so IDRlabs has released an alien stage personality test.
i decided to experiment with this test a little, and my conclusions are that this test is honestly not very good.
this personality test, provided by IDRLabs, asks you thirty questions. for each question, you are asked to mark how strongly you agree or disagree that a given statement accurately describes yourself; you are given the option to strongly disagree, disagree, remain neutral, agree, or strongly agree.
at the end of the 30 questions, the test returns 6 percent values describing what percent similarity you have with each character; the character with whom you have the highest similarity is your test result. the results are displayed as thus:
the test also provides a blurb describing the character with whom you have the highest similarities. the blurbs are provided below:
these blurbs were the first sign that all was not well in the state of denmark.
to determine how the results were determined from one's answers to the 30 questions, i decided to run some experiments.
experiment set 1 - understanding test structure
for my first set of experiments, i sought to understand the overall structure of the result calculation. i decided to observe what would happen if i varied my answer to only 1 question, and kept my answers to all other questions constant. I followed this procedure:
identify one relevant question to test.
for an agreement level n in the agreement levels {strongly disagree, disagree, neutral, agree, strongly agree}, repeat steps 3-5:
assign the relevant question to the agreement level n.
assign all other questions to be "neutral."
record the resulting percent similarities for each character.
i ran this procedure twice for 2 different relevant questions.
experiment set 2 - reverse engineering result assignments
for my second set of experiments, i wanted to see if i could reverse-engineer which questions were leading to which results. i followed the following procedure:
for each question, repeat steps 2-4:
mark the relevant question as Fully Agree.
mark all other questions as Neutral.
record the percent similarities with each character.
repeat this procedure for all 30 questions.
the first relevant question i chose was "I find it easy to move on from an unrequited love." the results are given below (the numbers in the character columns are the percent similarities given by the test):
the second relevant question i chose was "I defend my allies against any external threat." the results are given below:
the results of my second set of experiments, in which i attempted to reverse-engineer which questions led to which results, are given in the table below.
the results of each of my 30 iterations of the test fell into 1 of 2 categories:
percent similarity for 5 characters was 50%, and for 1 character was at 60%. (this character is marked in green on the table.)
percent similarity for 5 characters was 50%, and for 1 character was at 40%. (this character is marked in red on the table.)
based on the results of these experiments, i feel fairly confident in concluding the following:
each of the 30 questions is assigned to exactly 1 character, such that each character gets 5 questions. there are no questions that are assigned to more than 1 character.
each character specifically has: 4 positive questions, in which case answering that you agree increases your similarity, and 1 negative question, in which case answering that you agree decreases your similarity.
your percent similarity for each character starts at 50%. for each relevant question, going up the rungs of agreement starting from "neutral" causes your percent similarity to increase by 5%; going down the rungs of agreement starting from "neutral" causes your percent similarity to decrease by 5%. this is inverted for negative questions.
your percent similarity with a character is determined solely by how you answer the 5 questions assigned to them. therefore, your percent similarity with one character is completely unaffected by how you answered questions that were not assigned to this character.
each question's answer is weighted equally.
as even a cursory read through the results of the second set of experiments shared above will show, this test is Not Good. let's discuss why this is the case.
the main problem is that the test structure (described above) is fundamentally unsuited to the alien stage characters. alien stage functions thematically in part by drawing deliberate parallels between the characters; thus, many of the characters have many traits in common. however, the test is structured such that each question can only contribute to your similarity score with 1 character.
this results in illogical result calculations, in which questions that should obviously apply to more than 1 character are attributed to only 1 character. for instance, your answer to the question "I maintain a composed image to hide my insecurities" only affects your similarity score with SUA, even though there are also other characters this statement accurately describes. this also results in cases where the test seems to contradict itself: for instance, your answer to the question "I am open and expressive about my feelings" only affects your similarity score with MIZI and does not affect your similarity score with SUA, but your answer to the almost-identical question "I am comfortable sharing my feelings with everyone" only (negatively) affects your similarity score with SUA and does not affect your similarity score with MIZI; you can freely give contradictory answers to these two questions.
most egregiously, answering that you disagree with the statement "I find it easy to move on from unrequited love" only increases your percent similarity with TILL, and does not at all affect your percent similarity with IVAN. this is nuts. hello. hello.
on top of this, even separately from the problems introduced by the test structure, in many cases the test is simply incorrect. it is not incorrect merely because it is attributing to only one character a trait that should be attributed to multiple characters, but rather simply because it is incorrect. as a prime example, for some reason, answering that you agree with the statement "I avoid confrontation whenever it is possible" increases your percent similarity with TILL. this is deeply funny to me. TILL, the guy who keeps fighting the segyein to the point where they had to restrain him more severely than any other character? TILL, the guy who immediately started whaling on IVAN for stepping on his flower crown? this guy? really? in what planet, in what galaxy, in what universe?
also i'm pretty sure the art is AI-generated.