What is adjusted correlation

correlation

Correlation analyzes are used to examine relationships between variables, in particular the analysis of “the more, the more” or “the more, the less” relationships. MAXQDA Stats supports the calculation of bivariate correlations, i.e. the correlation of two variables.

To calculate a correlation in MAXQDA Stats, select either in the main menu

  • Correlation> Pearson, often used for interval-scaled variables or
  • Correlation> Spearman, often used for ordinal scaled variables

The following dialog then appears in which you can select all the variables for which pairwise correlations are to be determined.

Note: All variables of the type integer and floating point are always available, regardless of the scale level of the individual variables.

Variables can be selected in several ways:

  • Use the mouse to drag a variable into the right-hand window area.
  • Select one or, while holding down the Ctrl or cmd key, select several variables by clicking with the mouse and drag the marked variables to the right into the "Rows" or "Columns" area. Alternatively, click on the blue arrows.

To include the code frequencies of a code as a variable, switch to the "Codes" tab and select the desired codes. (This is only possible if MAXQDA Stats was started with the data of a MAXQDA project.)

Select the option Exclude cases by listto only consider those cases that have a valid value for all selected variables.

Results table

After clicking on OK a table of results appears which contains the pairwise correlations. Since all selected variables are listed both in the columns and in the rows, the output is symmetrical on the diagonal, i.e. contains all calculated correlations twice.

Three values ​​are output in each cell:

  • the calculated correlation coefficient - depending on the selected function, i.e. Pearson's r or Spearman's rho,
  • the p-value for the significance of this correlation coefficient as well
  • the number of valid cases for each combination of the two variables.

In the picture above, for example, the correlation between the variables “Age” and “Life Satisfaction Index” is highlighted in green. Pearson's r is 0.242 here, the p-value is 0.0153 (i.e. below the significance threshold of 5%) and the number of valid values ​​for this combination of variables is 80 cases.

The 1-sided p-value is output as standard. With the help of the list selection in the toolbar, you can switch to 2-sided p-values. The p-value is only determined from 5 cases and is always 1 for lower case numbers.

Note: With option selected Exclude cases by list the number of valid and missing values ​​is only displayed in the upper right corner of the window and the cells only contain the correlation coefficient and the p-value.

Highlighting significant correlations

Click on the icon Highlight significant correlations to highlight all cells with a p-value less than 5% in green. The lower the p-value, the darker the highlight color:

  • less than 5% à light green
  • less than 1% à green
  • less than 0.1% à dark green

Overview of the functions of the toolbars

In the upper part of the result window there are two toolbars that allow access to important functions:

Table view / diagram view - switches between the results table and the scatter graphs.

To update - calls up the dialog for creating the correlation table again.

Insert in output viewer - inserts the currently displayed table into the output viewer.

Copy to clipboard - copies the displayed table or the selection in the table to the clipboard, for example to paste it directly into Word.

To press - shows a print preview and prints the table.

Export - exports the table as a file in Excel, website (HTML), or rich text format for Word and other word processing programs.

Highlight significant correlations - Activate this option to highlight cells with a p-value less than 5%.

Undo changes - step-by-step resetting of changes to a table (deleting rows, combining rows and moving rows).

Restore changes - Step-by-step recovery after a change to a table has been undone.

Sort table and adjust columns

The table can be sorted by clicking on the column headings: the first click is in ascending order and the second in descending order. Another click restores the original sorting.

Tip: You can also move individual result lines. To do this, click on a line and move the line up or down while holding down the mouse button. Moving also works for several lines selected at the same time.

The width of the columns themselves can also be changed with the mouse and their position can be changed by clicking on the column heading and dragging with the mouse.

Delete lines

The results table is interactive and individual or multiple lines can be removed from the table, for example to limit the evaluation to individual variable values: Right click on a line and choose Clear. You can also select multiple lines while holding down the Ctrl or cmd key.

Deleting a column or row in a correlation table only affects the view. The numbers will not be recalculated even if this is due to the option Exclude cases by list would change the number of values ​​to be taken into account.

Tip: A column can be hidden from a table by clicking the column title with the right mouse button and Hide column choose. When the Frequencies function is called up again, the column is automatically displayed again.

Transfer the results table to the output viewer

After you have adapted and completed the table according to your needs, it can be transferred to the output viewer, from where it can later be exported together with other tables and diagrams:

  • Click on the icon Insert in output viewer to transfer the currently displayed table to the output viewer.
  • Click on the icon Insert all result tables in the output viewer to transfer all created tables to the output viewer.
Tip: The column widths of the table that you may have adjusted are retained during the transfer.

Export and print the results table

With the help of the symbols displayed at the top right, you have several options for exporting and printing the currently visible results table:

Copy to clipboard - copies the entire table or the current selection in the table to the clipboard, for example to paste it directly into Word.

To press - starts the printing process and shows a print preview in which you can make settings such as margin, orientation, header and footer.

Export - exports the table as a file in Excel, website (HTML) or rich text format for Word and other word processing programs.

Show scatter plots

Click the icon in the results table Diagram viewto display a scatter plot for each paired variable combination.

You can use the two selection lists in the upper toolbar to define the variable for the X-axis and the Y-axis. The selection lists contain all the variables that were selected for calculating the correlations.

It is not possible to adjust the appearance of the diagram. The size of the diagram automatically adapts to the size of the window. Enlarge the window for a larger diagram.

The created scatter diagrams can be transferred to the output viewer and saved for later export. To do this, click on the symbols

Insert diagram in output viewer to show only the current view to transfer or

Insert all diagrams in the output viewerto insert the scatterplots of all combinations of variables at once.

Note: The diagram is inserted in the output viewer in the same size as it is displayed, but not wider than 650 pixels.

To export or print a diagram directly, the usual symbols are available at the top right of the window:

Copy to the clipboard, To press and Export.