As my study is on Sexual behaviour of Truck drivers , i had assessed attitude towards condom using a Likert Scale based Multidimensionsal Condom Attitude Scale
I am doing a convenience sampling procedure and this MCAS scale has 25 items so how many truck drivers should i enroll so i can check the reliabilty and validity of the scale and further with its subset scoring make association with other factors
Please if possible do give me reference
Thankyou
There have been any number of studies that use simulation techniques to determine the stability of estimates from factor analyses. Although there is no firm statistical test for this, a common recommendation is 10 observations per estimate in your model, although I have seen some results that indicate as low as 5 may be acceptable.
You can find some references to these issues at the following location (where the last article I posted contains a simulation study on sample size):
https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items
Dear Ashish, it might be worth clarifying a few issues: Is this your instrument or did someone else create it? If someone else created it then it should already have been validated. Your title mentions both multidimensions and scale which I found a bit confusing. If your 25 items are from a single scale then there is only one dimension and you will need to use a scale reliability analysis. If they are from different scales then it is multidimensional and you will need to use a factor analysis. If it is your own instrument then these scales should have been designed in advance then validated using a factor analysis.
Convenience sampling is the least random form of sampling so you should bear this is mind when you are estimating sample size. Both factor analysis and scale reliability analysis have minimum sample size recommendations. I would start by looking at Andy Field's book Discovering statistics using SPSS.
Based on G.power calculation for correlation test, you need at least 82 participants. Assuming moderate effect size
Peter Sameuls - This is only one scale which is called UCLA Multidemsional Condom Attittude Scale
It has been validated not only in USA but also in other countries in diverse population and etnicity BUT not yet in India. For that i had translated the 25 items of the scale into HIndi(INDIAs lingua franca)
Now to test the reliability and validity of this scale - which statistical test should be used for testing both reliability and validity. What is the minimum sample size ....and conveience sampling based study(non-probability sampling)
Professor Muayyad Ahmad -Is correlation test for reliabilty or validity - what is the sample size required to test for reliabilty and validity
Ashish, please refer to the article by Beaton et al on cross cultural adaptation of self report measures if you are adapting a questionnaire that has been developed in some other country before testing the psychometric properties. The article is freely available.
If you wish to run a reliability test you can start with a sample of about 30 Participants since its scientifically valid to infer findings from that number and above, then run a reliability test to measure the reliability of your instrument, and may employ an experts Judgment for your validity measure!!
Wish you success~
- Badges
- Science topic
- Similar topics
- Medicine
- Public Health