PROGRAM INSTRUCTIONS FOR DATA ANALYSIS

The data analysis program requires a Java enabled browser, for example Firefox, Netscape or Microsoft Internet Explorer, running in Windows(9x/ME/NT/2000/XP), Linux, or Macintosh equivalent. If your browser does not have this capability, the "Return to Main Program" link will appear immediately under the title when you access the tool kit. Otherwise, when you access the tool kit, the program applet will be initialized and appear on the screen.

The data analysis program loads with a set of sample data. If you want to use the sample data for practice with the analysis programs, click the continue button, and three option buttons will appear to the right of the data entry box. The initial options are:

describe, analyze, and graph.

To enter your own data for ANOVA, autocorrelation, Mann-Whitney U, t test, and/or time series analysis, click the clear button. The "Data for Analysis" list will disappear, and you will see a box for entering new data. You have two options for entering data.

For direct data entry, type the scores and variable identifications in the data entry box, using the format below with one measurement per line. When you have finished, click the load data button and the data will be moved to the analysis box.

The other option, preferable especially when you have a great deal of data to enter, is to first prepare the dataset using the format below with your text editor or word processing program. Type the measurement scores and the variable identifiers with one measurement per line. When you have finished, copy that information to your clipboard and then paste the data in the data entry box.

The format for data entry in this program is as follows:

score space and then 1, 2, or 3 to identify the variable.

For example:

89 3   will be interpreted as a score of 89 on variable 3
893    cannot be interpreted because no variable is identified.
3.5 2  will be interpreted as a score of 3 and one-half on variable 2
-2 1   will be interpreted as a score of minus 2 on variable 1

If you make a mistake while entering the data, click the clear button and start again.

When you click the Bayesian, binomial, or chi square analysis button, a data entry box will be visible with instructions for entering the required data.

For the Bayesian analysis, you enter two p values. The first represents your 'prior belief' about the outcome based on prior study(ies). The second is the p value outcome from a new time series analysis. Enter the numerals with a space between them (e.g. .051 .119); p values are always positive and less than 1.00.

For the binomial analysis, you enter two values. The first represents the number of successes; the second represents the total number of trials. Values are entered with one space between (e.g. 11 19).

If you have at least eight data points in variable 1 for baseline, the chi square module will open with the number of data points above and below the celeration line in the baseline and the number of data points above and below the celeration line in variable 2. You can also use this module for a standard 2 x 2 chi square analysis, entering two values for the top row and then two values for the bottom row, each separated by a space (e.g. 11 19 30 30).

The time series graph and time series graph with celeration line plot up to 30 data points, beginning with those identified as variable 1, followed by any identified as variable 2, and then by any identified as variable 3. The alternating treatments graph plots a maximum of ten values on each variable.

Graphs are limited to plotting whole numbers and to values less than or equal to 100.

The time series and autocorrelation programs assume that you enter the data in chronological order. In the sample data set, the variable 1 scores are together representing the baseline measures, but the variable 2 and variable 3 scores are intermixed as would be the case in an alternating-treatments design.

Baseline data should be designated as variable 1 to insure proper identification in some of the analysis programs.

It is not necessary to use all three identifiers. For example, if you simply want to run a time series on one phase of the study, all data could be identified as variable 1.

The analysis results have been crosschecked with many datasets. Results are consistent (with occasional small decimal point differences from rounding conventions) with other analysis programs. This does not, however, preclude the possibility of an unanticipated program error from a particular dataset. If you obtain a result which does not appear at all consistent with the data, first check the "Data for Analysis" list to be sure your data were entered as intended.