Getting started
Open a sheet, check the design aStat suggests, press Run, and read the answer.
1. Wait for R
When the page opens, the top right corner says what R is doing. The first visit downloads R, which can take a minute on a slow connection; later visits take a few seconds. When it says R is ready, you can run analyses.
2. Open your data
- Open file reads CSV, tab-separated text, Excel (.xlsx and .xls), and OpenDocument (.ods) files, up to 50 MB. For Excel files, aStat opens the first sheet that has data.
- Try an example opens one of 23 synthetic datasets, one for each kind of design. They are simulated, not real animals or plots.
- Plan a study opens power and sample size, which needs no data.
You can type into the sheet, paste from Excel, and undo or redo with Ctrl+Z and Ctrl+Y.
3. Lay out your data
aStat expects the layout every statistics package expects, often called "long" or "tidy" data:
- One row per measurement. If each animal was weighed on three days, that animal has three rows.
- One column per variable. Treatment, block, animal ID, day, and the measurement each get their own column.
- A header row with a short, clear name for each column. Names such as
Diet,Barn,CowID,Day, andWeight_kgalso help aStat recognize the design. - Missing values can be left blank or written as
NAor.(as in SAS).
| CowID | Treatment | Day | Milk_kg |
|---|---|---|---|
| C01 | Control | 0 | 31.2 |
| C01 | Control | 14 | 32.0 |
| C02 | Fat | 0 | 30.8 |
| C02 | Fat | 14 | 33.1 |
4. Check the suggested design
Under What this sheet looks like, aStat lists the designs that fit your columns, most likely first, with the reasons and a question only you can answer, such as whether treatments went to individual animals or to whole pens. The rules are ordinary code, not AI, so the same sheet always gets the same suggestions.
Click Use this analysis on a suggestion, or pick any analysis yourself under Set up the analysis. If something does not fit (a missing column, too few animals), aStat says what is wrong before anything runs.
5. Run and read
Press Run analysis. You get:
- Answer: one sentence saying what the data show, with the test that supports it.
- Assumption checks: marked OK, Check, or Problem, with what to do if a check fails.
- Chart (for most analyses), which you can restyle.
- Full results: the tables a journal or thesis needs.
- The R script that produced every number.
See Reading results for what each part means.
Mixed and generalized models
The first time you run a mixed model or a count or score model, aStat downloads the R packages it needs (lme4, lmerTest, mmrm, or MASS). This takes from a few seconds to about a minute and happens once per browser.