But It Worked for Me: The Value of Self-Experimentation
Change your behavior the way a scientist would.
Posted August 9, 2016
We all have patterns of behavior we'd like to change. But whether it's smoking less or not at all, paying bills on time, not biting fingernails, eliminating that morning doughnut, chatting more with strangers, or taking a spin class, it can be tough to do something new.
If you have a problem specific to you (which reminds me—I should water my houseplants), you won't find much help from researchers. They usually study phenomena that affect groups of people, collecting data with standardized instruments, generating statistics that summarize the characteristics of a sample, and drawing conclusions about populations that may or may not have much in common with you.
Suppose, for example, you come across research on the relation of caffeine intake to sleep quality, and the results indicate that people who drink beverages with caffeine after 1:00 p.m. sleep, on average, 0.72 hours less each night than people who don't. (This is merely a made-up example—please don't tell people you learned this "fact" from me.) If you're not sleeping enough at night and you're a big-time coffee drinker, you might be thrilled to come across this research. "Maybe I should cut out caffeinated beverages before 1:00 p.m.," you think.
But what if your sleep behavior seems to be unrelated to the timing of caffeine consumption? You think back to those occasions in which you had coffee after dinner and slept like a baby. Or perhaps you had no coffee one day and spent the evening in bed staring at the ceiling. And what if you just love coffee so much that you'd rather try something else first? You might also acknowledge that your problem is sleep quality rather than uninterrupted sleep and wonder whether this study still applies to you.
Suppose you look for other explanations—the cop shows you watch on TV before bed, the snoring spouse, the LEDs on the clock radio, the bedtime snack, the cat that terrorizes you at 3:00 a.m. You might be disappointed if you look for empirical research on these potential predictors of sleep quality, because, chances are, there isn't any.
Instead of trying this or that to correct the problem, consider taking a systematic approach. Haphazard problem solving often brings more frustration and confusion than answers. Scientists use experimental designs to understand cause and effect relationships. You can use a single-subject (yourself) experimental design to do the same thing.
Single-case research has been unfairly dismissed by researchers and the public alike. Controlled manipulations are used by many professionals to influence individual behavior in a desirable way. I use quantitative assessments to evaluate the effectiveness of treatment for individuals in my clinical psychology practice. Animal learning researchers demonstrate clear and strong effects with n=1 designs. Athletes and fitness enthusiasts systematically manipulate training and dietary variables while tracking their responses. Self-experimentation doesn't have to mean anecdotal and biased. And concerns about sample size and generalizability are irrelevant when you're interested in the responses of an individual.
You don't have to approach self-experimentation with the same degree of rigor used by professional researchers, but if you really want to bring some clarity to your understanding of cause and effect relationships in your own life, here are some ways to get started:
- Specify your target behavior in objective, measurable terms. Is it clearly defined? Example: "Sleep quality" is subjective, but you won't have to rely on guesswork to record "minutes of uninterrupted sleep."
- Prepare to record your data. Create a spreadsheet or use a notebook to monitor behavior patterns.
- Measure the frequency of the target behavior before you try to change anything. This is your "baseline" phase. A few days or a week might be long enough, depending on the behavior.
- Generate a testable hypothesis about the cause of this behavior. What do you believe will directly influence it? What is your intervention? How will you administer it, and for how long? Is it clearly defined? If not, refine your idea.
- Implement your intervention and record your data. Don't trust yourself to be objective or consistent? Ask someone else to help.
- Examine your data. Graphs are helpful. Did you observe a meaningful change between your baseline and intervention phases? If so, nice work! If not, repeat the aforementioned steps with a new intervention.
For an excellent academic argument in favor of single-case research in psychology, please read this wonderful article by Dr. Matthew Normand.