in my research, I have to distribute a survey to the nurses in particular hospital. and I want to make a rough estimation for the number of surveys that should be returned back for analysis and the justification for that. The number of the nurses is around 300, and I will distribute the survey for all of them. what is the minimum number of the surveys that should I accept to get a meaningful results?
Rania -
Beware of sample size calculators online. They generally assume simple random sampling, yes/no data, the worst case for proportions, no finite population correction factors, and so can even sometimes suggest a sample size requirement bigger than the population size. They are irrelevant for perhaps most surveys.
You are attempting a census but expect substantial nonresponse. That is very different than random sampling, and I doubt you have auxiliary data for model assistance, except perhaps to categorize your data. You want data from each important category, and you still risk substantial bias. The sample size depends on such categories, and variability of your data, and you will still have biased results, but better if you study your results by category.
I suggest you research "response propensity"/"response heterogeneity" groups.
Best wishes - Jim
Hi Rania,
Please use following link to determine the sample size of your study. Just enter the population size, confidence interval and confidence level.
Sample size will vary according to them.
http://www.surveysystem.com/sscalc.htm
Good Luck!
Terans
http://www.surveysystem.com/sscalc.htm
Rania -
Beware of sample size calculators online. They generally assume simple random sampling, yes/no data, the worst case for proportions, no finite population correction factors, and so can even sometimes suggest a sample size requirement bigger than the population size. They are irrelevant for perhaps most surveys.
You are attempting a census but expect substantial nonresponse. That is very different than random sampling, and I doubt you have auxiliary data for model assistance, except perhaps to categorize your data. You want data from each important category, and you still risk substantial bias. The sample size depends on such categories, and variability of your data, and you will still have biased results, but better if you study your results by category.
I suggest you research "response propensity"/"response heterogeneity" groups.
Best wishes - Jim
PS - Within data groups/categories/strata/subpopulations, you can handle missing data by two general methods: imputation or weighting. I suggest you weight here. For groups with relatively high weights and/or relatively high variability within that group, you may want more data, as long as it is good quality data.
You may want to research "weighting adjustments for nonresponse."
To add to James' response:
The important categories would be things like race, years of service, sex, and types of patients. So the nurse that works with Dr. Jones might see more cardiac patients, while the nurse that works with Dr. Brown might see more late stage cancer patients. There may be an effect of location, so nurse Alice works with gunshot victims from the East side of town while nurse Frank works in maternity with patients from the affluent north end. If the nonresponse rate is different in the different groups, then all the numbers are biased. Getting a perfect census will not eliminate that sort of bias, but sampling effects make the problem worse. The important categories may (or may not) be a part of the survey, but they will affect the outcome.
Dear RG colleague,
Determining the sample sizes involve resource and statistical issues. Usually, researchers regard 100 participants as the minimum sample size when the population is large. However, In most studies the sample size is determined effectively by two factors: (1) the nature of data analysis proposed and (2) estimated response rate.
For example, if you plan to use a linear regression a sample size of 50+ 8K is required, where K is the number of predictors. Some researchers believes it is desirable to have at least 10 respondents for each item being tested in a factor analysis, Further, up to 300 responses is not unusual for Likert scale development according to other researchers.
Another method of calculating the required sample size is using the Power and Sample size program (www.power-analysis.com).
Regards,
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size