Day 12: STA205 - Introduction to Statistical Methods

Last time

Next time

• Announcements:
  • Using IMath exercises to review for the exam....

  • Reminder: stash your rectangles.

  • Your quiz is returned:
    • Hypotheses will generally be written with colons (:), and, in this case, with a mention of the population parameter $p$!

      You're sort of missing the main point if you don't mention the parameter.

      $H_o:\ p=.286$ and $H_a:\ p<.286$.

    • Because the ACS has spent a lot of time and money evidently ("after significant advertising campaigns"), we'd better be able to see a decrease. So let's set a high bar with a low $\alpha$ (perhaps $\alpha=.01$).

    • Two parts (and one important note): we don't usually know that we've committed the error. So some of you were worried that the ACS will be sued, will have their reputation sullied, etc. If they knew that they'd made the error, then they could unmake it! That's the real problem with these errors -- they're silent!
      • Type I
      • The ACS will believe that their campaign has succeeded, when it in fact didn't. They might ramp it up, and make it nationwide, and waste a boatload of money for little or no impact.
      • The citizens of KY will believe that their smoking rates are down, when they're not. They may continue to see advertising that is ineffective.

  • We have a test next time, through confidence intervals (Packet 5): the complete story of $p$!
    • Bring a calculator, and your z-table.
    • You will receive a formula sheet (which you can view on-line here).
    • You will also receive scratch paper, if you need it.

  • Your fifth IMath assignment is due tomorrow at 5:00 p.m.

    A reminder that there's some StatCrunch on it, which we looked at last time.

• Last time:
  • we mostly finished Packet 5, which has two major objectives:
    1. estimate parameter $p$ using statistic $\hat{p}$, with given confidence, and
    2. discover the sample size needed to estimate $p$ with given confidence.

    We still have one sample size calculation to do.

  • I'll start with some questions about anything related to the exam, then we'll finish that packet 5 with one more example of a sample-size calculation.

  • If we have time, we'll begin Packet 6, which has as its major objective determining whether two variables are independent of each other or are related to each other.

    I've mentioned this issue in other packets: for example, we're studying violent behavior in KY adults, but if our sampling frame excludes felons, say, we might be skewing our results (if there is indeed a relationship betweeen violent behavior and felon status -- so we might want to study that!).

• Links:
  • The Standard Normal Table of probabilities.