Saturday, February 21, 2009

Sampling and Surveys

What are appropriate purposes for survey, how are subjects selected, how is data collected and analyzed, and what kinds of generalizations are possible?
Unlike case studies and ethnography, survey is intended to “obtain descriptive information about readily observed or recalled behavior of very large populations” instead of making an in-depth study of a phenomenon to identify certain variables. Survey done with random sampling can be used to “achieve representativeness of large population.” Its purpose is to reduce the cost and efforts in conducting research on large population and still obtain representativeness. The researcher has to determine the large population and the appropriate size of the sample population (subjects). Size of the sample has to be balanced (neither too big nor too small) to achieve both representativeness and not to lose the quality of the data. The best way to achieve representativeness is to use random sampling so that every member of the large population has equal chance to be selected in the sample.
In survey different data collection methods can be used from questionnaire to interviews, but careful attention is required in the construction of survey and questionnaire. They have to be clear and unambiguous. As far as possible, the tested methods in the field of research have to be used and if any new method or instrument is used, that has to be pretested, edited, and reviewed to establish its validity and reliability. The main consideration has to be given in the method’s capability of eliciting high response rate. After data is collected, the major variables of it are determined and tabulated in terms of nominal, interval, or rank order data. The researchers may also use the measures of central tendency or dispersion to analyze the data. The results of sampling survey are largely generalizable to the population (N). However, it is difficult to claim cause-effect relationship as this is a descriptive research. Overall, sampling and survey are important in making the study of large population manageable and representative as well. But the researcher needs to be careful in subject selection, selection of methods/instruments of data collection, and data analysis.
Let’s examine Blokzijl and Naeff’s research. They conclude that the result of the survey suggests to keep the PowerPoint presentation sober. They are careful enough not to make a sweeping generalization on students’ preference about PowerPoint presentation. However, the researchers could have taken a much larger population, (N) incorporating students with varying levels familiarity with technology. Their results do not become generalizable to or representative of any students other than the ones they studied. They do not define the large population and the proportion of sampling size with the large population. The five-scale point may have affected the result of the research as it has a tendency to draw the responses towards the center.

No comments:

Post a Comment