Researchers cannot observe behaviors of interest continuously because it would be too costly

Question

As a result, they create blocks of time to study the behaviors and hope that the behavior in those blocks presents an accurate picture of what happens during non-observation times. Researchers have investigated optimal ways of using sampling blocks in observational research. How accurate are conclusions drawn from small, medium, and large samples of observations and with random versus cluster sampling?
What will be an ideal response?

Answer 1

In general, as the number of blocks of observations increases, the error in describing the behavior decreases. As Moore (1998) has shown, with few observation blocks (i.e., 4), the inaccuracy in describing behavior patterns was almost four times higher than for a large number of observation blocks (i.e., 30). Thus, it is clear that small samples, which many researchers use, generate inaccurate data.
When researchers decide how to sample, if they use random sampling with short observation periods, they can get accurate data on behaviors by observing as little as 10 percent of the total possible time. Cluster sampling, which involves observations in a few, long periods rather than more, shorter periods, leads to inaccurate data unless the total observation time equals half of the total possible observation time or more.