The latest MSNBC/WSJ poll has caused some confusion among DUers about its oversampling of African American voters.
I hope this clears it up.
TO: NBC News And The Wall Street Journal
FROM: Hart/McInturff
DATE: March 27, 2008
RE: Sample For March 24-25 Survey
As you know, the sample for the March 24-25 poll on race included an “oversample” of 100 African American voters. There has been some confusion as to exactly how these extra interviews were integrated into the survey; we hope this memorandum will clear up any misconceptions.
The main sample for the survey was a cross section of 700 registered voters nationally. As is the case with all of our usual polls, this sample is statistically representative of voters across the country, accurately reflecting the gender, age, educational, geographical, and racial makeup of the electorate. The column in the topline document labeled “All Voters”, as well as nearly all of the subgroups listed in the survey crosstabs, are among these cross section of 700. Eleven percent (11%) of these interviews -- or 77 interviews -- were with African Americans, which accurately reflects African Americans’ proportion of the electorate. Thus, African Americans are NOT over-represented in our national sample.
In addition to this national cross section, we interviewed an extra 100 African Americans to analyze the opinions of this group with a greater degree of statistical reliability. We combined these 100 only with the 77 African Americans that naturally fell into our national sample, for a total of 177 interviews with African Americans; these extra interviews were not combined with the full national sample of 700. The column in the topline document labeled “African Americans” shows the responses of these 177 respondents, as do the subgroups in the crosstabs for African Americans, African-American men, and African-American women.
The table below shows the margins of error for the three groups whose responses are shown in the topline document:
National cross section of voters: 700 interviews, +/- 3.7%
White voters: 520 interviews, +/- 4.3%
African-American voters: 177 interviews, +/- 7.4%
http://firstread.msnbc.msn.com/archive/2008/03/27/827746.aspxAn earlier explanation was published by Pollster.com based on a different poll altogether:
Pollsters sometimes "oversample" a survey sub-population in order to increase the reliability of the results for that group. More interviews means less potential random sampling error. Before tabulating the data for the full sample, however, they "weight" back the oversample its correct proportion with the larger sample.
I checked with Gary Langer, the director of polling at ABC News, and he provided a few additional details. The ABC Polling Unit started with a nationally representative sample of 1,803 randomly selected adults interviewed between March 29 and April 4. Of these, 660 described themselves as baseball fans (on the survey's first question). Of these, 64 were African-American.
The pollsters wanted a bigger and more reliable sampling of African-Americans. So they continued calling from April 5 to April 22 and interviewed another 476 randomly sampled African Americans, of whom 139 were self-described baseball fans.
Thus (adding everything up), the ESPN/ABC survey interviewed 799 baseball fans, including 203 among African Americans. Before tabulating the data, however, they weighted the combined sample of 2,279 (the original 1,803 plus the oversample of 476 blacks) in a way that reduced the proportion of African-Americans to its correct value as determined by the U.S. Census.**
This practice is not at all unusual. The intent is to generate more statistically reliable results by race, not -- as Brown puts it -- to "generate racially charged results."
http://www.pollster.com/blogs/even_polls_about_baseball.php