Are you planning on conducting a public opinion poll in the near future? Maybe you want to find out who in your neighborhood is against the new parking lot expansion. Or maybe you want to find out who everyone is going to vote for, for class president. Be sure to follow the following rules of polling, and, hopefully, you're poll will obtain reliable results.
1. The persons interviewed must be a random sample of the entire population to be polled. This means that any person in the population has an equal chance of being interviewed. Do NOT allow people to volunteer as volunteers' views are often different from non-volunteers.
2. The questions must be comprehensible. This means that one has to ask people questions of which they have some basis of knowledge and, therefore, some basis for forming an opinion. For example, an average citizen probably does not know about the specific foreign policy the US has with Austrailia yet if you ask them about a national issue they probably know a little more about it.
3. The questions must be asked fairly. Clear wording and the absence of "loaded" and "emotional" words is essential. The questions should also not be worded as to suggest a "right" answer.
4. The answer categories offered to a person must be carefully considered. If you provide two answer vs. three answer choices, you will achieve significantly different results.
5. Not every difference in answers is a significant difference. This is due to sampling error. You could poll two different groups of people with the same questions, etc. and obtain completely different results. Generally, the bigger the sample, though, the less the sampling error will be.
Three common errors that occur when polling are: sampling error, "measurement error," and "specification error." Sampling error is the least frequent of the three. "Measurement error" usually involves poorly worded questions. For example, if one was to use fair as possible answer choice, many people would question what exactly that word means as it means something different to everyone. "Specification error" usually means there is a bad motive behind the question. You may, therefore, get the "right answer to the wrong question," according to Gary Orren, associate professor at Harvard's John F. Kennedy School of Government.
Surveys are now an integral part of political life, and we are constantly bombarded with new poll data. When viewing such data, look at the following aspects of the poll to determine its validity:
1. Who is the author of the poll?2. Make sure the precise question to the poll is printed in full.
3. Look for how the sample was selected.
4. Look for a sample of at least 600 interviews.
5. Look for a sample where the voters are screened so that the results reflect teh attitudes of voters who are likely to vote, particularly in the primaries.
6. See when the poll was conducted
By keeping these six aspects of a poll in mind, you will be better able to assess whether the poll is substantiative.