1) Descriptive statistics for repeated measures
2) Program WITHIN
Objective 1: Descriptive statistics for repeated measures
For this objective, we will need a small data set. Unfortunately, I am having some technical difficulties with my web server, so you will have to type it in on your own. Here's the INPUT statement and the data: INPUT id pretest posttest; 1 100 135 2 112 150 3 95 112 4 150 190 5 118 130 6 138 158 7 120 170 8 88 95 9 107 133 10 124 147 Once you have this data entered, just follow steps 1-5 below. Don't forget to ask SAS for the output. 1) Use SAS to create a new variable (changeY) which is defined as: changeY=posttest-pretest 2) Have SAS compute means and standard deviations for all three variables. 3) Have SAS compute s (standard score) where s = SD(changeY)/SD(pretest) This is the relative effect size of the interaction 4) Compute Vw where Vw =[ [SD(pretest)]**2 + [SD(posttest)]**2]/2 5) Compute SDw where SDw = [sqrt (Vw)]**2
Objective 2: Using program WITHIN
For this objective, you will need to get new software from me. When you get to this point, bring me your disk, and I will copy the software to your disk. 1) You will need to create a data file. You can do this with any word processor or text editor (e.g., program editor in SAS). Line 1: mean (time 1) mean (time 2) Line 2: SD (time 1) SD (time 2) Line 3: N test-retest reliability Line 4: reliability of measure (t1) reliability of measure (t2) Note that line 4 is optional, but you need all the rest. If you want you can make up your own data. If not, here's a sample you can use: 100 104
10 10
50 .48 .64 .64 Make sure that you do not have any blank lines at the top and if you are using a word processor, make sure you save your file as an ASCII file (it's one of the options under SAVE AS...). Make sure you remember what you named this file. You will need it for the next step. 2) Now run program WITHIN a) Find the DOS prompt in the MAIN folder and double click on it b) type a:within 3) You want to print the output to your disk. When it asks you to enter the disk file name type a:within.out (this will be the name of your output file). When you get to analyses options, choose option #1 (two measurements; data on one group). The rest of the instructions on the screen should be pretty straight forward. 4) You will run program WITHIN 2 times. The first time ignoring measurement error and the second time correcting for attenuation 5) The output is pretty complicated. You can look at it by using any word processor or text editor to open up the file within.out which should now be on your disk. I suggest that you print the output and bring it to class on Tuesday. Ralph said that he will take some class time to go over it. Don't forget about the exam next week!!!