Feb20-Team4-P-value follow up

From Computational Statistics Course Wiki
Jump to navigation Jump to search

To Calculate

Experiment 1:

  • Data: #Males and #Females from the pride that is our focus.
  • Test statistic: #Females/#Males
  • Null model: The background information from our knowledge about other prides gives us an idea about how sex-ratios are distributed, and we use this distribution as our null hypothesis.
  • Pdf: The null model is going to have a pdf which is some function with support in [0, 1], with most of the mass centered around some number between 0 and 1, and very thin tails.

Experiment 2:

  • Data: We buy lottery tickets until we win, and then count the number of lottery tickets we purchased.
  • Test statistic: .
  • Null model:
  • Pdf:

Experiment 3:

  • Data: Coordinates from our GPS device which is kept fixed at a certain position.
  • Test statistic: Mean coordinates.
  • Null model: Our null model is that the device is unbiased, and gives location estimates which are normally distributed with mean at the exact position of our device, and a variance specified by the manufacturer.
  • Pdf: We expect the pdf under the null model to be a normal distribution.

Experiment 4:

  • Data: Failure time for light bulbs used in traffic lights.
  • Test statistic: Mean failure time.
  • Null model: Our null model would be a statement of what the mean failure time of the lights should be. This could be based on the manufacturer's specification.
  • Pdf: We expect the pdf under the null model to be exponentially distributed.