Why is it wrong to say Poe , Binay, Roxas are tied?

by Dean Bocobo, as originally posted on “WHY IS IT WRONG TO SAY POE BINAY ROXAS ARE TIED?”

READ BusinessWorld-SWS September 2015 Pre-Election Survey: Poe (26%), Binay (24%) and Roxas (20%) lead Presidential race

In the most recent SWS survey of 1200 respondents the statistical margin of error is said to be +/-3%. What this means without getting too technical is that the SCORE each candidate received has a range of 6% centered at that SCORE. So Grace Poe’s score of 26% could actually be as low as 23% or as high as 29%. Binay’s score of 24% could be as low as 21% or as high as 27%. Roxas’ score of 20% could be as low as 17% or as high as 23% (There is a more technical mathematical expression of these ideas but this captures a correct enough layman’s explanation–good enough for govt work as they say)

So what does it mean when statisticians say that there is a STATISTICAL TIE between two candidates. The conventional or textbook explanation would be to say that there is a STATISTICAL TIE between two candidates if the TOP of the RANGE of one candidate with a lower SCORE is greater than or equal to the SCORE of another candidate. So in the case of Poe and Binay there IS a statistical tie or dead heat because the top of Binay’s range is 27% which is greater than Poe’s SCORE.

But can we also say that there is a statistical tie between Roxas and Binay. The top of Roxas range is 23% but Binay’s SCORE is 24%. So unlike Poe and Binay, there is NO STATISTICAL TIE between Binay and Roxas!

POE____________ 23%_______26%______29%


I think that SCIENTIFIC POLLS are important to building strong democratic traditions around elections and to create an ever more intelligent and active electorate. Surveys play such a big role in our elections that the Media both social and commercial and opinion makers … should exert every effort to understand them well and not MISLEAD the public when they do not.

Leave a Reply

Your email address will not be published. Required fields are marked *