Choose a sample you can defend.
The two things a reviewer checks first are your sample and your data. This course takes you from a blank page to a sample-size calculation with every assumption written down, a defensible design, five data-quality checks, and a form that stops the usual field failures at source.
Narration by Amara Okafor · Watch the 1-minute overview below · Members watch in the dashboard
Choose a sample you can defend — and catch bad data before it reaches a report.
This is a practical, narrated online course that teaches you to size a sample you can defend, choose a sampling design and account for its design effect, run the five data-quality checks, and build a Kobo or ODK form that prevents the common errors at source. Across 135 slides you work through a single running example — a livelihoods survey of 2,000 households — from the sample-size calculation to a one-page data-quality note a donor will accept. It is built for M&E officers, programme managers and consultants, and assumes no prior statistics.
The 1-minute overview. The full course plays in your dashboard.
Where reports fall down is the foundations.
Not the analysis — the thinking and the numbers underneath it.
A number nobody can defend
A sample size picked by habit, with no margin of error, no design effect and no assumptions written down. The first thing a reviewer checks.
The group your sample missed
Convenience clusters, a stale frame, improvised replacements. The people most in need are quietly absent — and no formula can fix it.
The dataset that fell apart
Outliers deleted without a log, duplicates never found, cleaning done in the raw file. Three weeks of work you cannot explain.
The form that broke
Untested skips, no constraints, duplicate IDs. The errors arrive in a spreadsheet, long after you could have prevented them.
- Size a sample and state every assumption out loud
- Pick a design — and know what it costs you in precision
- Run the five data-quality checks and clean reversibly
- Set up Kobo or ODK so the usual field failures never happen
- Write a one-page data-quality note a donor will accept
- M&E and evaluation officers
- Programme and project managers
- Field research and data teams
- Consultants who design surveys
6 modules, one calm step at a time.
Open — how to use this course
What you will build, and the one survey we work through end to end.
Sample size in plain English
Precision, confidence, p, the design effect and non-response — with a calculator.
Designs & the design effect
Random, systematic, cluster, stratified, multistage and purposive.
The five data-quality checks
Completeness, consistency, accuracy, uniqueness, timeliness — and reversible cleaning.
Kobo & ODK setup
Skip logic, constraints, testing and daily monitoring.
Retrieval practice & next steps
Six questions to make the method stick, and how to apply it this week.
You leave with a one-page data-quality note.
The page that proves your number is defensible — sample, checks, and limitations.
SAMPLE · cluster, 2-stage · n planned 834, achieved 761 · response 91% COMPLETENESS · 97% required fields · worst enumerator 91% CONSISTENCY · 6 cross-field checks · 14 errors resolved ACCURACY · range checks · 9 outliers flagged, 3 removed (logged) UNIQUENESS · 2 duplicates found and merged LIMITATIONS · 4 villages inaccessible (security) · income refused 7%
The course ships the sample-size calculator, the checks, the cleaning-log template and this note as fill-in artefacts. Every number has a source.
The materials you keep.
Every lesson ends in a file. These are yours to download and reuse — included with the course and with membership.
Sign in to download — included with the course or membership.
Buy this course, or join and get them all.
One-time · yours to keep
- 135 narrated slides · 50 min
- The templates and the worked example
- Lifetime access in your dashboard
Cancel anytime
- This course and every future course
- The full ebook and template library
- The full skills library — 40 installable skills
Amara Okafor
Amara has written logframes, indicator sets and donor reports across dozens of evaluations. She built this course so that the documents every donor asks for take an hour — not a week.
Evaluator & M&E trainer. Every idea is shown on one real, worked programme you can copy.
The honest answers.
Who is this course for?
Anyone who designs or runs a survey — M&E officers, programme managers, field research teams and consultants. No prior statistics is assumed.
How long does it take?
About 50 minutes of narrated teaching across 135 slides. You can pause, rewind and work at your own pace.
What do I get?
The full video, the slides, the sample-size calculator, the design-effect table, the five checks and a one-page data-quality note template.
Is it included in membership?
Yes. Member and Pro memberships unlock this course and the whole library. You can also buy it on its own.
Does it cover Kobo and ODK?
Yes — a whole module on setting up Kobo or ODK to prevent the usual field failures, including skip logic, constraints and the three test submissions.
Do I need statistics?
No. The sample-size maths is taught in plain English with a calculator, and every formula is explained line by line.
Make your next logframe defensible.
Apply it to your next proposal this week — build the theory of change, then the logframe, and sign it.