Respected Seniors, I would like to share that I am working on a primary study with 3 major construct. I initially studied them individually ( Construct having approx 30 statements each) using separate sample groups for pilot study. After initial round of pilot study I have deleted almost 10 items per construct before clubbing the questionnaire. I would like you to suggest measures or techniques which I can use to further reduce the parameters and related statements. As of now 60 statements are there. Collecting response on a questionnaire with 60 statements is difficult and may lead to respondent bias.
Is it appropriate if I initially study these constructs separately when it has to be used in a clubbed way.
Hi Anoop,
First of all, consider the importance of content validity. So if your items do not represent the exact variable you want to measure, then there is a serious problem. When you do the analysis of principal component, you will probably get some of the items out of the scale as the work becomes clearer. In addition to this, you can also remove highly related items from each other.
Best wishes.
You would need to collect an initial small.sample to conduct pilot testing and then data reducttion by principle component analysis can be useful
Respected all, I have deleted almost 10 items per construct based on data reduction by principle component analysis only. I missed it in my explanation.
In my case, I had to go through each item one-by-one to determine how it would contribute towards the research questions; plus how it will be relevant for regression analysis and interpretation. I had wished to include as many as possible but was quickly reminded that my research cannot cover everything. So, I had to remain focused and it has taken me six months and am still modifying the questionnaire before even pre-testing. After that, probably more changes before large scale broadcast.
Besides the suggestions above, i would suggest you stream line the focus of your research. review items in your questionnaire in relation to your objectives and research questions. This will inform the relevant ones to retain and those to eliminate. Subject the retained ones to Pre-test. Hope this will be helpful.
Best wishes,
Chris.
Best to do pilot testing with small sample (30). Questionnaire can be long. Then do factor analysis, after which items that do not load onto the respective constructs, can be removed. The questionnaire would become shorter by the time you issue it the second round for the real survey.
There are multiple ways to reduce the questionnaire length.
For the reduction of items you can use various procedures. I recommend you: 1. Consultation of experts, using the Delphi technique. You can make consistency checks between experts. 2. Analysis of the internal consistency of the questionnaire (Cronbach's alpha, indicating which items can be eliminated). 3. Factorial analysis of the items. You can do confirmatory factor analysis (if it has precedents for this same analysis) or exploratory. You can remove items that do not have high factorial scores.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire