masters thesis I am evaluating nurse’s knowledge of dysphagia. I would like to discover where the main problems (lack of knowledge) are. In my questioner I included assessment of nursing practice, professional training and academic knowledge. At firs I also had 3 point scoring, but was than advised to use alt least 5 points scale for better reliability.

Good luck!

how can i do ordinal coefficient theta in spss or minitab?

thanks in advance

Dear Sedighe,

Check this out:

Ordinal Versions of Coefficients Alpha and Theta for Likert Rating Scales

Bruno D. Zumbo et al. (2007) Journal of Modern Applied Statistical Methods 4, 6,1

http://digitalcommons.wayne.edu/cgi/viewcontent.cgi?article=1121&context=jmasm