Know-How Exchange : Customer Behavior

If I understand your question correctly, the number of people in your sample per city will depend on the size of the city - it will not be merely 300/7.

The best way to select a representative sample will depend in part on the format and accuracy of your information. For example, if you have telephone numbers, they can be used to select folks.

Let's take this one step at a time. Let's assume your primary concern in sampling is adequate representation across cities, especially since that is your question. First, add the populations of all cities. Divide the number of people in each city by the total number in all cities, and multiply by 300 to give you the target number in each cities.

Next step is to determine your best sampling technique within each city. This is tough to reccomend without looking at your data sources. Click on my profile and give me a call, I'll be happy to take a look at the actual data, and post a suggested technique on the site.

Next you begin the survey process. Do you want to attempt to contact 300, or to actually contact 300? If you have a goal of 300 successful surveys, you might consider what we used to call a "cluster technique" where you create 300 lists of people, and call the first person on each list. If you are successful in completing the survey, you are finished with that list. If you have some form of failure (refused the survey, bad contact ifo, never available) then you replace that person with the second person on the list... and continue until you have your 300 surveys. If you have properly selected your sample and select the "first" and "next" contact from each list at random, you should have a very good sample.