The Walr Platform includes significance testing using both Chi-square (Chi2) and T-Test methods.
- Chi2 tests enable you to see which categories differ from total.
- T-Tests highlight differences between each of the other categories sampled. (The unicodes used to differentiate between the columns are in an Excel at the bottom of this article. There are 300+).
Tables can display values for Chi2, T-Test, or both at once.
Video Guide: https://youtu.be/I6O_g7sX5KA
TABLE OF CONTENTS
Chi-squared Test
From your crosstable click Options:
1. Click Statistical to display your statistical options:
2. Selecting "Chi2 auto select" allows the platform to determine the confidence level of the test automatically based on your data when the Chi2 Risk level field is 0.
Number of Cases | Confidence Level |
|---|---|
| 1-300 | 90% |
| 301-1000 | 95% |
| 1001-4000 | 97.5% |
| 4001-20000 | 99.5% |
| 20001 or more | 99.9% |
3. Entering a "Chi2 Risk level" will override the automatic levels based on number of cases. The value entered must be between 0-50.
Confidence level equals 100 minus Risk Level. For example, entering "10" will provide you with a 90% confidence level.
Values of the Chi2 test are shown in the table cells using green for values significantly higher than the total or red for values significantly lower than the total.
The confidence level of the test run is displayed in the documentation cell in the upper left corner of the table (in this case 95%)
T-Test
From your crosstable click Options:
1. Click Statistical to display your statistical options:
4. Selecting "T-test auto select" allows the platform to determine the confidence level of the test automatically, based on your data, when the T-test risk level field is 0.
Number of Cases | Confidence Level |
|---|---|
| 1-200 | 90% |
| 301-400 | 95% |
| 401-800 | 98% |
| 801-1600 | 99% |
| 1601-3200 | 99.5% |
| 1601-6400 | 99.8% |
| 6401 or more | 99.9% |
5. "T-test mode" will default to All possible pairs but you may select the test to default to Block.
- All possible pairs will check all columns against each other.
- Block will limit this pair checking to a specific question block.
6. "T-test characters between blocks" will reset the column label to "a" at the start of the next block. For example, The "Age" block may contain something like 18-34 =a; 35-54 =b; and 55+ = c. If the next block were genders, "Male" would restart at the letter "a" instead of carrying on the alphabet with "d" if this option is toggled on.
7. Entering a "T-test Risk level" will override the automatic levels based on number of cases. The value entered must be between 0-50.
Confidence level equals 100 minus Risk Level. For example, entering "10" will provide you with a 90% confidence level.
- The results of the significance test will be visualized as follows:
- Each exclusive column will have a symbol shown at the top of each column (“Column symbols for T-test”) This will typically be a letter from a-z.
- Tested pairs are shown with the exclusive column listed first followed by a question mark then which other column(s) were tested. In this example c?abd means that column c has been tested against columns a, b, and d.
- A greater than / less than sign (< or >) shows which cell is significantly higher or lower than the test column. In this example column c is less than column d (<d) by a statistically significant amount.
