18 Questions You Should Ask About a Poll

I was looking for some of the worst polls ever made , but I came across this rather insightful list of things you should ask yourself before taking a poll seriously. Most of this is things we have already learned or may know, but its fun to compile them all together.

1. Who did the poll?

Was is Gallop or Pickles-for-Pride?

2. Who paid for the poll and why was it done?

It's important to ask this question to avoid special interest polling. It's also healthy to surmise the intentions. Many people poll with a certain outcome they wish to see. In my opinion that can be dangerous. It may lead to push-polling.

3.. How many people took the survey?

Discussed in class, sample size counts!

4. How were those people chosen?

If the poll was conducted by a teacher simply using his class to fill out a poll, and worst yet, offering incentives like extra credit to take the poll, then you might want to reconsider.

5. What area or group?

State? Country? Teachers? Students? Cat-owners? You want to make sure the group/area is relevant. If you're surveying about cat ownership, you might not want to sample dog-owners. Or if you're surveying about desert climates -- you might not want to survey New York.

6.  Are the results based on all the people interviewed?

If you display just the Democrat's opinion I can't see how that would enlighten anyone.

7. Who should have been interviewed and was not? Do response rates matter?

Sometimes, a non-response rate isn't entirely important, but you want to make sure you get as many responses as possible. 

8. When was the poll done?

2012? 2011? or 1938?

Especially during a presidential campaign, even a week's time could dramatically change the results

9.  How were the interviews conducted?

person, telephone, online or by mail all have significant advantages and disadvantages to look out for. A person, or their voice and influence opinion, but they get more participants. Mail and online surveys have to be completed with minimal pressure.

10. Polls on the Internet?

Generally a no-go. The sample size is limited to how many people visit the website. MSNBC polls vs. Fox News? Very different.

11. What is the sampling error?

We were just about to cover this in class weren't we!

12. Who's ahead?

When is a candidate REALLY ahead of another? Is it 1%? , 2 points? 5? Sampling error calculations can help you decide this.

13. What other kinds of factors can skew this specific poll? 

Inadequate training of poll questioners? Time frame?

14. What questions were asked?

Push-polling? Biased? or Objective?

15. What is the question order?

We learned this in class. Question order can sometimes influence whether or not a person answers favorably or unfavorably.

16. What other polls have been done on this topic?

Do they say the same thing? Is this new information? Is there reason to believe these polls were skewed?

17. Exit polling?

Exit polling is important! They are meant to survey only voters! Just as other polls can be misleading, however, watch out!

18. What else needs to be reported when publishing results?

Context. Context context context!

Source: http://www.ncpp.org/node/4/#6