Week 7 (3/10/05 - 3/17/05)
Setting boundaries is a form of love. -- Fred Rogers (1928 - 2003)
UNIT 15
Comments RE: UNIT 14 HW
- Excellent job, continuing improvements
- Why I scrutinize some of your homework (Linda's art history class)
- Most serious error - missing analyses (rare)
- Specific error -- you cannot reliably reliably determine strength
from a scatterplot (unless all points fall directly on the line). See figure:
Two scatter plots of same data. Straight-line pattern in lower plot
appears stronger because of improper scaling and surrounding white space:
- It is the statisticians responsibility to present data in a responsible way
(e.g., yes, this require judgment, and trial and error, too)
- It really is OK not to calculate everything every time by
hand (Did I show the trick for calculating SS from s2?)
- Summarize your analysis
- e.g., Unit 14 summaries should mention linearity, direction,
strength, significance, outliers (if any)
- A summary of "significant" is rarely adequate or
advisable (we talked about this in relation to the other chapters, too)
- Let's continue to work on the more accurate interpretation of p values,
which is smaller p
--> greater confidence in rejecting H0 (better information than using
the artificial cutoff of .05 or .01 (We've discussed this in the past 5
classes, but I don't think its getting through to everyone.)
BREAK Please print this handout formulas1.pdf
Selected Methods and Interpretations
How to Prepare for the Exam