MATH 251, Probability and Statistics I, Fall 2001, Sept. 17, Day 8

HW 7 Comments:  If you get your QQ plot from Analyze>Descriptive Statistics>Explore: Plots: Normality Plots, the "Expected" axis gives you standard normal numbers.  If you use Graphs>QQ, you get raw numbers on the "Expected" axis.

HW Day 8, Monday Sept.17  Read:  2.1 Scatterplots   + handout on scatterplot smoothing by hand with Median Trace (+ handout of fig. 2.8, Nitrogen Oxides vs. CO)
Also SPSS manual pp. 57-61.  This is very compact, and incomplete.  Supplementary notes:
---Simple scatterplot menu p. 58:  Put a variable in the "Label cases by" box. Then in the chart editor click  left hand button on the lower toolbar (little rectangle icon, "Point ID")  and you can click on points and label them.  Chart Options, labels "off" will clean them away.
---To show 2 or more groups (text p. 209) Simple scatterplot menu p. 58.  Put the categorical variable of the groups in the "Set Markers by" box.  You will need to edit the point shape of one of the groups to print it, unless you have a color printer.
---Smoothing is done with the Lowess fit line (bottom picture p. 61)
        In Chart editor:  Chart>Options:Fit line (check Total or Subgroups):Fit options:Lowess.
           The smaller the percent, the narrower the moving "window" used.
---The overplotting of duplicate points is "solved" in SPSS by using "sunflowers" to plot several points in a local area as a flower with one petal per point.  Chart Options; you can have sunflowers-- or labels &subgroups-- but not both at the same time. (You can copy a chart using the Outline, format each one differently.)
---(For 2.2, 2.3)To get correlation or regression numbers for a group: Analyze/Regression /Linear.(p.59) Put your variable for subgroups into "Selection variable"; hit the Rule button. Type in the value for the group you want (you need to know what that is--M or F, or Male or Female?)

Read ahead,  2.2 (Correlation)
Hand in: Scatterplots (ch2.1)
p. 120, 2.3 car exhaust a and b. With it,
   do a median trace on fig. 2.8 (handout) 
Using SPSS: 
2.8 speed/fuel  Save your data file. 
2.10 body mass M//F (use "Set Markers by" sex) 
     Save your data file. 
2.11 botulism
2.17 pecking order You may use medians instead of means 
(getting SPSS to get you the means can be done but it's tricky.  You can eyeball the medians on your printed graph). See notes below.		 Read, discuss 
p. 117, 2.1, 2.2 
 
 
 
 
 
 

 

 Optional 
2.7 erosion,plot by hand if you need the practice.
Notes on Pecking order, 2.17:  As a .DAT file, the data set is given this way: Pecking order is the first
variable, and Weight the second.  (You might expect a third, for pen number.  The data is in order of pen number,
so you could put it in easily if you wanted.)
As an Excel file, it's very different--it looks exactly like the book's layout (Row A is an overall title, not part of data).  You can plot each pen's results separately, on the same axes, using Graph>Scatterplot: Overlay.  Use Help to see how.

Though you are supposed to study "the effect of weight on pecking order," it is visually easier to put pecking order on the x-axis, I think.  The simplest analysis here is to look at the median weights for each pecking order.  A more detailed analysis would connect up the dots for each pen so you could see how the 4 chickens in each pen relate.  You can do this with Lowess--try but don't spend too much time on it.

Sievers home  Math251-Fall01/DayP8.htm      9pm    9/16/01
This page belongs to Sally Sievers who is solely responsible for its content. Please see our statement of responsibility.