Performing multiple analyses

Comparison tests

For tests where the Y variables are compared to X variables you can use several methods for pairing the multiple Y variables and X variables.

There are three possible options to choose from, although OxonStat will only present you with the ones that are relevant for your selected variables and test.

Matched Y-X Pairs

OxonStat pairs the selected variables sequentially. It executes a separate statistical analysis for the first Y variable against the first X variable, the second Y variable against the second X variable, and so on. It is only available when the number of Y variables is equal to the number of X variables. For example:

In this example, the comparisons would be:

• Weight vs Height
• Height vs Age
• Age vs Gender

The number of analyses produced will be equal to the number of Y variables.

Exclusions

Certain tests will refuse to compare a variable with itself, since the results would have no meaning. For example Paired-t, Wilcoxon Signed Rank, Correlation / Regression and Spearman Rank. For these tests if the Y variable and X variable is a match the test will be excluded. For example:

In this example, the comparisons would be:

• Weight_1 vs Height
• Sys_BP vs Sys_BP would be excluded as the variables match
• Dia_BP vs Glucose

The analyses produced will be equal to the number of Y, minus the number of variables that in order are paired against themselves.

All Y-X Pairs

OxonStat pairs your selected variables comprehensively. The software executes a separate statistical analysis for the first Y variable against every selected X variable, then repeats this loop for the second Y variable against every selected X variable, and so on.

Warning, it is possible to generate enormous numbers of analyses (and printouts!) using the All Y-X Pairs option.

For example:

In this example, the comparisons would be:

• Weight_1 vs Height
• Weight_1 vs Age
• Weight_1 vs Gender
• Height vs Height
• Height vs Age
• Height vs Gender
• Age vs Height
• Age vs Age
• Age vs Gender

The analyses produced will be equal to the number of Y variables multiplied by the number of X variables.

Exclusions

Certain tests will refuse to compare a variable with itself, since the results would have no meaning. For example Paired-t, Wilcoxon Signed Rank, Correlation / Regression and Spearman Rank. For these tests if the Y variable and X variable is a match the test will be excluded. For example:

In this example, the comparisons would be:

• Weight vs Height
• Weight vs Age
• Height vs Height would be excluded as the variables match
• Height vs Age

The analyses produced will be equal to the number of Y variables multiplied by the number of X variables, minus the number of variables that are set as both Y and X.

Each Y vs. All Xs Combined

OxonStat pairs your selected variables multi-dimensionally. The software executes a single statistical analysis for the first Y variable against all selected X variables combined, then repeats this loop to run a single analysis for the second Y variable against all selected X variables combined, and so on.

In this example, the comparisons would be:

• Weight_1 would be compared with Age, Gender, Sys_BP and Dias_BP
• Height would be compared with Age, Gender, Sys_BP and Dias_BP

The number of analyses produced will be equal to the number of Y variables.