19th October 2010 at 1:46 am #4004roshayu mohamadMember
Does anyone have a good reference on this?
I am conducting a quan study on 5 public univs in Malaysia (out of 20 public unis in total)
Sampling frame: public higher education administrative staff and managers.
population size: 50,000.
Thank you.19th October 2010 at 2:15 am #4008Krishamurthy PrabhakarMember
The following article will help you to a large extent.
Organizational Research: Determining Appropriate Sample Size in Survey Research
Bartlett, Kotrlik and Higgins (2001). If you are planning to use Multivariate Analysis you may have to have minimum sample size of 500 to have a reliable analysis. I do not say bigger the better, what i suggest is if possible try to have a large sample size. Please refer to Multivariate Analysis by Hair et al for further treatment. Some times posting your objectives of research will also help to suggest suitable sampling method.
Dr.K.Prabhakar20th October 2010 at 12:02 am #4007roshayu mohamadMember
Thanks Dr Prabhakar,
I will be using structural equation modelling for my analysis. So currently using kerjecie and morgan table to justify my sample size. (its in Bartlett reference too). I do plan to recruit about 500 participant (according to the table sample size is 380).
By the way, given the big population, I wonder if the choice of 5 universities is enough? What do you think?20th October 2010 at 2:46 am #4006Krishamurthy PrabhakarMember
Dear Roshayu Mohamad, Nice to know that. I wish are all universities are similar or there are any differences from them. For example you may have only technical universities or universities that are only science and humanities. If all of them are comparable, you can use random numbers to select the sample. If they can be clustered, please go ahead and do it. Sample size of 5/20 is really good. As student of research, i wish you supplement your data with qualitative aspects also. You will add lot of richness. Please go to the website of William M.K. Trochim, whose work is completely available on net. Lot of your questions will be answered. Some times when you read Prof.Trochim, many questions disappear. for your analysis i suggest using R. It is created by statisticians for their students and research. It is robust and i take the opportunity to suggest as you are beginning your journey of research.
Wishing you all the best.
Yours brotherly,dr.kp15th December 2010 at 9:11 am #4005Sunny BoseMember
For SEM the practised norm is respondents should be15 – 20 times the number of attribuites and dimensions. Out of which the the first set should be 7 – 10 times to check for the relationship and the remaining for the validation.
Regarding number of universities, if your population is 50,000 then sample should adequately represent the population characterisitics whether you take it from one university or more than one university.
- The forum ‘Default Forum’ is closed to new topics and replies.