Interpreting Significance in Crosstabs

Overview

MarketSight assists you in interpreting your crosstabs by highlighting significant results with color, column letters, or both.

Significance Shading

Pink Shading

If a row variable header is pink in color, this indicates that there is a relationship between the column variable(s) and that row variable. If the row variable displays both mean and column %, the pink shading can potentially indicate the significant result from statistical testing of means, column %, or both. If the "Show significance pop-ups" option is enabled, mousing over a pink row will provide more information about the test used.

Blue Shading
If a cell is blue in color, this indicates a statistically significant result of a pairwise test. There is a significant difference between that cell and at least one other cell in the same row. If the "Show significance pop-ups" option is enabled, mousing over a blue cell will provide more information about the test used.

Yellow Shading
If a cell is yellow in color, this indicates a statistically significant result of a contrast test. There is a significant difference between that cell and the group of all other cells within in the same row. If the "Show significance pop-ups" option is enabled, mousing over a yellow cell will provide more information about the test used. When contrast testing is used, a larger font indicates the value in the cell is significantly larger, and a smaller font indicates the value is significantly smaller.

Column Letters

If the statistics option to use column letters is enabled, MarketSight assigns a letter to each column within the crosstab. Column letters are used to distinguish statistically significant results of pairwise tests. If one or more column letters appear in a cell, it indicates that the cell containing the letter has a value that is significantly larger than the cells referenced by letters.

When two confidence levels are used, pairwise tests with significant results at the lower of the two confidence levels will be denoted in lowercase letters. Pairwise tests with significant results at the higher Confidence Level will be denoted in uppercase letters.

For example, in the following crosstab, "Purpose of Usage by Gender," the column corresponding to Male is assigned the letter ”A” and the column corresponding to Female is assigned the letter ”B”.

11% of Males and 5% of Females answered that they ”Use it primarily for business”. The letter ”B” within the Male column for this response tells us that 11% is statistically significantly higher than 5% for this data.

P-Values

A p-value is a numeric indicator of the statistical significance of a test result. A small p-value (close to 0) indicates that the result of a statistical test is statistically significant (unlikely due to chance). A large p-value indicates that the results of a test are not statistically significant. In general, a p-value greater than 1 minus the confidence level is not statistically significant. For instance, if you have specified a confidence Level of 95% (or .95), a p-value greater than .05 is not significant.

Significance Notation in PDF Exports

In PDF exports, the Table of Contents will display an asterisk in front of each variable for which significant results are found. Column lettering is automatically turned on when exporting to PDF and cannot be turned off. Test names and p-values cannot be displayed in PDF exports.

Significance Notation in Excel Exports

MarketSight's statistical significance pop ups appear in Excel as comments.