I've read a few people mention a link between Te and unseeliness. So I decided to take a look at the database to see just how much of a link there was. If you want to see all the numbers, I've shared the link to the google sheet where I recorded them at the bottom of this post. Here's what I found:
Fi-leads are more likely to be seelie (71%) than unseelie (29%). It is not clear from the numbers whether or how much this is affected by whether or not Te is conscious.
Among those who have Fi as their second function, NeFi's are slightly more likely to be seelie (59%) than unseelie, while SeFi's are significantly less likely to be seelie (17%) than unseelie. Here are the development types listed in order of most to least likely to be seelie: Fi conscious and Te unconscious, both Fi and Te conscious, neither Fi nor Te conscious, Te conscious and Fi unconscious
Among those who have Te as their second function, SiTe's are about 50/50 seelie and unseelie, while NiTe's appear to be quite unlikely to be seelie given that of the 23 in the database, none are seelie. SiTe's have increased rates of seeliness when Fi is conscious and Te is unconscious, a very small increase in the rate when neither is conscious, and a decreased rate when Te is conscious but Fi is unconscious (no data on having both conscious).
Te-leads are more likely to be unseelie (79%) than seelie. Having Fi conscious significantly increased rates of seeliness (from 21% to 53%), while not having Fi conscious slightly decreased rates of seeliness (from 21% to 15%).
So, this data is definitely consistent with the idea that Fi is linked to seeliness and Te is linked to unseeliness. A few disclaimers should be made though. First of all, I am not a statistician and I haven't done any calculations on the statistical significance of these numbers. In some cases the number of cases I was working with were pretty small, which makes the numbers less reliable. Also, the database is not really a representative sample because of the various ways people are chosen to add to the database.
Here are my questions at this point:

• Is it possible that this link is being caused by the method of typing? Could there be some unseen overlap in the signals used for typing seelie/unseelie and Fi/Te that would result in these correlations?
• If it's not an artifact of the typing method, what is going on psychologically to cause these correlations?
• What is the effect of the perception axis? I was focusing on Te and Fi, but it also looks like those on the Se/Ni axis are more likely to be unseelie than those on the Si/Ne axis.
• What is the data for adaptive vs directive and Ti/Fe?

Here's the data:
https://docs.google.com/spreadsheets/d/1E1bYpL5GKkC3H80Jd3DF6NMaNcGxQP93xfn7LJ12j20/edit?usp=sharing