Sampling Process
The technique used to do the sampling is called: “Probability-Proportional-to-Size Two-stage Area Cluster Sampling.” For a general population sample: the first stage is to sample blocks and the second stage is to sample households. For a student sample: if you were sampling on campus you would first select dormitories and then dorm rooms; if you were sampling off- campus you would first sample apartment complexes and then apartments.
The procedure for the general population is described below and can be modified for a student sample. Additional information on doing a student sample is given at the end of this discussion.
Step 1: Use the random number table (on my Web site) to choose two random numbers for each person. That is, if the research team has seven members, then fourteen numbers need to be selected. You will need to select five digit numbers. The random number table is posted on my Web site. You may start at any place in the table. However, you must be consistent in your selection. Do not skip around the table in selecting numbers.
Step 2: After you have the random numbers selected, go to “Census 2000 Block Statistics” located on my Web site. Use the random numbers to select your blocks. You will record both a Census Tract number and a Block number. For instance, if the random number is “14001” then the block that is selected is Block number 1010 located in Census Tract 5.04.
Step 3: Once you have the Block numbers and their corresponding Census Tracts, you have to find their location. Go the “Census 2000 Reference Maps” link on my Web site.
First, click on Illinois. Next, click on McLeanCounty, and then click on Bloomington/Normal. Keep on zooming in until you can see the Census Tract numbers. Once you find the census tract, you will be able to locate the block by zooming in further.
Step 4: You will use these locations to do your sampling. At each location you will distribute and collect the following number of questionnaires. Take the number of people in your research team and multiply it by two. If you have seven people then you would have fourteen locations. Divide this number into 240. 240 / (2 x number of team members). For example: 240 / (2 x 7) = 240/14 = 17. Thus each person will need to distribute and collect 17 questionnaires at each location. Each member will go to two locations. [See additional note after Step 5 below.]
Step 5: At each location a “systematic random sample” will be done. We do not want to do a census. If the block is too small, combine this block with a contiguous block or two. Make an estimate of the number of households at this location. For example, let’s say we estimate that there are 50 households. Then take this number and divide it by the number in the previous step. For example, 50/17 = 3. This is called the “sampling interval.” This means that every third household will need to be selected. To get started you need to select a random number from 1 to 3. Let’s say that the random number selected is 2. Then I would go to households 2, 5, 8, 11, 14, 17, etc. At each location you go through this process again making sure you end up with the same number of questionnaires at each location. In counting, youshould always start at the same corner (e.g., south east) and count in the same direction (e.g., clockwise) at every location. If a household is not at home or one refuses to participate, go to the next one. However, resume your count where you left off. In our example, if you need to go to household 9 because household 8 is not at home, still go to household 11 when you resume your count.
Additional note on number of locations and surveys (HH/apartments) needed at each location
If we assume a sample of 240, the following chart tells you how many surveys you need to collect at each location. I will allow you to do 14-18 locations for your project.
# of locations sampled# of households or apartments needed
14 17
1516
1615
1714
1813
What if you have a student sample?
If you are using a student sample you need to use the list of “Bloomington/Normal Apartments” link instead of the “Census Block Statistics” listing on the Web site. You will need to choose four digit random numbers. This will identify the general location where you will sample.
If you are doing a general student population sample you will need to sample on-campus and off-campus. As a rule of thumb take 40 percent of your sample on campus and 60 percent of your sample off-campus. For example, if you were to go to 16 locations, then 6 locations should be on campus and 10 should be done off-campus. I will allow you to include one Greek house in your sample (fraternity or sorority).
For the on-campus part of your sample (after the dorms are selected), ideally, it would be best to go on the dorm floors from dorm room to dorm roomto draw the systematic random sample. However, if it is not possible to go to the dorm floors because of security precautions, you may ask to set up a table in the lobby of the dormitory to pass out questionnaires. Be sure you collect them on the spot. If you have to use this procedure, take every 5th person that walks by to try to make your sample as random as possible.
Caution: Do not make up your own sampling procedure!!
As this stage of the research project, it is necessary to do a probability sample. Above I have explained in detail how to do a PPS Two-Stage Area Cluster Sample. I have also explained in class how to do this procedure. You are expected to follow this procedure in drawing your sample. If you do a non-probability sample then your findings will only be exploratory and not descriptive. You will not be able to generalize your findings to the population of interest. In addition, there will be a significant deduction of points from your final report score. Feel free to ask if you are confused about drawing your sample properly.