Science Investigation Surveys and Controlled Surveys (page 3)

By — John Wiley & Sons, Inc.
Updated on Jan 1, 2011

Survey Variables

Like an experiment, a survey may include an independent and a dependent variable or other related variables. Consider an investigation based on this question: How much is the playing of musical instruments by children related to the playing of musical instruments by their parents? In conducting an experiment, we would have to rear children according to a prepared plan-some with parents who play musical instruments and some with parents who do not. After ten or twelve years we would compare the groups to see what we could learn.

No one, obviously, is going to do such an experiment. The only workable method is to take a survey of many parents and children. As in an experiment, the hypothesis would include a statement of two variables that the investigator suspects might be related: (1) the playing of musical instruments by parents and (2) the playing of musical instruments by their children. The hypothesis might be: "There is a statistically significant relation between the playing of musical instruments by children and the playing of musical instruments by the parents of these children."

As in an experiment, a survey may require that constants be maintained. For instance, in classifying children into two groups, those who play instruments and those who do not, we would not want to allow biases to creep into the findings. We would try to ensure, for example, that all of the children came from families of similar income level, similar educational level, and so forth. Or, to take a different approach, we would try to do a good job of randomizing the families so that all income levels, educational levels, and the like would be fairly represented in both groups of children.

Such a survey would require interviewing the children or their parents face to face, by telephone, by mail, or in some other manner, using a suitable questionnaire.

In the end, we would hope to have enough information on enough children and their parents to make a decision about the hypothesis. Would an affirmative answer ("Yes") to the hypothesis tell us that the parents' playing was a cause, or the main cause, of the children's playing? Probably not. The causes of such playing are complicated. There are many other variables that may be part of the answer. However, as a result of our survey, we may find that the two variables are related.

A Controlled Survey

A survey can also be controlled in much the same way as in a controlled experiment. The people in the survey are called the "sample" of the larger population. For example, if you survey 30 students in a classroom, out of all of the people in the world they are your "sample." In a controlled survey, you need to divide the subjects surveyed into two groups, the survey sample and the control or survey control sample. For example, if you were to investigate the effects of the hours spent watching TV have on siblings (brothers or sisters), you would create two samples, one with siblings (the survey sample) and one without siblings (the control sample). In all other respects the two groups should be as alike as possible. Give each person in the two groups a questionnaire that includes the question, "How much time do you spend watching TV per day?" When you get the questionnaires back, you can analyze them to learn if there is a significant difference between the survey group and the control group.

Survey science differs in this important way from experimental science: In doing a survey, you try not to change anything. However, you may use the results of the survey to change things. You may then need to do another survey to learn what effects, if any, the changes had.

Your undertaking of a controlled survey can be honest and scientific. It can also be an excellent basis for a display in a science fair.

View Full Article
Add your own comment

Washington Virtual Academies

Tuition-free online school for Washington students.