Skip to content
Curiotide

Plan a Backyard Science Investigation Worth Running

  • Ages 5-7
  • Ages 8-9
  • Ages 10-12
Illustration of a seedling with two leaves inside an arched window, on green waves

Most garden science is a demonstration: do this, watch it happen, agree it was interesting. An investigation has a question nobody in the house knows the answer to, and that changes how a child pays attention.

Ages 5 to 12

Time An hour to set up, two minutes a day for a fortnight

You'll need A notebook, whatever is in the garden, and a chatbot on your account

How it goes

  1. Start with something your child noticed. Why is the grass yellow there? Why are there more woodlice under that stone?
  2. Turn it into one changed thing. Same soil, same seeds, same water, different amount of light. Changing one variable is the whole method and it is worth naming.
  3. Ask for the design using the prompt below, then simplify it.
  4. Write the prediction down before starting. Your child’s, in their words, dated.
  5. Measure the same way every day and record it in the notebook.
  6. Compare against the prediction at the end, and say what you would change if you ran it again.

The prompt

My 8-year-old noticed that moss grows on one side of our garden wall and not the other. Turn that into a fair test she could run over two weeks using things in a normal garden. Say exactly what to change, what to keep the same, and what to measure. Keep it simple enough for a child to run alone.

What a good result looks like

A fair test names one variable to change, lists what stays the same, and gives a measurement a child can take: Change only the amount of shade. Keep the same wall, the same wetness, the same time of day. Measure by counting squares of a paper grid held against the moss.

If the design changes two things at once, or the measurement is “see how it grows”, send it back. Those two failures account for almost every unusable suggestion.

Check it before you use it

  • Is it actually fair? Models suggest tests with two variables changed. Read the design and check only one thing differs.
  • Is it measurable by a child? “Observe the growth” is not a measurement. “Count the squares on graph paper it covers” is.
  • Safety. Read any suggestion involving chemicals, heat, or wildlife handling before your child sees it.

Keep the human part

Going out to look, together, at the same time each day. The habit is the science, and no tool provides it.

If it is not working

If nothing changes over the fortnight, the timescale is wrong rather than the experiment. Moss is slow. Switch to something with a faster response, cress on a windowsill or condensation on a jar, and save the moss for a term-long observation.

If the daily measurement gets forgotten, tie it to something that already happens. Reading the gauge on the way to the bin works; remembering at four o’clock does not.

One thing to watch

When the result contradicts what the chatbot predicted, back the garden. That moment is worth more than any correct prediction, and our guide on teaching kids that AI can be wrong covers how to make it land.

Related: ask AI for an experiment, then actually test it is the kitchen version, and the backyard weather station is the long-running one. For quieter looking, design nature journal prompts your child will use.

Keep reading