Understand the relationship
The reasoning behind the result
The experimental unit defines independent replication
The experimental unit is the entity independently assigned to a condition or otherwise supplying an independent observation under the study design. Measuring it three times does not ordinarily turn it into three independent experimental units.
Technical repeats can characterize measurement variability and improve a unit's measurement. The planner preserves their multiplication into workload while keeping the independent-unit target separate.
Attrition inflation is a planning expectation
Nstart = ceil(Ntarget/(1 − a))
If the entered attrition fraction is a, expected retention is 1 − a. Dividing the target by that fraction gives the starting quantity whose expected retention reaches the target. Each group is rounded upward independently.
This is not a probability guarantee that the target will be met. Correlated losses, group differences and uncertainty in attrition can require another design. A 100% attrition assumption cannot yield a finite starting requirement and is rejected.
Measurements remain a separate resource total
Multiplying starting units by technical repeats estimates measurements if every starting unit proceeds through those repeats. Actual workload can be lower if units are lost earlier. Additional controls are counted as measurements and do not automatically increase independent sample size.
The entered target is not calculated from effect size, power or variance. Those inferential choices require an appropriate model and remain separate from this resource ledger.
Follow the numbers
A target of ten after 20% attrition
- For each group, 10/(1 − 0.20) = 12.5, rounded up to 13 starting units.
- Two groups need 26 starting units. At two technical measurements per unit, planned measurements are 52.
- Two additional controls bring the measurement count to 54; they do not change the target of 20 analysable independent units.
The plan distinguishes twenty target units, twenty-six starting units and fifty-four potential measurements.
Quick guide
How to use this calculator
- Enter independently justified target units for each group.
- Enter technical measurements per independent unit separately.
- Supply an attrition scenario and extra control measurement count.
- Read starting units and measurement demand as different totals; this planner does not perform a power analysis.
Calculation method
Calculation and interpretation
Translate a stated experimental design into a transparent unit and measurement ledger.
Starting units per group = ceil(target analysable units/(1 − attrition)); planned measurements = starting units × technical repeats + separate controls
Worked example
A target of ten after 20% attrition
The plan distinguishes twenty target units, twenty-six starting units and fifty-four potential measurements.
Starting units per group = ceil(target analysable units/(1 − attrition)); planned measurements = starting units × technical repeats + separate controls
Supported inputs
Precision and limits
Design arithmetic, not design approval
Entered allocations and counts do not establish adequate power, independent experimental units or control of confounding. Randomization must match the actual experiment and analysis; no biological procedure or treatment is selected.
Continue calculating
Related calculators
