Provides more easy input for SISA procedures which do pairwise analysis. It concerns the following procedures:
This procedure concerns three webpages which follow on one another.
For the pairwise tests you should have two variables in two columns which relate to two measurements on the same topic on the same person (or on a persons "pair"). What you want to know is if there is a change or difference between these two measurement on this single topic. For further details please consult the pairwise help page.
For the pairwise T-test the program calculates the mean change and the standard deviation of this change and sends you to a copy of the pairwise procedure page for further testing. For the T-test you need to input two columns of continuous interval level "means" data. Check the T-test option. The number of cases on which the t-test is based is the total number of valid cases, not measurements!
For two Wilcoxon Signed Ranks Test you input similar data, however, this data does not have to be normally distributed and can in someway be truncated. Check the Wilcoxon option. Note that the number of cases on which the Wilcoxon signed ranks test is based concerns the number of changed cases only!
For the McNemar you can basically input any two column two measurements data which indicates if someone changed or not and if this person changed in a negative or a positive direction. This can be interval data in which case the McNemar only considers the direction of the change, and not the extent of the change. It can also be nominal data indicating absence (’0’) or presence (‘1’) of a characteristic. Note that the number of cases on which the McNemar test is based concerns the number of changed cases only!
The data can be copied for example from a spreadsheet or word processor or the data can be typed in manually into the "Input data here:" field. The numbers in the input field have to be separated by spaces, or returns, or semicolons, or colons, or tabs, but can NOT be separated by comma’s or full stops. The data in the input field is read row wise, so first the data in the first row followed by the second, the third and all the other rows.
The formatting and tabulating of large datasets might take a while in which case there might be warnings, just select “continue” and in the end the computer will get there. Cases with unknown characters in the values are skipped, unknown are all non numerical characters except for the separator characters mentioned above. Information on the number of cases read and the number of invalid cases can be found in the address bar in the top of your browser.
This procedure is experimental!! Just give it a try and please let us know about your (negative) experiences and give us your suggestions @:
The data concerns a set of data on 10 cases measured twice. There is one invalid case and there is one case which did not change (20-20). Two cases change in a negative direction (37-34; 42-40), these two cases also have the least change of all cases.
For the t-test, mean change 5.8, number of cases 9, sd of change 5.33, perform t-test, 10 cases read, 1 invalid.
For the Wilcoxon, rank sum of the two negatively changed cases 3 (=1+2), number of changed cases 8, sd irrelevant, perform wilcoxon, 10 cases read, 1 invalid.
For the McNemar 6 cases changed in a positive direction and 2 in a negative direction, sd irrelevant, perform McNemar, 10 cases read, 1 invalid.