The aim of the study is to investigate the effect of leadership on academic staff performance.
The target population is academic staff in public universities.
The proposed steps to adopt stratified random sampling are:
1- a list of all universities will be identified.
2- the population of academic staff of each university will be determined, the total population will be obtained through the summation of the number of academic staff in each university.
3- the desired sample size will be determined based on the given population.
4- the samples will be drawn randomly according to the proportion in each university.
The researcher prefer to not include the faculties in the sampling procedure because it will consume of time and cost.
1- Are the steps listed above correct?
2- What is the best justification to ignore faculties in sampling procedure?
Thank you So much Prof.Dr.Peter A Kindle, for your valuable answer.
Regarding the measurement of the leadership: yes as you said I am going to measure the perception of leadership from the staff perspective and similarly for the staff performance performance.
Best regards
Hi, Mohammed,
I do not see any problems with your proposed methodology, but I do wonder how you are going to measure your key variables - leadership and staff performance? Both seem amenable to some sort of opinion or Likert scale measurement, but then you are not measuring leadership and staff performance, but merely the perception of leadership and staff performance.
It should be no problem to justify your focus on staff rather than faculty because the jobs are quite diverse. Faculty, at my institution, are assessed on several levels that include student evaluations, supervisor annual performance reviews, promotion and tenure committee evaluations at the appropriate times, and even by the growing list of academic productivity metrics (RG factors, Google Scholar citations, etc.). There is nothing comparable to any of this for staff evaluations except, perhaps, the annual supervisory review. I would expect access to annual evaluations to be confidential.
I believe that you will find some substantial problems in achieving satisfactory response rates, and you may need to resort to some sort of Freedom of Information (in the US) to access staff email addresses. Responses to online surveys are generally under 15%, so I question if the entire stratified random sample approach is really worthwhile? The most likely outcome is to spend a lot of time identifying your target population only to have a response rate that is so low that external validity cannot be asserted.
Thank you So much Prof.Dr.Peter A Kindle, for your valuable answer.
Regarding the measurement of the leadership: yes as you said I am going to measure the perception of leadership from the staff perspective and similarly for the staff performance performance.
Best regards
You have a multilevel design - staff within universities; this software will give some help to how many you need at each level for reliable inference
Optimal Design for Longitudinal and Multilevel Research Documentation for the “Optimal Design” Software
and although it is in a very different area , this paper thoughtfully considers stratification in such designs and analyses
Designing multi-level studies:: sampling voters and electoral contexts
Article Designing multi-level studies
https://pdfs.semanticscholar.org/968e/e295121b114db964b384456237e7d7263e0b.pdf
I take it that other commentators are using Faculty in a different way to you - that is as academic staff not as groups of disciplines. You could do as a three level design : individuals within Faculties within universities - and this will important if Faculties are important institutional setups with their own cultures ; see my previous response here
https://www.researchgate.net/post/What_is_the_sufficient_sample_size_in_multilevel_analysis
The Optimal design software takes account of the 'costs' of obtaining data at each level.
The effective sample size is based (mainly ) on absolute numbers and not proportions.
Good question and so the answers. What about the concentration of universities in the different regions of Malaysia, it could be factor to consider when choosing universities. As for the leadership and performance, you can also access already designed 5-level Likert scale questionnaires which are validated from other researchers. You may then adjust these to your culture with permission from the designers.
I agree with the very detailed and cogent view of Prof. Peter Kindle. Best regards
Dickson
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size