How must a poll be done for it to be effective?

Answer 1

Polling is part science and part art. To be effective a Poll must be statistically valid and must be free of bias.

• The Statistical validity of the poll can be verified by examining the data for the poll. Pollsters usually indicate the number of participants (for example 1024 people in Canada were surveyed) , the make of the poll participants (this was a random poll which had approximately 100 people from each province) , the accuracy of the result (+/- 2.3%), The confidence interval (19 time out of 20); Be wary of polls that do not provide this level of information.
• Bias is subjective or a flaw in the methodology and could take many forms which skew creating sampling errors, for example.
  • Nonresponse bias: skewed results because of those who do not respond
  • Response bias: Skewed because of those who respond to often.
  • Wording of questions: Skewed data because of how a question is phrased.
• • Coverage bias: Skewed data because of the group sampled was not representative.

To be accurate an representative the poll must be statistically sound, and be free from bias.