?? POLLING STANDARDS A quick checklist for using polls in the paper: 1. Before any poll can be cited in the paper or on our web site, our reporter must have access to the full set of questions asked, the results and the methodology. It is essential to find out: who sponsored the poll, who conducted the fieldwork, how many people were interviewed, how the respondents were chosen, how the interviews were conducted (i.e. by telephone, in-person, by mail etc.), what dates the interviews were conducted, and what the margin of sampling error is. It is best to have the poll vetted by News Surveys. 2. If a poll does not meet Times standards for disclosure, sampling and/or methodology, but it is significantly impacting the news or changing the dynamics of a campaign, it can be cited using language that makes clear to the reader that the poll itself is not reliable and why. The citation should not include any percentages. Here too, it is best to check with News Surveys. This is an example: Mr. Smith was clearly unnerved by recent polls that found him neck and neck with Ms. Jones. Some of those polls do not meet the standards of The New York Times and other news organizations because they relied on automated telephone calls. But they have energized the Jones campaign and brought it new support, not least from the antitax Tea Party movement and other conservative groups. 3. With few exceptions, in order to meet Times standards for publication, polls should be based on probability samples. That means everyone in the population has to have an equal chance or at least a known chance of being selected. 4. If the poll is not a major media poll such as ABC/Washington Post, NBC/Wall Street Journal etc., then it is important to know who paid for the poll and why the poll was conducted. In general, polls sponsored and paid for by political campaigns or special interest groups do not meet Times polling standards. 5. Overnight reactive polls to events like the State of the Union Address generally do not meet Times polling standards. 6. For now, polls conducted over the Internet with the exception of those that use telephone sampling to recruit respondents do not meet the sampling standards of The Times. 7. Polls conducted using automated voices sometimes referred to as interactive voice response polls or "robo-polls" do not meet Times polling standards. 8. The article should include the basic information about the poll. The findings and the margin of sampling error should be rounded to the nearest whole percentage point. "The nationwide telephone poll was conducted June 1-4, with 1,029 adults and has a margin of sampling error of plus or minus three percentage points." Occasionally, when a passing reference is made to a poll, details on methodology may be excluded. It is best to consult someone in the polling department. 9. Give the reader the context of the poll, "the poll was taken three days after the President announced new health care initiatives." 10. Reporting of election polls should follow the guidelines outlined in the Polling Standards Document. 11. New York Times polls are preferred over those done by outside organizations when both are available. 12. It is best to have polls vetted by someone in the News Surveys Department.