Uploaded image for project: 'OASIS Open Document Format for Office Applications (OpenDocument) TC'
  1. OASIS Open Document Format for Office Applications (OpenDocument) TC
  2. OFFICE-916 Public Comment: ODFF: yet more suggestions
  3. OFFICE-2228

FTEST: does it test the variances, or whether two datasets come from the same population? Is it 2-tailed?

    XMLWordPrintable

    Details

    • Type: Sub-task
    • Status: Closed
    • Priority: Major
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: ODF 1.2
    • Component/s: OpenFormula
    • Labels:
      None
    • Proposal:
      Hide

      FTEST
      Summary: Calculates the probability of an F-test.

      Semantics: Calculates the two-tailed probability that, based on two samples from two normal distributions, these normal distributions have different variances.

      Suppose the first sample has size n1 and sample variance s1^2 and the second sample has size n2 and sample variance s2^2. If s1^2>s^2 FDIST returns twice the area of the right tail of the F-distribution with degrees of freedom n1-1,n2-1 beyond s^1/s^2. If s1^2<s^2 FDIST returns twice the area of the left tail of the F-distribution with degrees of freedom n1-1,n2-1 below s^1/s^2.

      Show
      FTEST Summary: Calculates the probability of an F-test. Semantics: Calculates the two-tailed probability that, based on two samples from two normal distributions, these normal distributions have different variances. Suppose the first sample has size n1 and sample variance s1^2 and the second sample has size n2 and sample variance s2^2. If s1^2>s^2 FDIST returns twice the area of the right tail of the F-distribution with degrees of freedom n1-1,n2-1 beyond s^1/s^2. If s1^2<s^2 FDIST returns twice the area of the left tail of the F-distribution with degrees of freedom n1-1,n2-1 below s^1/s^2.
    • Resolution:
      Hide

      Changes applied:

      FTEST
      Summary: Calculates the probability of an F-test.

      Semantics: Calculates the two-tailed probability that, based on two samples from two normal distributions, these normal distributions have different variances.

      Suppose the first sample has size n1 and sample variance s1^2 and the second sample has size n2 and sample variance s2^2. If s1^2>s^2 FDIST returns twice the area of the right tail of the F-distribution with degrees of freedom n1-1,n2-1 beyond s^1/s^2. If s1^2<s^2 FDIST returns twice the area of the left tail of the F-distribution with degrees of freedom n1-1,n2-1 below s^1/s^2.

      Show
      Changes applied: FTEST Summary: Calculates the probability of an F-test. Semantics: Calculates the two-tailed probability that, based on two samples from two normal distributions, these normal distributions have different variances. Suppose the first sample has size n1 and sample variance s1^2 and the second sample has size n2 and sample variance s2^2. If s1^2>s^2 FDIST returns twice the area of the right tail of the F-distribution with degrees of freedom n1-1,n2-1 beyond s^1/s^2. If s1^2<s^2 FDIST returns twice the area of the left tail of the F-distribution with degrees of freedom n1-1,n2-1 below s^1/s^2.

      Description

      Broken out from OFFICE-916

      FTEST

      • "Summary: Returns the result of an F test (the probability that two datasets
        have come from the same total population)"
        Excel says it tests the probability that the variances are not
        significantly different (which is not the same as coming from the same total
        population).

      [ed:] TODO: who can provide insight?

      • "An F-test returns the one-tailed probability" Excel currently describes
        this as the 2-tailed probability. I believe that is correct (see
        http://www.coventry.ac.uk/ec/~nhunt/pottel.pdf). Old versions of Excel eg 97
        describe it as one tailed. It would be worth someone with real knowledge
        checking this out.

      [ed:] TODO: who can provide insight?

        Attachments

          Activity

            People

            • Assignee:
              aguelzow Andreas Guelzow
              Reporter:
              erack Eike Rathke (Inactive)
            • Watchers:
              0 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: