While we are excited about the chart and data pages debuting today on Pollster.com, please keep in mind that the site remains under development. We are planning constant upgrades to the Polls section, and we will update this post to reflect changes and additions. For our launch, we are only posting data for the most competitive races. Within the next few days we will be adding poll data, where available, for other statewide races.
To find a chart and data for a particular race: Pull down the menu on the "Find-a-Chart" box in the upper right corner of the page. Choose a state. Then choose a race, Senate or Governor.
Each chart page will feature a table at the top displaying the candidate names and party affiliations. The first column of data will show the average percentages favoring each candidate from the last 5 public polls released for that race.** To the immediate right, the table will also display the margin separating the top two candidates. The last column will display an average of the last 10 public polls in each race.
The chart that appears below includes three elements which users can click on or off using the controls that appear just below the chart:
- Poll results -- dots representing the percentages for each candidate
- Polling trend -- a line for each candidate showing the trend in the last-5-poll average over time
- Confidence intervals -- two faint lines showing the upper and lower range of the reported margin of error for each poll result.
The color of each chart element corresponds to the party affiliation of the candidate. Democrats are in blue, Republicans in red and independents, where they appear, in grey.
At Pollster.com, we intend to report the results of every public political poll that claims to project the views of the electorate, including those conducted using non-probability Internet panel samples. While the tools are not yet ready, we hope to provide readers with the ability to compare and contrast the results from newer survey methodologies.
For now, we have included the "margins of error" reported for non-probability Internet surveys, such as those conducted by Zogby International. The practice of reporting a margin of error for such surveys has been officially condemned as misleading by the American Association for Public Opinion Research (Interests declared: Mark Blumenthal serves on AAPOR Executive Council). I have written about this subject previously and will have more to say about it in the coming months.
A table appears just below each chart displaying results from all of the polls plotted in the chart. A click on any of the column headings will sort the table accordingly. Click on "pollster" for example, and it will sort on the pollster name. "Method" in the far right column indicates the methodology of the poll:
- "Phone" indicates polls conducted using conventional telephone methodology involving a live interviewer
- "IVR" indicates polls conducted using an automated (or Interactive Voice Response) methodology rather than a live interviewer
- "Internet" indicates polls conducted over the internet, usually involving non-probability sample methods
- "Mail" indicates a probability sample conducted by U.S. mail
For our launch, we have included a feature that allows users to remove individual polls from the chart. Click on any row of data in the table and its background color will change to grey, indicating that the chart above no longer plots its data. Users can continue to click to remove as many polls from the chart as desired. Clicking the "reset" link just above the "method" column in the top right corner of the table will restore all data to the chart.
Please note that clicking to remove data from the chart will not recalculate the averages in the header bar at the top of the page.
** Also note that the 5-poll average reported at the top of the page may differ from the averages we provide to Slate. The Slate/Pollster averages exclude results from Internet surveys conducted using non-probability samples.
Finally, if you are having any problems with the chart or would like to recommend features for future upgrades, please email us as firstname.lastname@example.org