Start with the study question — not the statistical test.
VetSample helps identify the experimental unit, recognize paired or clustered observations, and select an appropriate sample-size calculation using veterinary-language questions.
Veterinary study design assistant
Start with your research aim
VetSample can identify the appropriate design from a few simple questions, or you can choose the method yourself.
1What type of value is your main outcome?
For example: body weight or serum concentration = continuous; pregnancy or mortality = yes/no.
2What is your study design?
Choose whether different animals form the groups, the same animals are measured repeatedly, or both structures occur together.
3How many groups, factors or repeated measurements are there?
VetSample uses these numbers to distinguish t tests, one-way ANOVA, two-way ANOVA and mixed ANOVA.
1At what level do you want to estimate prevalence?
Example: proportion of individual cattle positive, or proportion of herds positive.
2How will units be sampled?
Animals sampled within the same herd/farm/enclosure are often correlated and may require a cluster adjustment.
3Is the total target population size known and reasonably small?
If yes, the calculator can apply a finite-population correction when appropriate.
1Which diagnostic property is primary?
You may plan for sensitivity, specificity, or require adequate precision for both.
2How will animals enter the study?
Target-population sampling and known disease-status sampling have different implications for prevalence and predictive values.
3Is an acceptable reference standard available?
Sensitivity/specificity calculations assume disease status can be defined by a reference standard.
1How is the first variable measured?
2How is the second variable measured?
3Does each animal/unit contribute only one independent pair of measurements?
Repeated or clustered pairs usually require a model that accounts for within-unit dependence.
1What type of outcome are you trying to explain or model?
The current guided regression calculators cover continuous outcomes. Example: milk yield, body weight, serum concentration.
2What is the main question of your regression model?
Choose whether you want to test the model as a whole or the additional contribution of newly added explanatory variables.
3How many explanatory variables (predictors) are planned?
Count the explanatory variables entered in the planned model; the intercept is not counted.
Independent unit, natural clustering, technical replicates and assumption source
Recommended approach
Study design check
Method preview
Veterinary example
Typical study structure
Key planning inputs
Choose this when
Common settings
Design and analysis settings
Step 1 of 2 — Review the common settings
Confirm the analysis goal, significance level, target power, sidedness, expected loss and any clustering adjustment before moving to method-specific inputs.
Optional adjustment for natural clustering (ICC-based)
No clustering adjustment is applied unless you explicitly activate it and enter defensible assumptions. This approximation is intended for natural clustering when the individual animal/unit remains the analysis unit; it is not a substitute for design-specific planning when treatment is assigned to the herd, cage, enclosure or other cluster.
Step 2 of 2 — Enter method-specific inputs
Now enter the assumptions required by the recommended method. Explanations and examples are provided next to the fields.
Calculation result
Interpretation
Calculation details
Suggested methods text
Adapt this text to your protocol or manuscript.Scenario analysis
Sensitivity table
Power explorer