#MEAL #statistics #sampling #evaluation

We are thrilled to announce the launch of the EvalCommunity Sample Size Calculator – a dedicated tool designed specifically for monitoring and evaluation practitioners. Whether you are designing a baseline survey, planning a midterm evaluation, or conducting a high‑stakes compliance audit, this tool helps you determine defensible sample sizes using the same formulas that appear in major evaluation guidelines (CDC, WHO, MEASURE Evaluation).

Why a dedicated M&E sample size tool?

Most online sample size calculators are generic. They ignore the finite population correction (critical when you sample more than 5% of a small programme population), offer no support for stratified designs (essential when you need disaggregated data by region or beneficiary group), and omit risk‑based attribute sampling used in data quality assurance and audits. Our tool fills these gaps. It speaks the language of evaluators.

Three calculators in one

The interactive app (embedded below the article on the tool page) offers three tabs. Here is what each does and when to use it.

1. Simple random sampling (with optional FPC)

The classic formula: n₀ = Z²·p·(1‑p) / e². You choose confidence level (90–99%), expected proportion (p), and margin of error. If you enter a population size N and enable FPC, the tool checks whether the sampling fraction exceeds 5% and automatically applies the correction. It shows both unadjusted and adjusted n, plus the actual coverage percentage. A gauge visualises the sampling fraction. This is perfect for household surveys, output monitoring, or any homogeneous population.

2. Stratified sampling – proportional, equal, or optimal allocation

Stratification is a cornerstone of M&E: you want to report separately for different regions, programme components, or gender groups. The stratified panel lets you add multiple strata, each with a population and (for optimal allocation) an estimated standard deviation. After you pick a confidence level and margin of error, the tool calculates the total sample using the SRS formula (with FPC on the total population) and then distributes it according to your chosen method:

• Proportional – sample ∝ stratum size (maximises overall precision).
• Equal – same n per stratum (ensures enough cases for subgroup comparisons).
• Optimal (Neyman) – weights by N·σ (minimises variance when stratum variability differs).

A detailed table shows per‑stratum sample sizes and sampling percentages. A minimum of 2 per stratum is enforced to allow any meaningful analysis.

3. Risk‑based QA sampling (attribute / hypergeometric)

In audits, data quality checks, and compliance reviews the question is not “what is the average?” but “can we detect a problem if it exists?” This panel implements the formula n = log(1‑C) / log(1‑p), where C is the confidence level and p is the assumed defect rate (e.g. 2%, 5%). It also applies FPC when the sample exceeds 5% of the lot. A unique feature is the risk classification: low, medium, high, or critical. Based on your selection, the tool suggests a multiplier (up to 1.5× for critical risks) and explains the reasoning — exactly what you need when drafting an audit sampling plan.

Methodological rigour you can trust

All formulas are displayed in a dedicated Formulas & Methodology section directly below the calculator. We also include:

• A table of Z‑scores for the four most common confidence levels.
• Guidance on margin of error (e.g. ±3% for impact evaluations, ±5% for standard surveys, ±8% for rapid assessments).
• An explanation of the finite population correction rule (n₀/N > 5%).
• A note about the design effect (DEFF) for cluster sampling – we recommend multiplying the calculated n by 1.5–2.0 when cluster sampling is used.

All of this is presented in a clean, accessible interface. The design uses a calm, professional palette (navy, teal, warm neutrals) and is fully responsive – it works on phones, tablets, and desktops.

Practical examples from the M&E cycle

Example 1 (Baseline survey): You have a programme reaching 2,800 farmers in three regions. You want 95% confidence, ±5% margin of error. You don’t know the prevalence of a key indicator, so you use p=0.5 (conservative). Without FPC, n₀ = 385. But because 385/2800 = 13.8% > 5%, the finite population correction reduces the required sample to 336 – saving you 49 interviews.

Example 2 (Stratified endline): Your population is 4,500: Region A (2,000), Region B (1,500), Region C (1,000). You need sufficient cases in each region to compare outcomes. Equal allocation gives you about 60 interviews per region (total 180), while proportional gives 80/60/40. The tool lets you toggle and immediately see the implications.

Example 3 (Data quality audit): A partner has 500 project files. You want 95% confidence that the error rate is below 5% (i.e. if errors exist in more than 5% of files, you’ll detect at least one). Base n = 59. If the risk is classified as high (e.g. financial transactions), the tool suggests a 1.25 multiplier → 74 files. If critical, 1.5× → 89 files. This builds in professional judgment directly into the calculation.

(Bookmark this link – you'll use it again.)

Designed for the M&E community

This tool is part of our ongoing effort to provide high‑quality, freely accessible resources for evaluation professionals. The sidebar of the calculator page includes quick reference benchmarks, use‑case tags, and links to key references (CDC framework, WHO LQAS, BetterEvaluation, MEASURE Evaluation). We also link to the EvalCommunity Academy for those who want to deepen their skills in applied statistics, AI in M&E, and evaluation methodology.

We welcome your feedback. If you encounter any issues or have suggestions for additional features (e.g., cluster sampling adjustment, finite population for each stratum, integration with power analysis), please reach out via the contact page on EvalCommunity.

About EvalCommunity

EvalCommunity is the largest global network of monitoring and evaluation professionals. We provide jobs, career guidance, a directory of consultants, and a rich library of resources. The new Sample Size Calculator joins our growing set of practical tools designed to strengthen evaluation practice worldwide.