While writing all of my recent posts on Steemit, a thought kept occurring to me. It seems that no matter what content is provided, people who joined earlier are a higher level than those who joined later. Eventually after having a conversation about it with a friend, I decided to investigate. Are the high levels really high levels because they provide quality content? Or is Steemit just a large time game? Do we just wait it out until we become large and powerful?

Many of Steemit's high level users provide quality content, and deserve the followers and level they have, I am not denying that. However, the reoccurring theme of older users being a higher level was too obvious to ignore, and I decided to do some research.

Assumptions and Notes:

• All data was from taken from SteemDB which after 365 days have passed, uses 'moon' (month) measurements. I assumed that 13 moons was 390 days, and that 14 moons was 420 days.
• I assumed that all data on SteemDB was correct and accurate within a few days.
• I assume that every user I plotted on this graph and included in this research was a real person. I selected all the users on this graph using a random method. I took all users (the first 60) from the 'new' page of Steemit and searched their SteemDB. Any user that appeared to not be real (a bot) was removed. However, there is the possibility some of these data points could be bots.
• I assumed that all users only had positive reputation growth and had not lost reputation due to down voting. Any negative level users, or users with posts that were flagged were removed. I only want to look at users with growth, as flagging and down voting is unrelated to time anyway. Once again though, due to the fact that all work was down by myself, there is the possibility or errors.
• There are no double ups of data. All data represents a separate and unique profile.
• I used a data set of 60 users so as to balance time and accuracy. If I could be provided with a bigger data set, I would be more than willing to do this research again. However, due to the time constraints I have settled on 60 users which I believe to be a fair and representative amount of the population.

So where are the results? After plotting the top 60 (unique and non-bot) users in the 'new' tab (20 at a time at 3 seperate times) using SteemDB here are the initial results.

At initial glance, there is most definitely a correlation between days on Steemit and reputation level. However, with this small data set, data is rather scattered. Besides one obvious outlier (level 25 at 420 days, the dot above the title is not a data point), the data does seem to follow a positive trend.

Adding a linear trend line allows us the see the trend in the data. However, as seen in the image below, the coefficient of determination (R^2) is not very strong, showing a weak relationship between the two sets of data. The R squared value will always be between 0-1 with 0 showing no relationship and 1 being a perfect and absolute relationship. The value provided here (0.5878) shows a relationship exists, but a weak one.

Something else to consider is outliers, which especially at this data size, can skew the data and cause inaccuracies. To account for this and see the differences, I removed the most obvious outliers from the graph.

Removing one outlier greatly changes the results! The R squared value rose from 0.5878 to 0.7352. Is there a relationship between time spent on Steemit and level? I think so! (maybe that was obvious though...)

Removing two more rather unrealistic pieces of data (level 70 at around 100 days and level 25 after around 60 days), shows an even more obvious correlation.

While outliers will always exist and we have to incorporate them into our data, at this small data set they can disproportionately affect the data, and therefore, the results. For this reason I decided to remove the most obvious outlier and these two other pieces of data that seem unrealistic for most users. While one can argue that these pieces of data should have remained, I simply wanted to show the affect of removing them from the data set... After all, a bigger data set would be far less affected by these outliers.

In Conclusion:

While it seems obvious that people who have been using Steemit the longest would be the highest levels, in my mind it also shows that Steemit is definitely not all about quality content. It's about when you signed up, and how long you have put into the site. While this may have seemed obvious to some, I wanted to go into a bit more detail and have a look.

This research was biased by the fact that every user was under the 'new' tab, meaning that they (if they are a high level) probably regularly post, and it doesn't account for less active users. However, I did not want to include inactive users in my data as this would have provided even more biased results.

Another possible factor that may greatly have affected these results, is the hard fork change. I am considering doing a report of user who joined only after the change, but I am curious if anyone is interested to see such an analysis.

If anyone is able to provide me with, or show me where a bigger data set is, I would be more than happy to analyse that too. I am curious how the data and trend would change when more users are added to the mix. However, in the meantime, I think this post shows a definite correlation between Steemit reputation level and time.

Do the whales really deserve the power they have? That's my question for everyone today.