Dear Pedro

I believe that both must be considered, that mean that you have some dimensions with high alpha values and others not, low internal consistency might cause non logic results in component analysis later.

good luck

Dear Pedro

To me, as long if the Cronbach alpha value > 0.7 it consider to acceptable in term of internal consistency. However, you may delete any single item if  it is recommended to increase the alpha value. I think the critical issue in PCA at convergent and discriminant analysis (i.e loading factors, eigen values, communality etc).

Hi Pedro, 

I'm writing for a bit of ignorance about using Cronbach's alpha with the PCA, but I would think that the C's alpha you use would depend on what you are interested in assessing. Are your principle components groupings which you are using to make sense of the data (i.e. saying 'everything here has some internal consistency')? In that case, I would expect that you would want to look at C's alpha for each component individually. Please keep in mind that I have no direct familiarity with Cronbach's alpha with PCA - but have used both techniques separately. 

All the best for your analysis

Bec

Estimado Pedro,

Aprovecho este contacto para invitarlo a que participe en las sesiones de Bioestadística, econometría e informática del V Congreso Internacional de Producción Animal que se realizará en el Palacio de Convenciones de La Habana, Cuba del 16 al 20 de noviembre de este año. en este evento se presentarán algunos trabajos relacionados con la estadística multivariada que veo es su línea de trabajo. Por favor, si tiene interés en este evento, puede escribirme al mi correo para enviarle toda la información necesaria. Mi correo es: [email protected]

Espero que se motive y que podamos encontrarnos en Cuba. Saludos cordiales

Dr. C. Raúl Cobo Cuña

Investigador, ICA

Thank you all for the answers,

After increasing the number of components, I see that the total cronbach increases also. So I better use the individual one. 

This Q&A from research gate is the best, dont you think?