After correcting for bias, Clinton's national lead in CBS News poll evaporates

CBS News is now pushing the pro-Hillary Clinton polls in rapid fashion.  Late on Wednesday, this media outlet released the results of a national poll claiming to show that Clinton holds a 6% advantage over Donald Trump in the direct head-to-head matchup.

Apparently, the poll "was conducted by telephone June 9-13, 2016 among a random sample of 1,280 adults nationwide, including 1,048 registered voters[.] ... The data have been weighted to reflect U.S. Census figures on demographic variables."

However, based on the polling details, the final weighted sample of 976 registered voters is made up of just 28% Republicans and 35% Democrats.

In the unweighted sample of registered voters, the relative percentage by party was 29% Republicans and 35% Democrats.

So, during the weighting process, the poll increased the Democrat-Republican spread from 6% to 7%.  This relative weighting should have been headed in the other direction.

According to nationwide polling data, Republican Party affiliation has averaged 28% for 2016 so far, and it also averaged 28% since the start of May.  This agrees well with the CBS News poll's composition.

On the other hand, the last time the Democrats were at 35% was early March...of 2013.  Since May of this year, Democratic party affiliation has averaged 29%, just 1% higher than the Republicans, not 7% higher.

As a result, this CBS News poll appears to be biased in favor of the Democrats by the same margin that Clinton purportedly holds over Trump, meaning that if the bias is removed, so is Clinton's lead, and we have a statistical tie.

CBS claims that "[t]his poll release conforms to the Standards of Disclosure of the National Council on Public Polls."  If so, those standards need to be higher, since the polling release should also include all detailed demographic breakdowns for age, sex, race, education, geographic location, income, ideological leanings, etc.  This would allow us to look for even greater bias beyond the obvious R. vs. D. party affiliation divide.

CBS News is now pushing the pro-Hillary Clinton polls in rapid fashion.  Late on Wednesday, this media outlet released the results of a national poll claiming to show that Clinton holds a 6% advantage over Donald Trump in the direct head-to-head matchup.

Apparently, the poll "was conducted by telephone June 9-13, 2016 among a random sample of 1,280 adults nationwide, including 1,048 registered voters[.] ... The data have been weighted to reflect U.S. Census figures on demographic variables."

However, based on the polling details, the final weighted sample of 976 registered voters is made up of just 28% Republicans and 35% Democrats.

In the unweighted sample of registered voters, the relative percentage by party was 29% Republicans and 35% Democrats.

So, during the weighting process, the poll increased the Democrat-Republican spread from 6% to 7%.  This relative weighting should have been headed in the other direction.

According to nationwide polling data, Republican Party affiliation has averaged 28% for 2016 so far, and it also averaged 28% since the start of May.  This agrees well with the CBS News poll's composition.

On the other hand, the last time the Democrats were at 35% was early March...of 2013.  Since May of this year, Democratic party affiliation has averaged 29%, just 1% higher than the Republicans, not 7% higher.

As a result, this CBS News poll appears to be biased in favor of the Democrats by the same margin that Clinton purportedly holds over Trump, meaning that if the bias is removed, so is Clinton's lead, and we have a statistical tie.

CBS claims that "[t]his poll release conforms to the Standards of Disclosure of the National Council on Public Polls."  If so, those standards need to be higher, since the polling release should also include all detailed demographic breakdowns for age, sex, race, education, geographic location, income, ideological leanings, etc.  This would allow us to look for even greater bias beyond the obvious R. vs. D. party affiliation divide.