Stat 112: Prediction intervals and hypothesis testing

• Administrivia:
  • Office hours moved to 4:10 tomarrow (typically 1:30)
  • JMP Homework due Thrusday.
  • Any questions about the homework? (2.58 data) (2.65 data)

• Basketball example
  • attendence = 15000 + 8 * dollars charged, with error of 3000.
  • Increasing by 1 dollar brings in 8 more customers!
  • It would be nice to see a plot

• Summary of testing proceedure
  • compute t = b₀/SE
  • If |t| > 2 it is significant
  • If not, the true value could be zero
  • Alternate proceedure: look at p-value < .05

• Once we have a good regression, we want to make predictions
  • guess = b₀ + b₁ x
  • Accuracy = +/- 2 * sigma

• Olympic winning times (jmp data)
  • Assumptions: Curvature violated?
  • What would the winning times be in 776BC to 393 AD
  • When will they break the speed of light? (i.e. zero time)

• Cottages example (jmp data)
  • Assumptions: notice influential outlier. Should we delete it or not? Both have their advantages.
  • Consider building a 3000 square foot cottage
  • Would it be profitable?
  • Make prediction

Last modified: Thu Feb 10 08:52:31 2000