This page tests without the raw data. The summary statistics are entered directly.
It is used in three specific situations: a supplier report that provides only the mean and standard deviation, a publication whose conclusion you want to verify, and a verification calculation before running a test.
Three tabs:
- Quantitative : Enter the following for each sample: size (n), the average and the’standard deviation (sd). Ellistat tests as if the data were there; ;
- Qualitative : Enter the number with the characteristic (k) and the size (n) to compare proportions; ;
- Independence : a row × column contingency table, used to test the relationship between two categorical variables.
General Settings: the alpha risk, and the’alternative : bilateral, left unilateral or right unilateral.
💡 Bilateral or unilateral?
Bilateral asks, «Is there a difference?»,
The unilateral asks, «Is this better?» The unilateral is more powerful, but The direction must be selected before viewing the data. Choosing it after determining the direction of the spread doubles the alpha risk.
On the "Independence" tab, Ellistat selects the appropriate method based on the sample sizes and displays it: *n=1: exact binomial, n=2: Fisher's exact test, n≥3: χ²*. The validity of the χ² test is checked by the Cochran's rule, in automatic mode, Ellistat groups items together as needed. The expected headcount are displayed, color-coded by standardized residual: the most heavily colored cells are those that show the dependence.