I have used a 5-point likert scale for my research on 'catalysing spiritual transformation'. The scale has 20 items divided equally across 4 domains (factors). It is a dual response scale and the first response rates the goal while the second response rates the accomplishment. It is a proven scale which has been validated for content and construct across continents. However, since I am using a translation for the first time, the author of the instrument who approved my translation, suggested that it is proper to do a fresh 'construct' validation for the translation. Accordingly I prepared to do CFA and found that my sample size after joining pretest and posttest data was only 174. I would like to join the two sets of responses of each questionnaire and double the sample size to 348, considering the fact that both sets of responses have identical structures though with different foci. I also noticed from the correlations matrices for the two sets of responses and the combination, that the correlation coefficients are significantly better for the combination and are all positive and > 0.5. Will it be scientifically sound to join the data of the two sets of responses and double my sample size as above?
Look forward to your valuable thoughts.
Thankfully
Lawrence F Vincent