Samples and Surveys… Some Simple Random Thoughts….

     A sample is simply a smaller group meant to be representative of a larger population, the group of people or items that have a specific characteristic that is being studied. However, the process of selecting an appropriate sample can be tricky. There are a number of different methods that are used and it is important to examine what method was used to choose a sample along with the size and composition of the sample group.

In order to draw valid conclusions from a survey, you need to look at how the sampling was done.  In random sampling, each member of the population has an equal chance of being included in the survey.

 

      The validity of the survey would then depend on whether the sample size was large enough to represent the population but small enough to accurate assess.

In systematic sampling, the population is arranged and then samples are taken at regular intervals. For example, a survey of pets on a particular street, might survey every fifth house rather than randomly selecting five. Random selection depends so much on chance that it can result in less accurate data, especially in smaller samples. This hold true whether  it is simple random sampling or a modified version such as systematic.        

 

Convenience sampling, also known as grab or opportunity sampling, is something many of us have experienced first-hand. This type of sampling involves getting samples that are near at hand. Have you ever been in a mall and someone has approached you to ask if they could ask you a few questions? Or have you ever been called and asked if you were the main grocery shopper in the house? These are examples of convenience sampling where it is much more difficult to validate the sample as truly representing the larger population. So any time we are given survey results, it really is necessary to ask ourselves, “Who were the survey respondents? How big was the sample? Is the sample truly representing the population?”   

When designing an experiment such as the one referenced in the text concerning envelopes, it is also important to make sure that the variables are simple and controlled. So although this study was large, the element of using three different envelopes and make two of them very similar, appearing to contain paper, and one to look like it contained money, there was no control over the variables.  The people who found envelopes and opened them probably returned fewer envelopes because they could see that there was really nothing of importance in them. It is not possible to truly test people’s honesty when only the envelopes with the fake money would provide a temptation to the people who found them.   The obvious pattern that exists in the data is that more wealthy people returned 82% of the envelopes which seemed to contain money.  The middle class returned 66% and the poor returned 56%. I would guess that the poor and middle class people opened the envelopes and perhaps finding that the money was fake decided not to bother returning the envelope. It cannot be assumed that they were merely dishonest. 

Advertisement