Validity of research sample

Reliability & validity in research
This subject seems too complex to deal with adequately here. I don’t know of a formula as such. An important issue is how representative your chosen sample is. What are the variables? These might include age, gender, geographical location, occupation, social group, income, ethnic origin et cetera. If your chosen sample is highly representative of the population (10,000) then that improves the validity of the data and means that you can generalise the results to the whole population. However, if your sample is not representative then your data will have poor validity and generalisation to the population will not be possible.

For example, if your chosen sample is 75% female and 25% male, but the total population is 25% female and 75% male, and gender is a significant variable, your sample isn’t representative.

If you intend collecting data using postal questionnaires, a response rate of 50-60% returned questionnaires would be regarded as excellent. It is often much lower than this but there are a variety of techniques that can be used to maximise response rates. But it means that if you want 1,000 completed questionnaires then you will need to send out at least 2,000, and that assumes that you have taken steps to maximise the response. Otherwise, 4,000 questionnaires may only result in 1,000 being returned, and possibly even less.

3 Responses

TrainingZone says:

February 18, 2003 at 9:49 am

reliability and validity
In anserwing your question, it depends on what research method you decide to use. Cohen, Manion and morrison (2000) have two chapters dedicated to reliability and validity. The book is called research methods in education. I hope this helps you.
Mark Barton
User deleted says:

February 19, 2003 at 12:12 pm

Reliability & validity in research
This subject seems too complex to deal with adequately here. I don’t know of a formula as such. An important issue is how representative your chosen sample is. What are the variables? These might include age, gender, geographical location, occupation, social group, income, ethnic origin et cetera. If your chosen sample is highly representative of the population (10,000) then that improves the validity of the data and means that you can generalise the results to the whole population. However, if your sample is not representative then your data will have poor validity and generalisation to the population will not be possible.

For example, if your chosen sample is 75% female and 25% male, but the total population is 25% female and 75% male, and gender is a significant variable, your sample isn’t representative.

If you intend collecting data using postal questionnaires, a response rate of 50-60% returned questionnaires would be regarded as excellent. It is often much lower than this but there are a variety of techniques that can be used to maximise response rates. But it means that if you want 1,000 completed questionnaires then you will need to send out at least 2,000, and that assumes that you have taken steps to maximise the response. Otherwise, 4,000 questionnaires may only result in 1,000 being returned, and possibly even less.
User deleted says:

March 11, 2003 at 12:14 am

Have a look at this web page
This webpage has a breakdown of formulae and may be of help

http://www.sportsci.org/resource/stats/ssdetermine.html

or enter “formulae for validity of research sample” into Google and go to the link “New View of Statistics: Sample Size Formulae”

Happy number crunching!