An experiment, properly done, is just a structured way of finding out whether changing one thing changes another thing.
You do not need a lab, a control group, or statistical significance to run a useful personal experiment. You need a clear question, a deliberate change, enough time to observe the outcome, and the honesty to accept whatever you find.
The Difference Between Noticing and Knowing
Most people who suspect their environment affects how they feel are in the noticing stage. They have observed something, perhaps several times, but they have not systematically tested whether a specific change makes a difference.
An experiment moves you from noticing to something closer to knowing. It does not give you certainty, but it gives you a better basis for decision-making than an impression alone.
How to Structure the Experiment
Choose one variable. The most common failure mode in personal experiments is changing several things at once and being unable to attribute any change to a specific cause. Pick the single thing you most want to test.
If you suspect the router in the bedroom is affecting your sleep, test the router. Move it out of the bedroom (or set it to turn off overnight) and leave everything else the same. Worth knowing going in: the World Health Organization has found that typical router exposure sits thousands of times below international safety limits, the kind of limits set by bodies like ICNIRP, so a change in how you feel is more likely tied to sleep hygiene factors like light and habit than to the signal itself.
Establish a baseline first. Before making any change, track your relevant outcomes for one to two weeks without changing anything. This gives you a real comparison point.
Make the change and hold it. Run the changed condition for at least two weeks. One or two good nights after a change are not meaningful. A sustained shift is.
Record throughout. Keep a simple daily log before, during, and after the change. Sleep quality, energy on waking, any physical observations. Without a record, memory will fill in gaps in ways that tend to confirm what you expected.
Evaluate honestly. At the end of the experiment, look at the record and ask: is there a measurable, consistent difference? Or not?
What Counts as a Result
A clear positive: the changed condition consistently produces a measurable improvement over the baseline.
A clear negative: there is no meaningful difference between the two conditions.
An ambiguous result: some improvement but inconsistent, or hard to distinguish from normal variation. This is common. It may mean the variable you tested is a minor factor, or that the experiment was too short, or that other variables are obscuring the signal.
All three of these are useful results.
What to Test First
The most commonly tested and most commonly productive variables:
- Router moved out of the bedroom or turned off overnight
- Phone removed from the bedroom
- All devices in the bedroom unplugged overnight (not just on standby)
- Work done in a different location (different room, or outside the home)
- One day per week with significantly reduced device use
Start with the variable that is easiest to change and most likely to be a factor based on your pattern observations.
Frequently Asked Questions
You will not get certainty, but your log is the best evidence you have. A placebo effect tends to be strongest right after a change and fade within a couple of weeks; an effect that is still holding up in week three or four of your log is more likely to reflect something real. Comparing your baseline numbers to the changed-condition numbers side by side, rather than relying on how convinced you feel, is what actually separates the two.
Test that too. If reversing the change reverses the outcome, that is stronger evidence that the variable mattered. If it does not reverse, the initial improvement may have had another cause.
After completing a test of one variable, yes. Not simultaneously. Sequential experiments are harder to sustain but give you much clearer information.
If the experiment shows a consistent improvement with a specific change, keep that change in place. If it shows nothing, rule that variable out and move to the next most likely factor. Either way, you know more than you did before.