What should be sample size in a study on community attitude towards mentally ill to be conducted in a town with population of one hundred thousand?
It depends on the value of error that you can tolerate and on the confidence level of your interval. If the attitude can be either neutral or negative, then you can use the following formula: n = (z_{1-\alpha/2})^2 \frac{\hat{p}(1 - \hat{p})}{(err)^2} (insert this formula to LaTex). Here err is the value of error that you can tolerate, i.e. the half-width of your confidence interval, z_{1-\alpha/2} is the (1-\alpha/2) - quantile of N(0,1), \alpha is your confidence level and \hat{p} is the ratio of (for example) negative responses and all responses. For instance, if you put confidence level of 0.95, the value of error of 0.01 and your \hat{p} is equal to 0.4, then you need approximately 9200 responces.
Depends on how much detail you are asking for and the accuracy you expect from your results.
You have a survey with one question: "What is your opinion of the mentally ill?"
Or you have a survey with 20 questions including gender, education, income, presence of mentally ill person(s) in immediate or extended family and so forth. Then another 20 questions that are more personal from what pets are at home to whether you have ever been married or various personality traits.
Are you planning a bunch of t-tests, or a factor analysis?
The answer to such questions will affect the answer. Generally the more variables in your survey the larger the sample size.
Beware having too much confidence in sample size calculators and formulae. They presume that your estimate of error is accurate, and accuracy is dependent on sample size and methodology in collecting the sample.
Since this is is a prevalence study, you can have a fair idea of how much sample you need using this citable, easy-to-use formula http://www.kck.usm.my/ppsg/statistical_resources/SSCPSversion1001.xls
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude