Skip to main content
    • Aa
    • Aa

Should More Polls Be Interpreted as Too Close to Call?

  • Walter W. Hill (a1)
  • DOI:
  • Published online: 01 March 2013

During political campaigns the spread between the popularity of the candidates is a common metric capturing the state of the horse-race feature of the campaign. One candidate is said to be ahead of another by an indicated number of percentage points. If the difference is less than the margin of error, the race is considered too close to call. In two-person races, however, the spread corresponds to a much smaller confidence level than is usually reported because the two numbers used to compute the spread are not independent. The size of the confidence interval that is typically reported is incorrect by a factor of two. Therefore, some spreads that are reported as decisive are races too close to call.

Linked references
Hide All

This list contains references from the content that can be linked to their source. For a full set of references and notes please see the PDF or HTML where available.

Adam J. Berinsky 2006. “American Public Opinion in the 1930s and 1940s: The Analysis of Quota-Controlled Sample Survey Data.” Public Opinion Quarterly 70 (4): 499529.

C. J. Wild , and G. A. F. Seber. 1993. “Comparing Two Proportions from the Same Survey.” American Statistician 47 (3): 178–81.

Recommend this journal

Email your librarian or administrator to recommend adding this journal to your organisation's collection.

PS: Political Science & Politics
  • ISSN: 1049-0965
  • EISSN: 1537-5935
  • URL: /core/journals/ps-political-science-and-politics
Please enter your name
Please enter a valid email address
Who would you like to send this to? *