Confidence Interval for the Variance of a Normal Distribucion:"Equal Tails" Interval versus Optimal

Jose M. Pavía
Departamento de Economía Aplicada
Universidad de Valencia
Esta dirección de correo electrónico está protegida contra los robots de spam, necesita tener Javascript activado para poder verla

  • Abstract
    When estimating the confidence interval for the variance of a normal distribution, textbooks weigh simplicity above optimality in selecting the solution. This fact, understandable in a pre-computational age, has been translated to almost all the statistical software. This work offers a very simple Microsoft Excel macro to obtain the most popular confidence interval and the two principal optimal solutions proposed in the literature, making easier the use of optimal solutions teaching and practising. A comparative analysis is also performed to assess the value of each one of the intervals. The 1937 Neyman's confindence interval is the option that shows the best properties.
  • Keywords: Shortest interval, Minimum Ratio, Confidence level, Sample size.
  • AMS Subject classifications: 62F25, 62-01.
  • PDF PDF (964.06 KB