Kindly use the Naing et al. sample size calculator (http://www.kck.usm.my/ppsg/statistical_resources/SSCPSversion1001.xls). It calculates sample size estimations with both finite and no finite population corrections.
Ref:
L. Naing, T. Winn, and B. N. Rusli, “Practical Issues in Calculating the Sample Size for Prevalence Studies,” Arch. Orofac. Sci., vol. 1, no. Ci, pp. 9–14, 2006.
Kindly use the Naing et al. sample size calculator (http://www.kck.usm.my/ppsg/statistical_resources/SSCPSversion1001.xls). It calculates sample size estimations with both finite and no finite population corrections.
Ref:
L. Naing, T. Winn, and B. N. Rusli, “Practical Issues in Calculating the Sample Size for Prevalence Studies,” Arch. Orofac. Sci., vol. 1, no. Ci, pp. 9–14, 2006.
As a rule of thumb, five responses against each item makes acceptable sample size. Means if your questionnaire has 50 items you should multiply this number with 5 which will make 250; this will be a acceptable sample size. 8 responses against each item makes good sample size, 10 responses gives very good, and 15 responses leads to excellent sample size.
Understanding your population is essential even if you cannot estimates its size. This can be approached by defining the population before the experiment (suggested method) or afterwards. How you sample defines the actual population. If you stand outside a bar and hand out a survey to people who enter that bar, then your population is "those people who go to that bar on the evening that the sample was collected."
Your ideal population is dependent on the kinds of conclusions that you want to make. "I want to know about the people who went to the bar on August 5th. I will now describe them." This is in contrast to "I want to know about people who go to bars and I would like to predict if a new person who becomes of legal age to go to bars would do so." The sampling design appropriate in the first case is wrong for the second case.
You need as many surveys as possible. Five surveys per question will not do much good. You need to think about the types of questions and how the results will be analyzed. If you have two questions, and you would like to understand how the response to question 2 relates to income level (question 1), and question one has four levels (