hi,
for my quantitative research, i am analysing the pilot survey data. While, my cronbach is looking good, but Tstat, few outer loadings & Pval doesnt look good.
Which are the key parameters i must check [eg cronbach, loading etc], before i roll out the final survey>
i have 2IVs, 2DVs & 2Moderators.
also i would like to check the reciprocity in the relationship between 2DVs.
, thanks in advance
Hi! The pretesting can help to refine the items. Sometimes the ambiguity in the items make the respondents confused. Thanks.
Good- day, i agree with Azamussan Syed
i think you are on the right track for your pilot and remember, there are some tests that wont be able to be performed on pilot data and/or even your pilot results can differ from the actual study. This due to response rate, sample size and even interpretation of the questions. I often advise my students not to be over-critical of pilot data, but use it as a guide. The real results will come from actual data. Even with actual data collection, you may think you will obtain a certain amount of responses, but this is all dependent on the respondents themselves.
goodluck.
Thanks a lot Sachin Suknunan & Azamussan Syed .
I've been dwelling deep into the analyses on my pilot for over a week now.
While cronbach& discriminant validity is good, Tstat, SRMR, NFI, Pvalu after bootstrapping & 17 out of 40 outer loadings are less significant.
I did some changes in the path in PLS which improved them a bit but there are some which are really insignificant.
Do I worry about it or proceed since cronbach & HTMT is green..
Rgds
55.= sample size, 6 variables inclu 2moderators. totally 40 items
cronbach PSS- .778 POS .865 AC= .923 WB=.941 Res*= .768 POP*= .666
SRMR .223 NFI= .498
Path coefficient 0.22 1.042 -0.34 0.76 -0.25 0.038 0.655
R2 AC =.662 OS= .429 WB =.290
Path coefficient 0.22 1.042 -0.34 0.76 -0.25 0.038 0.655
Azamussan Syed thanks.. This is helpful
so will it be prudent to say, i could go ahead with my final survey based on the cronbach& discriminant inputs?
Azamussan Syed thanks a lot for your inputs.. best wishes for these times
Azamussan Syed , i had a question on my moderators, the path coefficient for my moderators is -0.34 & -0.25
could that be an issue later ?
Dear Negi,
You must check reliability and validity of each variable (sub-scale) separately. Cronbach alpha values should be around .8. Conduct PCA in Factor Analysis and examine eigenvalues greater than '1' (or scree plots). If, only one component is extracted (indicating one dimension) then your sub-scale is valid for further statistical analysis on the construct. Else, you have to eliminate one or two items in the scale and again conduct PCA.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey