How to Use Self-Experiments to Live Your Best Life
DIY investigations with three easy starter experiments you can start right now
As a kid, I was always fascinated by the mad scientist archetype. In movies and comic books, the mad scientist works in his lab, developing an experimental super-soldier serum. Eventually, he decides he’s got it right, gets impatient to test it out, and injects it into himself. And lo and behold, he becomes superhuman!
Well, usually a super-villain, if we’re being honest. Ahem.
As a kid, I always wanted to do that — the superhuman part, not the villain part. And then as an adult, I learned that I can — minus the needles, experimental steroids, and insanity. Running experiments on yourself, though? That’s totally possible — and done right, it can be life-changing.
Unlike traditional scientific experiments, self-experiments don’t have a large number of research subjects. They just have one: you. That means they aren’t very useful for finding out what works for people in general, but they’re great for finding out what works on you, specifically.
In fact, I’ve run over twenty self-experiments lasting one to four weeks each, and many of them have produced massive improvements in my life and my health. If you want to learn how to do this for yourself — or copy a few of my most successful experiments — read on.
How to Design Your Own Self-Experiments
Step one: Formulate your experimental question
This is essentially the one-sentence summary of your experiment, or what the title would be if it was a professional experiment published in a scientific journal.
A few examples:
- How do I feel after eating different breakfasts?
- Which workouts make it easiest to fall asleep later that night?
- Which workouts are the least fatiguing?
- Which meals give me the most sustained energy level afterward?
- Under what working conditions am I the most productive?
Notice that each of these titles identify the independent variable — the “input” of the experiment that’s being manipulated — and the dependent variable — that is, the output or result that’s being gauged.
However, the variables are not yet clearly defined. That comes in the next two steps.
Also note that they are all phrased self-referentially — this is a self-experiment, so your goal is to find out what works for you, not what works for people in general.
Step two: Define your evaluation criteria
Next, you’ll select and clearly define one or more dependent variables of the experiment — the results you’re watching for.
These may have been mentioned, though not clearly defined, in step one. In step two, you’ll identify your dependent variables and come up with an operational definition for each one.
What is an operational definition? Unlike a dictionary definition, it’s what a given term means for the sole purpose of this experiment. A good operational definition should be clear and specific — it should be absolutely crystal clear what it refers to.
It should also be clearly measurable — the operational definition should include how each variable is to be measured.
In the first example from step one — How do I feel after eating different breakfasts? The dependent variable(s) would be how you feel. While this could be a single variable, it makes more sense to divide “how you feel” into at least two separate constructs — mood and energy level. Each of those can be affected by what you eat, and experiences tells us that they don’t always track with each other.
So your two dependent variables are mood and energy level — but now, how do you measure them? Since they’re totally subjective, you’d probably want to use a self-reported five-point scale, with five being highest, or best.
You also need to define when this self-assessment will be made. Let’s say you do it not just once, but every hour after breakfast until lunch.
There– now you have clearly defined criteria for evaluating the success of your experimental conditions.
Step three: Select your experimental conditions
Next up you’ll define your experimental conditions — the independent variable(s), or combinations thereof, that will be manipulated and tested to see their effects on the dependent variables.
The rules here are largely the same as in step two: the dependent variables, or experimental conditions, must be as specific and measurable as possible. However, since you have to manipulate them, there is an additional requirement: they must be controllable.
As an example, what you eat for breakfast can be precisely controlled. How many hours and minutes you sleep at night cannot be precisely controlled, although most people, with discipline, can at least control it to within an hour or so.
In the above example of testing the effects of different breakfasts on your mood and energy level, the different experimental conditions would be different meals. Low-carb, vegan, high-carb, a protein shake, and no breakfast at all would all be viable experimental conditions here.
Step four: Control for confounding factors
Confounding factors are things that are not independent variables — you’re not trying to manipulate them — but that nonetheless can affect the outcome of the experiment.
In a traditional scientific experiment, the use of multiple subjects reduces the influence of these confounding factors. That’s one reason why having a large sample size is so helpful. But in a self-experiment, with only one subject, any major confounding factor can ruin the whole enterprise.
Therefore, you need to control these confounding factors — keep them constant — as much as possible. And barring that, you may need to selectively pause the self-experiment when a confounding factor is clearly going to disrupt it.
As an example, consider the above breakfast experiment. The most obvious confounding factors here are how well you slept the night before and what you’re doing in the morning up until lunch.
During this experiment, you’ll want to control those as much as you can — maintain a regular sleep schedule and keep doing the same things between morning and lunch. You probably want to run this experiment only on weekdays, since you’ll naturally sleep differently, and be doing different things, on weekend mornings.
And finally, if something unexpectedly causes a night of terrible sleep, or if you know you’ll have a very unusual morning — perhaps you have a flight, rather than working at your desk as usual — you may want to skip the experiment for that day, knowing that a major confounding factor would invalidate your results for that day.
Three Self-Experiments for You to Try
Here are three of the experiments that have made the biggest impact on my life, as well as the lives of my readers and clients. Each can be performed in one to three weeks.
Experiment one: The Breakfast Test
This is the one I just used in my examples. Let’s see how this experiment looks when you put it all together.
Experimental question: How do different breakfasts affect my mood and energy level throughout the morning?
Dependent variables: Mood and energy level. Each of these will be self-scored on a five-point scale, every hour starting after breakfast and up until lunch.
Since this also lets you see how much your mood and energy levels fluctuate throughout the morning, you could potentially add two secondary measures, mood and energy stability, if you wanted to.
Experimental conditions: Different breakfasts, all eaten at the same time of day, on weekdays. A few likely options:
- Continental breakfast — toast, jam, coffee, fruit, and maybe a hard-boiled egg
- Low-carb — eggs, sausage, bacon, vegetables
- High-carb — pancakes, toast, fruit, and a few eggs or low-fat turkey sausages
- Standard American breakfast — eggs, bacon, toast or pancakes, hash browns, coffee
- No breakfast — Just a cup or two of tea. Start self-reporting an hour after whenever you normally eat breakfast.
- Vegan — Tofu, salad, fruit
Obviously there are a tremendous number of possible variations here since you have several “levers” to manipulate: the overall size of the breakfast, protein content, carbs vs. fat, degree of processing, and caffeine content, to name just a few.
Confounding factors: Sleep, what you do in the morning, and where you eat.
This is why you only want to run this experiment on weekdays — weekends will naturally be different. You’ll also want to control for sleep as best you can by maintaining a regular sleep schedule, and if you suffer from insomnia, you should probably improve your sleep hygiene before doing this experiment.
You’ll also want a regular morning routine, like every morning you sit at your desk and work on… whatever it is you work on. But the same sort of stuff every morning to minimize variety.
And for the sake of the experiment, you’ll want to make all of these breakfasts at home so you can know exactly what’s in them, and keep your eating conditions consistent.
If anything happens that would unavoidably confound the experiment — a flight leaves you jet-lagged, or you have meetings all morning instead of working at your desk as usual — you should probably just skip the experiment for that day, knowing that day’s results will be tainted.
One last potential confounder to be aware of is caffeine. I mentioned that as a possible independent variable, but it doesn’t have to be. If you have no interest in testing your caffeine intake, you can instead keep it constant and have, say, one cup of coffee every morning.
It has to be one or the other though. You can either manipulate caffeine intake as an independent variable or control it as a confounding variable, but the one thing you absolutely cannot do is simply not think about it, because it has a huge effect on your mood and energy level.
Experiment two: Working conditions
In this experiment, you’ll test different working conditions to see the effect they have on your productivity.
Experimental question: Under which working conditions am I the most productive?
Dependent variable: Your productivity. Measuring this can be tricky, and has to be specific to the type of work you do.
If there’s a straightforward way to quantify your output — lines of code written, words written, a number of sales made — do that. You could argue that numbers like these aren’t the best way to measure productivity, but they’re at least objective, which counts for a lot.
If you do something where your output is harder to measure, you may need to use a more subjective measure for your productivity. In that case, you’ll want to keep a to-do list for every day and start journaling your activities for the day so you can look back and see how fast you got everything done, and how well you stayed on-task.
You’d then rate your productivity for the day on a five-point scale, as you did for your mood and energy level in the breakfast test.
The difficulty here is that since you’re using a subjective measure, your own biases can prevent you from being honest with yourself. In this case, the best solution is to enlist an accountability partner. Have that person help you set your to-do list every day, and then talk to them after work every day about how productive you were that day.
And then, have them rate you, rather than doing it yourself. This also allows you to “blind” the experiment by not telling your partner what working conditions (the independent variables, discussed next) you were using that day.
Independent variables: Standing vs. sitting, lighting, sound, and maybe a few others.
Standing vs. sitting is a huge one, and I highly recommend everyone tests it. Most people find they work faster when standing, although in practice you will need to alternate standing and sitting throughout the day.
Lighting is also worth looking into — a very dark environment is almost always bad, but moderate vs. very bright light is an open question. This variable isn’t always easy to manipulate, but you could always adjust the brightness of your computer screen.
As for sound, outside noises are obviously distracting, but what about music or ambient noise? Some people work best in absolute quiet, while others are more productive when listening to a single song on an endless loop. This is worth looking into.
If you have a choice about where you work — from home vs. in the office, or at home vs. in a coffee shop — absolutely experiment with that as well.
Other factors specific to your line of work may be relevant. Monitor size, or using a two-monitor set-up, is often worth playing around with for people who have to switch between multiple screens and tabs a lot, like programmers. In fact, there are quite a few studies showing that bigger monitors, or having two screens, improves productivity for such people.
Note that the combination of these factors produces a tremendous number of possible combinations. This experiment may well need to run for a whole month.
Confounding variables: These will mainly have to do with the type of work you’re doing, and, again, with how well you slept the night before. And also possibly what you ate beforehand.
You’ll want to eat more or less the same meal every time before doing this test — so the same breakfast, if you run the test in the morning, the same lunch if you run it in the afternoon, or both if you run it all day. As with the breakfast test, you might skip days where conditions can’t be held constant.
This also requires you to be consistently doing the same kinds of work when you run the test. If your job has a lot of variety to it, you might choose to only run this test on days when you’re doing that particular type of work.
Experiment three: Evening routines
In this experiment, you’ll test various evening routines to see which one helps you to sleep the best.
Experimental question: How should I spend the last two hours before bed in order to get the best night’s sleep?
Dependent variables: Sleep duration and your energy level the next morning. Possibly also sleep quality, as measured by time spent in deep and REM sleep vs. light sleep, if you have a way to measure that.
Your energy level would be a five-point subjective scale again. Sleep duration could be estimated, but this experiment really works best if you have a Fitbit or something like that to monitor sleep. A free app like Sleep Cycle could also work.
And of course, sleep quality can only be measured with a fitness tracker like FitBit — some phone apps claim to measure it, but they’re wildly inaccurate.
Experimental conditions: Various evening routines, each of which covers the last two hours before bed. A few possibilities:
- Shower, brush teeth, watch TV for an hour, then read for a half-hour.
- Eat a light snack, do yoga, then read.
- Work for an hour, then read for an hour.
- Watch a movie before bed.
- Work out at the gym, come home, shower and change, have a snack while reading or watching TV (pick one), brush your teeth, then get to bed.
There are nearly endless variations here, but they should all cover the same s=time period before bed. And of course, they all need to include the necessities, like getting changed and brushing your teeth, although the order in which those happen can change.
Confounding variables: Sleep schedule, caffeine, and stress.
You should be going to bed and waking up at consistent times during this experiment.
Caffeine intake should be controlled — both how much you have, and at what time of day. 1–2 cups of coffee a day, always before lunch, is a good example. That provides a little flexibility without being likely to majorly confound the experiment.
Stress is harder to control, and might unavoidably confound your experiment. On the other hand, if stress is often a problem, then finding ways to deal with it should actually be considered part of the purpose of this experiment. In that case, many of the evening routines you try should include stress-fighting habits like exercise, yoga, or meditation.
Live the Experimental Life
I suggest starting with these three experiments since they tend to produce big wins for almost everyone. After that, start brainstorming experiments of your own.
Do protein bars give you more energy than protein shakes? Does putting cinnamon on your food keep you full longer? Are you happier when you play with your pet for fifteen minutes before going to work? When you allow yourself to get creative, not only can you dramatically upgrade your health, productivity, and lifestyle, but the process of doing so can be very fun.
By turning your life into a series of one to two-week experiments, you can greatly improve your quality of life in a short period of time. Start with the obvious big wins, like what you eat for breakfast and how you start and end your day — then find other areas of your life you think could be improved and start experimenting with them.
Do you have an idea for a great self-experiment? Share it with your fellow readers in the comments.