I’ve been practicing intermittent fasting since September 2015. At first, I experimented with a few different methods, but I quickly landed on the 16/8 method–which means that I fasted everyday for 16 hours and left myself an 8 hour window in which to eat. In reality, I often stretched my window to 10 hours for a late night snack but–with the rare exception for a special occasion–I completely stopped eating breakfast. I simply waited until lunchtime (between 11am and 1pm) to start eating everyday.
Prior to starting intermittent fasting, I had completed a “Biggest Loser” challenge at work in which I dropped 40 pounds. I did it pretty much by calorie restriction–eating only 1000 to 1500 calories per day for a few months. I decided to try my hand at fasting in order to maintain weight. Over time, it has worked. I gained back maybe 5-10 pounds and then plateaued at a solid 155.
Well, you know me. After a few years on the same old boring eating regiment, I had to try another experiment…
Although the research is still developing, fasting has proven to be an extremely helpful exercise for many in losing weight–despite the misconceptions some people have. Nevertheless, there has been some research indicating that certain methods of fasting don’t really work, because people simply overcompensate by eating more during the periods when they aren’t fasting. Anecdotally, it is for this reason that some have observed Muslims actually gaining weight during the month of Ramadan.
The Fasting Project
So, which is it? Does fasting actually help you eat less or does the increased calorie intake during non-fasting periods cancel out the amount you don’t eat while you’re fasting?
Well, science to the rescue! Or, at least a science-y like personal data project. Obviously, I can’t generalize this to anyone else but, at least for me, I think I’ve found an answer to the question. To determine whether fasting actually works, I alternated fasting each day at three different intervals for a period of 6 weeks:
- 14 days fasting 19 hours a day and only eating within a 5 hour window (skipped breakfast and lunch)
- 14 days fasting 14 hours a day and only eating within a 10 hour window (skipped breakfast but ate lunch and dinner)
- 14 days fasting 9 hours a day and only eating within a 15 hour window (didn’t fast at all–except when I was sleeping; ate breakfast, lunch, dinner, and any other time during the day that I felt like it)
To cancel out any effects of variation in eating based on the day of the week (maybe I would eat more on weekends?), I was sure to fast twice on each day of the week for each fasting method.
Also, I know there tends to be a bias in calorie counting. People tend to eat less when they’re paying attention to how much they’re eating. As they see the calories add up, we either explicitly or subconsciously start to refuse more food. To mitigate this effect, I did not count up my calories until the end of each day; I just kept track of all the food I ate in my notepad app and waited until the day ended to see how many calories it added up to.
Above, you can see a table showing a statistical summary of the project’s results. All in all, I consumed an average of 1300 calories on days when I skipped breakfast and lunch, 2000 calories on days when I skipped only breakfast, and a whopping 3000 calories on days when I did not fast at all. Interestingly, the average per hour calorie intake only increased when I skipped both breakfast and lunch. Just a guess, but I’m betting that’s because my body was already used to skipping breakfast and so I only felt compelled to eat more when I skipped lunch as well.
One thing I wanted to make sure of was that there wasn’t a huge difference in how much I consumed based on the day of the week in which I was eating. Looking at the bar graph above, nothing in particular jumps out. I ate the least on Saturday when I wasn’t fasting at all (15 hour window), on Wednesday when I skipped breakfast (10 hour window), and by a small margin on Thursday when I skipped both breakfast and lunch (5 hour window).
In the bar graph, you can also see the averages for each window pretty clearly–and there does seem to be a consistent difference. Above, you can see a scatter plot of my daily calorie intake as well. The different colored points and lines represent the daily intake and moving average trend, respectively, for each method. Just by looking at the scatter plot, it seems pretty clear that there are significant differences between the three fasting windows.
When looking at differences like this, though, you can’t just look at the averages; you have to look at the entire distributions. In each method of the summary table, you can see that the maximum of the method with the smaller window slightly exceeds the minimum of the method with the next larger window. So, the real question isn’t whether the averages are different but, rather, whether the entire distributions are different. That’s what a histogram is for. When you look at the distributions above, it seems fairly obvious that there are significant differences between each method.
To be sure, though, you’ve got to perform a statistical test. Above is the R output for an analysis of variance on the data set. Without going into an explanation of what the results mean, suffice it to say that there is indeed a difference between the three methods. But, how much of a difference is there between each method? There’s another statistical test for that, Tukey’s honest significant difference…
In this table, the “p_adj” shows that the relationship between each fasting method does have a statistically significant difference. The “difference” column shows how many more calories are consumed using the fasting method to the left side of the hyphen in the “methods” column vs the fasting method on the right side of the hyphen. For example, I on average consume 1,749 fewer calories on days when I skip breakfast and lunch than on days when I eat all day (don’t fast at all).
Another way to look at the distributions for each method is by using a box plot. The boxes represent the interquartile range (where 25-75% of the data fall) for each method. The line going through the middle of the box is the average. The lines extending from the boxes represent the extreme ends of the data–with the points being the outliers. This visual clearly shows the differences between the methods. But it also shows this: the larger the window I give myself, the more varied my calorie intake is.
During my project, I didn’t only track calories; I also tracked 5 other key nutrients. I’m not going to bother with doing the statistical tests but, just looking at the box plots, there seem to be significant differences in at least some of the methods for every nutrient. And, as for the carbs and the fat, the trend seems to be identical to that of the calories–with there being a clear difference between each method.
The real question about this project was whether or not I overeat on days that I’m not fasting to overcompensate for the days that I am. To test this, I looked at the “calories per hour” and compared them across the different methods. In this case, there only seems to be a difference on the days that I skip both breakfast and lunch. When I just skip breakfast (versus not fasting at all), I tend to consume the same amount of calories per hour and only eat less because I have a shorter window. When I skip breakfast and lunch, though, I consume about 50 more calories per hour. It’s still not enough to overcompensate for the amount I eat when I’m not fasting, but there is a difference.
You can see this difference more clearly in the box plot. When you look at my calories per hour, there is virtually no difference whatsoever in eating all day versus skipping breakfast. I don’t eat any more to compensate for missing breakfast. However, when I skip both breakfast and lunch, I eat significantly more calories to compensate for not eating all day.
That being said, the additional 50 calories per hour doesn’t really matter when I’m only eating within a 5 hour window. Take a look at the comparison of the box plots above. Eating the extra 50 calories per hour with the narrowest window adds up to eating about 1,250 calories versus 2,000 on days when I skip only breakfast and 3,000 calories on days when I don’t fast at all.
So, after all of that, the answer for me at least is a resounding: YES! Fasting works. And so, after 6 weeks of data collection and analysis, here’s my conclusion: I’m pretty much going to keep doing what I’m doing…