ANALYSIS: ACCURACY OF MAJOR POLLING INSTITUTIONS
Final poll predictions for the 1996 presidential campaign as a guide to accuracy.

Poll	Sample Size	Clinton	Dole	+ / -	Margin of Error	Margin Exceeded?	Error %
CBS/ New York Times	1519	53%	35%	18%	2.4%	YES	+ 9.5%
Pew Research	1211	49%	36%	13%	2.6%	YES	+ 4.5%
ABC/ Washington Post	703	51%	39%	12%	3.6%	Close Call	+ 3.5%
Harris	1339	51%	39%	12%	2.6%	YES	+ 3.5%
NBC/ Wall St. Journal	1020	49%	37%	12%	2.9%	YES	+ 3.5%
USA Today/ CNN/Gallup	1200	52%	41%	11%	2.8%	Close Call	+ 2.5%
Reuters/Zogby	1200	44%	37%	7%	2.6%	NO	- 1.5%
Hotline/ Battleground	1000	45%	36%	9%	2.8%	NO	+ 0.5%
Actual Results	96,211,883	49.3%	40.7%	8.5%

ANALYSIS: Two of the most quoted polls today were the least accurate in the 1996 Election. Four major polls exceeded their advertised margin of error, while two others, had a close call. Only two polls had results which fell well beneath their advertised 'margin of error'. It is also interesting to note that the number of people polled is not necessarily an indication of accuracy. Apparently, method is more important than number of calls made.

The most accurate polls: Zogby was closest without going over,
while the Battleground poll was closest overall.

It is also important to note that many of these polls sharply reduced their numbers in the days just prior to the election... indicating they may have been way off for the entire year.

[Front Page]