Open to the Public
Announcement of Presentation on November 20 2019
The Electoral Season always brings up the vote intention from polls (or surveys) that candidates, pundits and journalists use (and even misuse) to assess who is ahead in each political contest. We often hear or read that “candidate X is ahead, with 47% of vote intention, to only 45% for his opponent”. Such statement is statistically incomplete and incorrect. In addition, there are other technical complications.
For example, many people use cell instead of land phones, which are mostly unlisted. Hence, these people are difficult to reach by surveys. Others, use answering machines and caller Id devices as screening mechanisms, to identify callers before deciding to take their call. This curtails access to many potential survey subjects, which can seriously invalidate (or bias) the sample and its results.
Such problems have resulted in some polls grossly overstating or understating the status of the candidates. Recent cases include the US 2016 presidential election and the British referendum on Brexit The current presentation will discuss surveys and their problems in layman’s terms.
Note: Dr. Jorge Luis Romeu is a professional statistician with 45+ years of experience in teaching, research and consulting. He retired Emeritus from SUNY Cortland, worked as Senior Engineer, for IIT Research Institute, in the Reliability Analysis Center, Rome Lab, and was, for 16 years, a Research Professor, at Syracuse University. He currently teaches part time graduate statistics courses at SU.