Multiple Regression Analysis

The multiple regression analysis tool works in two distinctive ways, depending upon whether data fields are selected for initial analysis (guided analysis vs unguided analysis).

Unguided Analysis

If no data fields are selected, it performs regression analyses for each set of two data fields and presents the results in a table. In the DataPlus Console, open the data file "Series_2" but don't choose any data fields. The multiple regression launch icon (fourth from the left) should be active (if more than one data set is present, it will be dimmed).

Clicking the multiple regression launch icon will bring up a floating window containing a table with the data.

The 'X-Values are shown as a row across the top, and the 'Y-values' are the columns on the left side. Each cell (unless they are the same) has the same parameter, which is selected from the menu at the bottom. The default parameter is Pearson's r-value. Because the r-value is symmetrical, the opposite pairs will be identical, but other parameters (slope, intercept) may not be.

Guided Analysis

If a value is assigned to the X field, the multiple regression tool behaves differently. The X-value is placed in the first column and tested against all other values in the set. If any of the correlations are significant, a new column is generated, which is the residual [f(x) - y] of the two values. This is tested again against all other fields, and the testing will continue until no further significant correlations are found. Since this can occur with a number of fields, the results can look a bit confusing. In the following example, the X-field was "Result",.

There were two significant correlations, with "Population" and with "Result 2". This generated two residuals columns. To see what they represent, holding the mouse over the header will show a tool tip box explaining the makeup of the field.

As can be seen here, there were no further significant results, so additional residuals were not calculated.

A particularly nice feature is the ability to double-click on a box and automatically launch a regression plot tool with the fields represented in the box. For example, double-clicking on the lower-right box (r-value 0.1296) gives the following:

This example just shows what we already knew to be true - there is no correlation, but in other cases showing the results can be very useful

This is the end of the tutorial at this time; more will be added as other tools are completed.

Back To Index