## Sampling -- Sample of Convenience

A "Sample of Convenience" is a sample drawn from a portion of the population that happens to be conveniently accessed by (i.e., is conveniently available to) the experimenter. It is often the case that some portion of the population is more available than is some other portion of the population.

One example of this would arise if we were faced with obtaining samples of some manufactured products that were already boxed, wrapped, and stacked onto pallets. Using specific numbers, we might have 24 pallets, each containing 8 layers of boxes where each layer is 6 boxes long and 10 boxes wide. Our population is the 24x8*6*10 items. That is 11,520 items. We want to get a sample of 60 items. For a "Simple Random Sample" each item would have to have an essentially equal chance of being selected. However, in our situation boxes that are on the top of each pallet are much more convenient than are boxes in lower layers. A "Sample of Convenience" might involve selecting random items from the top layers of the pallets. There are 1440 (that is 24 pallets times the number of boxes in a layer, 6*10) boxes from which we will draw our sample. Clearly, using this methodology the 1440 top boxes have "essentially" the same likelihood of being selected, but the other boxes, all 10,080 of them, have no chance of being in the sample.

As another example, if we want a sample of 60 students from the 11,413 credit students in the fall term at the community college, we might just ask ten of our good friends among the faculty to help us achieve this by
1. giving us a list of the students in their classes
2. creating a merged list of the different students
3. using a random number table or generator to choose 60 students from that merged list
4. having our friends get whatever information we need from the 60 selected students.
Such a sampling methodology would be quite convenient for us. It is true that we have spread the selection out over many students and we have randomly selected the sample from that group. However, any student who is not in one of the classes taught by our friendly fellow instructors has no chance of being in the sample. We need to recognize that even though we think we have a broad spread of students because our "friends" represent a broad spectrum of teachers (we have friends in lots of different departments), because our methodology eliminates some students from the possibility of being in the final sample, we have biased that sample.