Scientifi c Sampling for Online Research

To view this page ensure that Adobe Flash Player version 9.0.124 or greater is installed.

Please login or register to make a comment!

Sampling Capabilities The rst step to answering your research questions with greater accuracy is to start with a better sample. At Polimetrix, we have created a radically new approach to generating online samples that provides a high level of accuracy at affordable prices. This innovative approach, called sample matching, enables companies to be as accurate as telephone based RDD sampling methods, while fully leveraging the speed and ef ciency of the Internet.

Our online samples provide greater accuracy than other panel providers by incorporating a scienti cally enhanced and validated approach. We operate PollingPoint.com, a centralized portal for those interested in sharing their views about public policy and other current issues. Since 2004, we have conducted over two million interviews within this channel, creating a sense of involvement and community that translates into increased participation.

Simultaneously, we have developed the 8Target Matrix 9, a compendium of national, state and local voter lists as well as other consumer databases that provide us with a large percentage of the US population against which we can de ne a truly representative sample. Using a proprietary and patent-pending technique called sample matching, we select panelists from our PollingPoint panel, aligning those invited to participate in ... more. less.

our online surveys with the Target Matrix across a number of demo- graphic, geographic and socio-economic linkage points. By modeling response and participation rates, we insure that the sample that is drawn, the panelists that are invited and the interviews that are completed are well balanced, representative and highly accurate. Observe the discussion anonymously through Observer accounts. Clients can enter and view an ongoing discussion room without being viewed or heard. Replay the session as many times as a client would like using our Flash-based playback tool Applications Public Opinion Polls Marketing Research Studies Academic Experimentation Media Assessment & Analysis Proximity Matching " Scienti c Sampling for Online Research © Polimetrix, Inc. All rights reserved. Case Study During the fall of 2005, Polimetrix successfully predicted the election results of 7 out of 7 Propositions within the California Special Election, surpassing all other online providers and closely challenging all but one of the telephone based methodologies. Our results were more stable and more economically obtained than any other. According to Mark Blumenthal of Mystery-Pollster.com, cThe newest and most innovative player in Califor- nia polling this week is a company called Polimetrix. d Polimetrix Validates Innovative Method of Online Sampling in California Special Election; Pre-election Internet Survey Predicted Seven out of Seven Races Correctly Six different polling organizations, including some of the best known names in survey research, polled Cali- fornia voters intensively in the weeks before the California special election on Tuesday November 8. Among these was a new company, Polimetrix, that has developed a radically different approach to collecting opinion data. Polling on ballot propositions is notoriously dif cult, but the Polimetrix approach proved itself to be a good replacement for traditional methods. Polimetrix forecast the outcome of seven out seven propositions correctly. Only one other polling organization (the venerable Field Poll) avoided making a miscall. The Polimetrix forecasts had a high degree of quantitative accuracy (with a root mean square error of just 3.1% and an average absolute error of 3.0%). The largest percentage error of any of the Polimetrix forecasts was 6.3% (on Proposition 79). Each of the other orga- nizations which polled on seven or more propositions had at least one error that was larger in magnitude. The two Polimetrix surveys, one conducted October 21-28 and the other conducted November 2-5, produced consis- tent results without any wild week-to-week swings. cAs the new kid on the block, this was an important test for us, d commented Polimetrix President and CEO Douglas Rivers. cIt showed that the methodology that we have developed really works and provides a viable alternative to telephone surveys that are becoming increasingly dif cult and expensive. d Polimetrix conducts its surveys using its PollingPoint panel and its proprietary sample matching process. Samples are drawn from Polimetrix 9s Matrix database (which contains records on over 15 million California voters) and the closest matching respondents in the PollingPoint panel are selected for interviewing. Respon- dents are matched on a wide variety of demographic, political, and consumer characteristics contained in the Matrix database. All interviewing is done online, without any interviewers or telephone calls. One major advantage of the Polimetrix approach is that the samples contain representative numbers of Democrats and Republicans, young and old, northern and southern Californians, black, white and Hispanic, and nearly any other characteristic. This means that the samples are not only representative, but stable from week-to-week. While all of the polls performed well at the end, even the earlier Polimetrix survey, conducted in late October before the nal weekend of the campaign, provided an accurate indication of how the election would turnout. The results of the Polimetrix and other surveys were posted on various Web sites (such as The National Jour- nal Hotline and Real Clear Politics) as the data were collected. cThere is no way to rig a test like this, d Rivers stated, cbut the results speak for themselves. d © Polimetrix, Inc. All rights reserved. Comparison of Final Polls to Election Outcome Percentage Intending to Vote cYes d (Omitting Undecided Voters). Proposition (Election Result) Field PollKnowledge Networks Los Ange- les Times PPICPolimetrixSurveyUSA Proposition 73 (47.4% Yes) 45.6%44.0%56.7%46.7%44.3%52.0% Proposition 74 (44.9% Yes) 46.8%46.0%48.9%48.9%46.4%45.5% Proposition 75 (46.5% Yes) 44.4%52.5%44.0%50.0%49.5%45.5% Proposition 76 (37.9% Yes) 34.8%34.0%34.1%32.6%41.7%39.8% Proposition 77 (40.5% Yes) 40.7%40.0%37.8%41.9%44.1%42.3% Proposition 78 (41.5% Yes) 44.4%49.0%46.9%37.5% Proposition 79 (38.9% Yes) 44.4%49.0%46.9%37.5% Root Mean Square Error 2.1%3.6%5.1%3.4%3.1%2.9% Average Absolute Error 1.8%3.0%4.5%3.0%3.0%2.6% Incorrect Predictions 011101 Data from RealClearPolitics.com and the California Secretary of State (100% of precincts reporting, 12:28 p.m. November 9, 2005). Entries are the percentage in each survey intending to vote in favor of each proposi- tion, with undecided respondents allocated proportionately. About Polimetrix Polimetrix is a venture-funded startup that develops analytics, infrastructure, and databases for survey and political research. Polimetrix offers a unique solution for opinion measurement, using a combi- nation of web surveys, large-scale databases, and cutting edge statistical techniques. Representative samples of respondents are drawn from the PollingPoint panel and interviewed on the Internet to collect data with accuracy, speed, and ef ciency. The company was founded in 2004 and is based in Palo Alto, California. It currently has a staff of twenty experts in survey research, political science, statistics, and information technology. It is backed by Alloy Ventures, a leading early-stage venture capital investor, and YouGov, a multinational polling and research organization. Find out more at www.polimetrix.com. 364 University Avenue Palo Alto, CA 94301 (650) 462-8000 1010 Vermont Ave NW, Suite 715 Washington, DC 20005 (202) 386-7695

less