MYP Design Success Criteria and Testing Methods Notes

Success Criteria Are The Specification With A Test Attached

1. A specification point says what the product must do, and a success criterion says how you will prove it does.
2. The two are written side by side, because criterion B asks your specification to outline the success criteria for the solution.
3. Pairing turns a wish into something with a result, so "must be stable" becomes "tilt the loaded shelf to 10 degrees and it must not topple".
4. Every criterion needs three parts, which are what you measure, how you measure it, and the number or outcome that counts as a pass.

Why The Pair Gets Written Together

1. Write the specification now and the tests six weeks later, and you will invent tests you can pass instead of tests that matter.
2. Writing both at once also catches points that cannot be tested at all, which is the fastest way to spot a weak specification.
3. A point you cannot test with the equipment in your school needs rewriting today, not discovering in Evaluating.

One Point, One Test, One Number

1. Give each specification point its own row, its own test and its own target, so nothing gets tested twice and nothing gets missed.
2. The measurement is the thing you read off, such as mass in grams, time in seconds, a count of users, or a yes and no.
3. The method is the equipment and the procedure, such as a digital scale readable to 1 g, a stopwatch, or a 5 kg bag of sand.
4. The target is the line between pass and fail, such as under 800 g, under 8 seconds, or at least 8 of 10 users.
5. Add a sample size to anything involving people, because one friend saying it seems fine is not a result.
6. Repeat physical tests three times and record all three readings, since a single drop test tells you almost nothing.

Four Kinds Of Test And When Each One Fits

1. Measurement tests use an instrument and give a number, such as mass, length, current draw or page load time.
2. Performance tests load the product the way a user would, so stack 5 kg on the shelf, fold the hinge 100 times, or submit the form with an empty field.
3. User tests hand the product to real people with a task and a score sheet, such as five students assembling it while you time them.
4. Inspection tests are a checklist you tick by looking, covering things like no sharp edges, both brand colours present, and every label at least 12 pt.
5. Match the test to the point, because asking users to rate strength is a measurement test dressed up as an opinion.

Decide The Test Before You Build, Not After

1. Tests written after the product exists always drift towards whatever the product happens to do.
2. Finish a stool and then decide the criterion is "supports a 40 kg child", and you have described your stool rather than tested it.
3. Tests fixed in criterion B shape how you make, since knowing you will load the shelf with 5 kg changes the joint you choose later.
4. Many tests can be run on a card or printed prototype weeks before the final build, which is far cheaper than finding the failure at the end.
5. Put the whole testing plan in your design folder while the specification is still on the page, and date it.

Write The Method So Someone Else Could Run It

1. A method is complete when a substitute teacher could run it from the page without asking you anything.
2. Name the equipment and its precision, so "digital kitchen scale readable to 1 g" rather than "weigh it".
3. State the conditions, including the surface it sits on, the room temperature, the phone model used or the browser version.
4. State how many times you repeat it and what you do with the numbers, such as three runs recorded and a mean taken.
5. Give the exact wording of any question you ask users, since "is it comfortable?" and "rate the comfort from 1 to 5" produce completely different data.

Your Testing Plan Is A Table, Not A Paragraph

1. A table forces one line per point and makes gaps obvious at a glance.
2. Three columns are enough, being the specification point, the criterion with its target, and the method.
3. Leave a fourth column empty now and fill it with the real result when you reach Evaluating.