I want to collect response from teachers of three districts. The population of each district is known as per following;
District. Male Teachers Female Teachers Total Teachers
1 2404 1488 3892
2 1606 707 2313
3 571 384 955
Total 7160
I want to obtain representative sampling proportionate to the population of each district. So which sampling technique would it be or suitable?
That depends on how you draw the sample. You might draw a simple random sample, in which everyone is placed in one group and has the same probability of selection and there are no other constraints, or you might draw a stratified sample, in which each group is sampled separately as a way of ensuring that each group has the desired sample size. Given the number in each district, SRS should work fine, and is simpler in terms of calculating standard errors. Another approach is implicit stratification; sort the teachers by gender and by district, pick a random starting point, and a sampling interval, and you will get an appropriate mix across all groups.
I agree with Mr. Chaney. SRS will help in this very well. If you have a target population already, if you have calculated sample size, to cover it, get a simple lottery method. Put all teachers names on paper or number them and pick the numbers till you get your minimum sample size. You may double it if you want to.
I recommend Stratified random sampling method. First you have to calculate the required sample size. then based on the proportion of male and female teachers from 3 districts, you have to calculate how much proportion of sample should be males and how much proportion of sample should be females.
Selection of number of male and female teachers from each district should also follow the proportion of male and female teachers from each district.
So the sampling will be a stratified random sampling based on the proportion.
For example
If the required sample size is calculated as 100, 64 should be males and 36 should be females (because 64% of total population is males).
Out of 64 males, 53% should be from district 1, 35% from district 2 and 12% from district 3 based on the proportion of male teachers from each district. So please select 34, 22 and 8 male teachers from each of the 3 districts respectively
Out of 36 females, 58% should be from district 1, 27% from district 2 and 15% from district 3 based on the proportion of female teachers from each district. So please select 21, 10 and 5 male teachers from each of the 3 districts respectively.
Now as Bradford and Ashok suggested, you can randomly select the required number of male and female teachers from each district from the list of male and female teachers which are prepared separately - using a random number generator or by lottery method.
Thank you
I agree with previous recomendations and can only add that for optimum representative sampling you should take in considaration also age proportions
Thanks for good suggestions, however the main problem is that I do not have a list containing the name of each respondent, so while adopting SRS, how would I make it random while I dont have each respondent's name. I just know the teachers (male & female) in numbers from each district. Can it be possible that after knowing the exact sample size I may stratify the population and then use convenient sampling?
You face the same issues whether you stratify or not. You should randomize in some way. Ideally, the districts would give you the names or IDs. If not, you may be able to get them to do the sampling for you (e.g., assign a random number in Excel, sort by that number, and pick the top 25 (or whatever number is appropriate for that district).
Your comparing study groups should not be signifinicantly different as regards percents of persons even though at three age groups: 59
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India