Exploring incentives: An Investigation of thousands of logs reveals exactly what most Actifitters aim for

If you don't know Actifit, it is an activity counter that rewards you in STEEM votes and AFITs (their own token) if you make at least 5000 daily movements. The maximum daily reward is reached at 10000 movements and every 1000 movements in between increases your reward.

I noticed a tendency in how I use Actifi. If by the end of the day, I am just below 10000 Activity Counts (AC) (or another whole number), I will see if I can get in some extra movements to get over the threshold. I wondered if I was alone and decided to investigate this empirically.

So what better way to do that than to look over a few thousands of Activity Reports. I managed to get 4507 before my browser crashed :)

Without visually looking at the data, I saw that the median AC was 10136 - conveniently just over the highest reward. The lowest AC was 1004 (1000 is the threshold for uploading) and some crazy actifitter made it to 85964. 50 % of the logs were between 6373 and 12678 AC, though and 97% had made less than 25000 AC. If we naively simulated data to fit these parameters, it should look roughly as follows:

This is how it would look if people just coincidentally happened to make around 10000 steps daily with a lot of individual variation. The bars represent the number of observations and the red line represents a smoothed distribution. As you can see the curve is roughly normally distributed with almost even variation to both sides of the mean.

However, if the incentives influence how people behave, there should be some bumps on the red line. I expected a bump to be at around 10000 with way fever observations just below 10000. Let's look at the data:

Wow, that was even more pronounced than I had expected. We see that there are 2 peaks: One at 5000 and one at 10000. It seems that most people either push themselves above 5000 Activities or do not bother to post their activity if they will not get a reward.

If we zoom in further, you can even see a boost at each thousand between 5 and 10k. It seems that if people have made, say, 7743 steps, they are likely to push it above 8000 before uploading their report.

So the incentives that Actifit is offering surely seem to influence the users' behaviour (I was not so special after all). An optimist would say that they are good in pushing people above 10k and a pessimist would maybe point out that most people seem to get much lazier after that. Maybe they could consider some incentives such as an extra chance to win a super upvote for each additional thousand steps logged?

In any case, this was a fun little example on how a question can be answered relatively quick if one knows the scientific method and is willing to put in a little work. It was also a relatively rare case in which no statistics were needed - the results are apparant to the naked eye.

I hope it was interesting to read for you too. If you have any comments or questions, I will be happy to hear them. Please check out @actifit at actifit.io if you haven't already, it has a very friendly community.