Survey Sampling

A population is an entire set of subjects
A sample is a group of people that represents a population.

There are two types of samples: good samples and bad samples.

Bad Samples

  • Voluntary Response Samples - sample has choice to respond to survey or not.
  • Convenience Samples - sample chosen based on convenience.
  • Biased Samples - sample chosen with ulterior motive.

Good Samples

  • Probability Samples - sample chosen by chance probability.
  • Simple Random Sample - equal chance of selection for sample. Can be done by labeling each member of population (ex. 001 - 100) and using a statistical table to select participants.
  • Systematic Random Samples - sample chosen by, for example, randomly selecting a number (let's say 3), so every 3rd person would be drawn.
  • Stratified Random Sample - sample chosen by dividing the population into groups (ex. by sex) and then doing a simple random sample on the group.
  • Cluster Sample - sample chosen from a segment of the population (ex. by election districts).

Problems with Sampling

  • Undercoverage - not enough sample participants for study
  • No response - response not high enough
  • Response bias - ex. participants can't remember what happened, and cannot therefore answer survey questions accurately.
  • Wording effect - can lead to misunderstanding question, which can also lead to bias.

Lurking Variables

Some survey-takers think that if they intentionally provide false and silly information during a survey, the final results will be skewed. This is untrue. If assessed properly, results that are "off the map" are simply labeled as "lurking variables" and because of their strong influence on x and y, they are simply not included in the final study results. Therefore by answering surveys honestly, your voice truly will be heard and the results of he study you participated in will help shape tomorrow.

opinion Voice your opinion - be a part of a sample!

back to main Back to surveys article listing