Users can select questions to save to their iPOLL folder by checking the boxes next to the questions on the search results page, then clicking the “Add to My iPOLL Folder” button.  Individual questions can also be added from the question-level screen by clicking the plus icon in the bottom-right corner. Users can view all questions in their folder at any time by clicking “View folder.” The contents of the folder can be downloaded as a .txt or .csv file. Users who are logged in to their personalized accounts can download up to 500 questions at once; others will be limited to 50.

The contents of My iPOLL Folder will be cleared at the end of a user session. Please download any data you intend to save.

iPOLL is the most comprehensive, up-to-date source for US nationwide public opinion available today. A full-text retrieval system, the iPOLL online database is organized at the question-level. The system allows for users to sift through over 600,000 questions archived from national public opinion surveys since 1935. The database is updated daily.

iPOLL includes data survey results from academic, commercial and media survey organizations such as ABC News, the Gallup Organization, Pew Research Centers, Kaiser Family Foundation, and many more. The data come from all the surveys in the Roper Center archive that have US national adult samples or samples of registered voters, women, African Americans, or any subpopulation that constitutes a large segment of the national adult population. iPOLL does not include state samples or foreign samples (see Catalog of Holdings/Search for Datasets), however, surveys of these populations are available from the Roper Center.

To retrieve all questions from an entire survey, regardless of when the questions were published, simply search for the survey organization using the organization pull-down menu and the beginning interview date using the date boxes on the search form. It is important to note that the date criteria used on the search form searches on the FIRST date of the interview period.

The Search for Datasets includes only those surveys for which the Roper Center holds the datasets (microdata or individual level), with brief descriptions of each. These include national, state, and foreign surveys, as well as special samples such as leaders or youth. iPOLL contains the aggregated question-level results from the US national surveys in the catalog, as well as a great deal of additional survey data from news releases and survey reports. Click here to find out what the difference is between a top-line and a dataset.

Like Search for Datasets, iPOLL can help you identify U.S. datasets that are rich in specific subject areas. However, because iPOLL provides the text and topline responses for all the questions, it makes selecting datasets easier and offers an appropriate context in which to review data, by comparing various questions and sources on the same topic and trend lines. Questions in iPOLL are linked to the Catalog study descriptions and often the questionnaires.

Typically internet surveys are not included in iPOLL, though several studies have been included as part of larger collections such as CBS News Polls. The methodology involved in internet polling is still a subject of contentious debate among survey professionals. We do have the results of some Internet surveys at the Roper Center, but they are not at present entered into the iPOLL database on a regular basis. (See Statement About Internet Polls by NCPP Polling Review Board.)

Publications based on Roper Center data collections should acknowledge the Roper Center and the organization(s) that originally sponsored and collected the data. Click here for more information on how to cite our data.

iPOLL search results are sorted by the most recent “Beginning Field Date” therefore on occasion you may see something seemingly out of order.

Member university faculty and students may access Roper Center resources while off campus by logging into the proxy server or Virtual Private Network (VPN) typically available through your university library.  If you have created a personalized account you can access all services directly from anywhere.

See if your university is currently a member of the Roper Center by visiting List of Members. If you are still unsure of your university’s status, email

iPOLL Plus includes not only how the full sample responded to the question, but also some standard demographic groups replied.  The groups likely to be included are men, women, those who identify themselves as Republicans, Democrats or independents, and those of different regions, ages, incomes, or educational backgrounds.

There are a couple of explanations for this:

  1. The iPOLL Plus feature is the latest enhancement to iPOLL, and therefore the 100,000 questions available in this format are from surveys archived with the Center over the last fifteen years. Staff are working back in time to earlier datasets and continuing to process the most current studies in this form.
  2. In order to provide the group data, the Center has to have archived the dataset file, about 35% of the questions in iPOLL are from surveys that are not archived at the Roper Center.

The period covered by iPOLL Plus is approximately 1995-present, and represents about 1/6th of all of the questions in iPOLL, or just over 100,000 items.

When a table has a column of that is empty, this indicates that there were fewer than 100 respondents in that particular group and the Center has suppressed those results. Small samples, including subgroups of less than 100 people, increase the margin of sampling error significantly.  Suppressing these small sample groups prevents the user from inadvertently drawing conclusions that overstate the findings.

Survey organizations attempt to draw random samples—that is where every individual in the population has an equal chance of being a participant. For various reasons samples will over represent or under represent certain groups in the populations. As a result, survey firms will adjust the sample by a process called weighting.  Every survey will have a weighing variable and in order to have a representative sample, any analysis must include this weight.