Calculations in Cross Table

In the calculations tab, you can select the calculations to apply to the selected questions. Only relevant calculation types are shown here. This means that the Categorical questions have one set of available calculation types, and Numerical questions have another set of calculation types.


Calculation types

  • The categorical calculation types ‘Percentage share’ and ‘Count’ are only shown if at least one categorical question has been selected
  • The categorical calculation type ‘Mean’ is only shown if at least one categorical scale question has been selected
  • The Significance testing options are only shown if at least one categorical question has been selected
  • Numeric calculation types are only shown if at least one Numeric question has been selected
  • The Correlation analysis option is only shown if at least one Numeric or one Single-choice categorical question has been selected. If only a Multiple-choice categorical question or an Index question has been selected, then the Correlation analysis option will not be shown


Standard Calculations panel


In the standard calculations panel, you will find the following calculation types:

  • Categorical questions: Percentage share, Mean, and Count
  • Numeric questions: Sum, Mean, Median, Min, Max, and Percentile
  • To use a calculation, simply select which ones you would like to use. As shown in the image below, you can also adjust the number of decimals per calculation type. If you would like to add any suffix/prefix, like a “%” sign, to a table, then you can define that here. The image below shows what the standard calculations panel looks like when categorical scale questions have been selected in the Rows and Columns tab.



Same for the next section “Standard Calculations panel”


The image shows what the calculations panel looks like when Numeric questions have been selected in the Rows and Columns tab. At the top of the standard calculations panel, the weight can be turned on and off. If the project contains multiple weight variables, then the weight variable to use is in the calculation is also selected in this area. The image below shows the setting used to turn the weight on and off.


Correlation Analyses

The correlation analyses use the same Pearson’s formula as the StoryTeller chart and table. The correlation analyses can be applied to the tables as soon as any Numeric or Single-choice categorical question has been selected in the Rows and Columns tab.


Step by step

  1. Select the variables for Rows and Columns (any Numeric or Single choice categorical).
  2. Go to Calculation Tab and in Correlation analysis and enable the calculation.
  3. Select if the R or the R2 value will be calculated and question the test will be performed against.
  4. Adjust the number of decimals, Unit and Suffix you want to see in the table with results.
  5. Press Generate to perform the analysis as per the settings selected.
  6. The information above the table shows which question the correlation analysis is made against.
  7. The correlation result will appear as a last row in the table.


Significance Testing

In the significance testing panel, you can turn on significance testing. The settings are similar to creating a chart or a table in the StoryTeller. This means that the following must be defined:


Step by step

  1. Select the variables for Rows and Columns.
  2. Go to Calculation Tab and in Significance testing enable the calculation.
  3. Select the significance level from the dropdown menu (1%, 5% or 10%)
  4. Select the base size option to use in the calculation (weighted, unweighted or effective base).
  5. Select if the test will be made across columns or rows.
  6. Select Comparison columns.
  7. Select If any Base limit or Respondent Count limit must be fulfilled to perform the test.
  8. Select Results to show.
  9. Press Generate to perform the analysis as per the settings selected.



Note: The base size option to use in the calculation is only shown if the result is weighted. If the result is unweighted, then an unweighted base is always used in the calculations (It is strongly recommended that these options are limited by the administrator during the setup so the report user doesn’t need to select which one to use.)



Additional Options

Visualize Significance Test with Colored Values


The significance test results are shown with symbols such as +/- or with colored cell backgrounds. There is also an option to allow visualization of the results with colored values. When all columns are tested against all columns (or all rows against all rows) the visualization of the result is always made by “symbols” (like a+, b-, c-, etc.) since each value/cell can be both positive and negative. When only testing against one column (or row) the result can be shown with symbols or as colored cell backgrounds since each cell can only be positive or negative. This option is not shown when all columns are tested against all columns or when all rows are tested against each other.


To use colored background to visualize the significance test, simply select the “Colored background” option in the in the “Show result as” list.

The significance test result can be visualized with colored values when a value is tested against another value. This option is not available when all values in the table are tested against each other. The value colors for positive and negative results are defined in the Project settings page, as shown in the image below.



Benchmarks

In the Benchmark panel, you can turn on the benchmarking calculations to benchmark one result against other results. The settings are similar to creating a chart or a table in the StoryTeller. As with the other calculations select the variables for Rows and Columns first.



Step by step

1.         Go to Calculation Tab and enable benchmark calculation.

2.         Select calculation type (Units, Percentage share or Index/Conversion rate).

3.         Select if the original result will be hidden or shown together with the benchmark result.

4.         Select if the benchmark calculation shall be made across columns or rows.

5.         Select Comparison columns.

6.         Select if the comparison column will be hidden or not.

7.           Press Generate to see the table with Benchmark analysis.



Three types of benchmark calculations are available:

  • Units: 50-80=-30 (B-A)
  • Percentage share: (50-80)/80*100=37,5% (B-A)/A*100)
  • Index/Conversion rate: 50/80*100=62,5% (B/A*100)

Above are descriptions of the formulas. In this example, result A is 80% and result B is 50%. The B result is benchmarked against the A result:


All three calculation types can be applied to all calculation types such as %, Mean values, and Count & Correlation analyses result.


Settings in Cross Table

In the Settings tab, you will find the base size options and some formatting settings. In the base size options panel, you can select if the base size should be shown or not. If the result is weighted, then both the weighted and the unweighted base size can be shown, or just one of them. If the result is unweighted, then only the unweighted base size can be shown. These settings are made in the “Display base” area. To show the base size in the table, simply select the Weighted base and/or the Unweighted base options in the Display base area.



In the area “Hide and warn for series with low base size/result” you will see settings used to hide results due to a low base size/result and the function for warning due to a low base size. These settings can be turned on/off, and the limit can also be entered in the input fields. The default limit and the default setting can be defined in the setup phase. If the result becomes hidden or if there is a warning due to a low base size, then an information message will appear above the result table in the Generate tab (see example image below). The displayed text (“*Low base size” in the image below) comes from the project settings page, so it can be adjusted to suit the needs of the project (see the second image below).The image below is an example of a table with a low base size, as explained by the displayed text “*Low base.” This text comes from the project settings page.


The image below shows the project settings page. The following texts will appear if the base size is low or if a result is hidden due to a low base size.



Additional Options

In the “Hide empty series” area, you can hide rows and columns without data/results. The following three options are available. They can be used simultaneously if needed.

Hide date gaps: Use this option to remove intervals (weeks, months, etc.) without data. See example 1 below.

Hide empty series: Use this option to remove splits without data, and answer options with a count = 0. See example 2 below.

Hide splits with No Data: Use this option to remove a split option which without data. See example 3 below.

Example 1: By activating “Hide date gaps,” all the highlighted columns will be hidden.



Example 2: By activating “Hide empty series,” all the highlighted rows and columns will be hidden.



Example 3: By activating “Hide splits with No Data,” all the highlighted columns will be hidden.