A spreadsheet with several columns is a starting point, but the experimental design determines what those columns mean. Before opening the workspace, write down the question your experiment was designed to answer.

One outcome, one grouping factor

Ordinary one-way ANOVA tests whether the population means of two or more groups are equal. It uses one categorical factor, such as treatment, and a continuous outcome, such as concentration. A rejection says that at least one mean differs; it does not identify every differing pair. See the NIST ANOVA table and hypotheses.

Synthetic design example: independently assign culture preparations to control, low-dose or high-dose treatment, then record one concentration summary per preparation. This could fit a one-factor comparison if the preparations are independent experimental units. These are illustrative conditions, not evidence from a study.

Check the assumptions against the design

The ordinary model assumes independent errors, normally distributed errors within groups, and a common error variance. Examine residual plots alongside knowledge of the experiment. A median-centered Levene test can help assess variance differences, but a nonsignificant diagnostic is not proof that assumptions hold. Independence comes from how the experiment was conducted, not a plot. See NIST's one-factor model and Levene guidance.

Recognise designs this workspace cannot handle

Moleculytics currently supports independent groups only. Do not use this workspace for measurements before and after treatment in the same subjects, repeated time points, or cells nested within animals. Designs involving multiple factors, blocking, covariates, or unequal-variance methods also need an appropriate analysis elsewhere. Unequal group sizes are accepted here; that does not remove the common-variance assumption.

If you are unsure what counts as an independent value, read the experimental-unit guide before importing data.

Prepare for an interpretable result

Record the measurement unit, group definitions, experimental unit, exclusions and missing values. Choose your significance level with the study question in mind. Then open one-way ANOVA, inspect the import preview, and confirm the design.

The app accepts 2–20 groups, at least two included observations per group, and up to 10,000 included observations. These are software limits, not a recommendation for adequate sample size. Beta numerical restrictions and an unresolved high-offset precision failure remain: read validation.

References