This is a hard to answer question, as there isn’t any statistically useful database. CALS has a bunch of data, but only a small subset of conlangers have ever entered their data there. There, there seems to be a clear preference for the creation of Artlangs, i.e. creating languages simply for the sake of creating art, not to fulfill any purpose. A priori langs seem to be favoured over a posteriori, likely because the latter is quite a bit more work-intensive (every new vocabulary item needs to be researched if one wants to do a good job at it…).
A larger sample size can be found on reddit. However, there are no statistics there, and the sample is skewed in favour of beginner conlangers (only few experienced conlangers ever frequent it. source: personal experience as a moderator). To get an idea of the data, we can use Google search results. With the added restriction site:reddit.com/r/conlangs
here are some results (first number total of following search results):
Auxlangs: 893
- Auxlang: 469
- Auxiliary Language: 212
- IAL: 212
Artlangs: 430 (this number is problematic because this category appears to be taken as the default by many, yielding no search results)
- Artlang: 191
- Artistic Language: 239
A priori: 1'170
A posteriori: 328
Naturalistic: 363
“Romance” languages: 1'403
- romlang: 363
- romance: 1'040
“Germanic” languages: 1'210
Indo-European: 730
Creole: 364
As you can see, on reddit, taking inspiration from existing European language families is very common. From personal experience, among more experienced conlangers there are very few people trying to create auxiliary languages, most people striving to either create naturalistic languages or to push the boundaries of linguistics in some way or another. But I cannot provide any data for this.