Psychology 815:Computer Lab #6
The inference probability; Big D, little d, and r
October 11, 1996


Objectives:

  1. The inference probability
  2. Big D, little d, and r
  3. Convert d back to r and compute the BESDub and BESDlb

Objective 1: The inference probability

In chapter 7 (page 18) of Ralph's notes, he explains how to calculate the inference probability. Let me give you a practical short-cut: The inference probability (IP) for a positive correlation is the probability of rho being positive. Well, the probability of rho being positive is:

1- the probability of rho being negative.

So how does this make your life easier? I'll tell you. The probability of rho being negative is the probability (i.e., the p-value) of the Z-test for the correlation. In other words, all you have to do to figure out the inference probability is
 

  1. run the Z-test for the correlation: Z = r * (sqrt (N-1)) (see lab #5 for details)
  2. calculate the pvalue for r (see lab #5)
  3. IP = 1- pvalue
Simple, right? Ok, have SAS calculate the inference probability for the correlation between FITNESS and AGE from the data that you used in your exam. Hint: Don't forget the sample size!


Objective 2: Big D, little d, and r

a.) Get the means and standard deviations for FITNESS scores for men and women (test data)
b.) Tell SAS to create a data file from the output. Note: This is new; pay attention!!

Here's the SAS code:
 

PROC MEANS;

VAR FITNESS;

BY GENDER;

OUTPUT OUT=WORK.DATANAME     *creates a SAS data file;

MEAN = MEAN1-MEAN2           *Creates variables mean1 and mean2;

STD = STD1-STD2;             *Creates variables STD1 and STD2;
Note 1: It is not Mean1 minus Mean2. Think of it as Mean1 through Mean2.
Note 2: Two lines do not have semicolons. This is by design!

c) Tell SAS that you want to work with this output data file that you just created. For example: if your output statement created data file WORK.MIKHAIL, then your next line should be: DATA WORK.MIKHAIL;

Final note: Unfortunately, although the above program will give you the means separately in your output, it will only put one mean for each variable into the output data set. Therefore, we will have to do this objective in two steps. You've already done STEP 1. Look at your output. Now, for STEP 2, just punch in the values yourself for the two means and the two standard deviations. Then, you can go on to part d below.

d) Create variable Big D, which is defined as: Big D = mean1-mean2 (now it's minus)
e) Create variable little d, which is defined as: little d = Bid D/SDw

note: SDw is defined as: SDw = SQRT ((STD1**2 + STD2**2) / 2)


Objective 3: Convert d back to r and compute the BESDub and BESDlb

Ok, that's it for this week. I hope you guys had fun with the first exam :)