Rick's New Socionic Test Online

[Slacker]

I was worried about those questions being too confusing. Basically, you're given two jobs and need to choose which one of them you would better be able to tolerate. I wanted to give people differentiated choices instead of just one or the other.

Ok so it's between 1 and 2, and then between 3 and 4, etc? But I have to have an answer under both 1 and 2, and an answer under both 3 and 4. Or do I just put "much better" under #2, and not put anything under #1?

[Slacker]

Ok I think I get it but I think they should be number 1a and 1b instead of 1 and 2, because the directions say "answer each question" which makes me think I have to answer both 1 and 2.

IEIYour resultIEEThese types might also be consideredSEISEEEIIThese types are not very likely EIEESIESEILIThese types are quite unlikely ILESLISLELIIthese types are extremely unlikely LIELSILSE

[Elro]

Ok so it's between 1 and 2, and then between 3 and 4, etc? But I have to have an answer under both 1 and 2, and an answer under both 3 and 4. Or do I just put "much better" under #2, and not put anything under #1?

It's the latter case - if you try putting one in both, whichever answer you select last will stay and the other answer will disappear. I only figured this out the second time I took it. (Or maybe it only worked the second time). So I guess just consider 1-2 one question, with a continuum that bends like a horseshoe.

[Creepy-Cyclops]

I think the questions in the work related stage are better this way. I think it leaves less chance of people second guessing the answers to give the answers they would like rather than what they are.

[Exodus]

What was your result, by the way?

I got ILI, but that was before the fix. Post-fix:

Your result
IEE
These types might also be considered
IEI
ILE
ILI
These types are not very likely
EIE
EII
LIE
LII
These types are quite unlikely
SEE
SEI
SLE
SLI
these types are extremely unlikely
ESE
ESI
LSE
LSI

Weird, because my answers on the ego element questions, at least, were very NT.

Is the question "I like to understand the status of my relationships and to think and talk about what can be done to improve them." supposed to be ethics/logic or rational/irrational? And what are the questions about intentions vs. actions supposed to mean?

I think the ego element questions were the best by far. The job ones were difficult for me to answer - none of the alternatives (except for the intuitive one) were appealing.

[Park]

Ok, here's my 2nd try: (1st time I got ILE)

Your result : ILI
These types might also be considered : LII ILE LIE
These types are not very likely : SLI IEI LSI EII
These types are quite unlikely : SLE IEE EIE LSE
These types are extremely unlikely : SEI ESI SEE ESE


Wow, I didn't know I was so "intuitive".

[Cuddly McFluffles]

*Still gets LIE*

I really want to swear at the test (the test won't talk back or get angry), but I wonder whether it's faulty calculation/questions or me.

[xyz]

I got SLE


SLE Your result

ILE These types might also be considered
LSE
LIE

SLI These types are not very likely
SEE
ILI
IEE

LSI These types are quite unlikely
ESE
LII
EIE

SEI these types are extremely unlikely
IEI
ESI
EII

[xyz]

Does this mean Herzy and I are BOTH DA SAME PERSON?!

[xyz]

ZO MA GAWD YOU'RE MAI DEAD RINGER LOLZ!!

ZOIKS!!!

Ok, here's my 2nd try: (1st time I got ILE)

Your result : ILI
These types might also be considered : LII ILE LIE
These types are not very likely : SLI IEI LSI EII
These types are quite unlikely : SLE IEE EIE LSE
These types are extremely unlikely : SEI ESI SEE ESE


Wow, I didn't know I was so "intuitive".

I like how both Winterpark and I got SLI as "Not very likely" ha.

[Jonathan]

Home Visie Werkwijze Mizami Match Key™ Contact Forum
Test results
ILE Your result
IEE These types might also be considered
ILI
LIE
IEI These types are not very likely
LII
EIE
SLE
EII These types are quite unlikely
SEE
SLI
LSE
SEI these types are extremely unlikely
LSI
ESE
ESI

[Slacker]

ha, yes I do!

something must be wrong with these tests then, because I'm no IEE

Yeah and it said I'm IEI and I'm quite sure I'm IEE.

[xyz]

Someone switch with me being a SLE is tiring

[hellothere]

Someone switch with me being a SLE is tiring

you can take my IEI, switching with your dual will be like switching bodies with a member of the opposite sex (if that appeals to you)

[Ms. Kensington]

tell me anyone, HOW THE HELL CAN I BE CONFIDENT IN THE TEST RESULT LOL? THIS TEST DOES NOT PROVIDE ANY MEASURE OF CONFIDENCE PEOPLE CAN HAVE IN THESE RESULTS AND THIS F* SUCKS.

you really remind me of misutii (in general)

[munenori2]

tell me anyone, HOW THE HELL CAN I BE CONFIDENT IN THE TEST RESULT LOL? THIS TEST DOES NOT PROVIDE ANY MEASURE OF CONFIDENCE PEOPLE CAN HAVE IN THESE RESULTS AND THIS F* SUCKS.

Geez, dee. It's just a test. Hell, it's just a test of a test. It seems like it's got some merit. Not everyone is getting their self-perceived types as the main result, but it seems to fall into either that or their considered for the most part. Nothing's perfect and this seems quite a bit better than most of the things we've been occupying ourselves with. I mean I can't even figure out what would satisfy your complaint of the confidence level on this thing. Take it all with a grain of salt. It's a work in progress.

[Ms. Kensington]

well, o... k...? so what.......

i couldnt find your type thread, sorry

[hellothere]

i mean if you don't know what's going on inside the test, how can you be sure in it, they might as well just spit random numbers and that's it! like say check out my excel sheet tests and excel is a standard computer program lol, you can see the algorithm, you can mess with it, do anything you wan. i mean COME THE FUCK ON!!! what's the point of posting a test if people can't believe the firkin results!!.........

you mean, you want it to be scrutable? that is a reasonable request, but at the same time there is the disadvantage that being able to see how the result is obtained may influence how you answer the questions.

[Rick]

i know what you're talking about, but i think there are other ways other than fucking up your test potential by not surrendering the algorithm.

"Surrender the algorithm?" Do you even know what you are talking about?
Since when is providing all personal socionics work for your scrutiny required to have "credibility" at this site?
Nobody here is interested in the "algorithm" but you, and I have no idea what your intentions are for digging into the test mechanism, so why would I want to give you a 38 page document explaining how to create the test? What are the chances you would use it to help me improve my test? I think about zero. It would just give you another reason to go on ranting and raving for another ten pages in this thread or copy it and make ten new tests of your own, but worse.

Here is a sample excerpt of the "algorithm."

(...removed...)

[munenori2]

i know what you're talking about, but i think there are other ways other than fucking up your test potential by not surrendering the algorithm.

Maybe I'm misunderstanding what you're saying here, but if you know what the answers will lead to, then you can get whatever result you desire (what hellothere said). The test is supposed to tell you something you don't know, rather than simply confirm your own suspicions. If you're that interested in the calculations of the test, if that's what you need to even begin to contemplate that it has any validity whatsoever, then send Rick an e-mail and I'm sure he'd be happy to privately discuss the issue with you.

Saying something along the lines of "I don't know how the results are calculated, therefore how can anyone have a mote of confidence in it" stings of questioning Rick's credentials or ability which, while the question itself may be legitimate depending on the situation, is pretty tactless to go nuts about on the forum in this way. Perhaps you're not being all that serious in what you're saying. I don't know, but if someone said these things about one of your tests (and maybe they have at some point) I do not think you'd take that kindly to it. All I'm asking for here is a little courtesy. I'm not even saying that I've represented your views accurately here. What I am trying to get across is that Rick asked for serious and legitimate feedback on his test, just as you do with yours.

[Rick]

...

Yes, I agree (but you said it with much more Fi )

I've already explained enough about the algorithm as is necessary for feedback in this thread.

[munenori2]

Yes, I agree (but you said it with much more Fi )

I've already explained enough about the algorithm as is necessary for feedback in this thread.

Yes, you have.

[Rick]

fair enough.

@ Rick: the reason i ask for algorithm is since if people UNDERSTAND it, i.e. get to know that it really should work, if it's a good algorithm, they have better confidence their answers have not gone to waste.

as for test-making and biased answers, i mean if a question is straight forward enough and you Can actually provide an honest answer, then i don't see how this factor should count at all.

of course even at the best socionics test you can just play with the buttons.

Confidence in a test comes from getting accurate results. That's why this thread exists. It looks like it's got about 30% accuracy now, so there is more work to be done. I can't see how giving everyone a 38 page description of how to write the program for the test would help me make the test more accurate. I've already explained the basic idea behind the test, and I don't see how talking about that more would help me improve the test.

[Rick]

do you want to discuss your algorithm or not?

Not at the moment. It is very straightforward, so it's not the algorithm that needs to be modified, but the questions, especially if there are questions that everyone is answering similarly regardless of their type.

[hellothere]

hey Rick, have you thought about putting a question on the first page asking for what the test-taker believes to be their type (with an 'I don't know' option of course)? I've seen this on other internet tests and it may help you see where the test/questions are going astray

[hellothere]

But if he's doing this

Confidence in a test comes from getting accurate results. That's why this thread exists.

then he's validating through self typings anyway, so I thought maybe if you recorded the self-typing together with the other demographic info it would provide a more precise way to assess the effectiveness of the test

[munenori2]

You can always decide to toss out info, but you can't always ask for it after the fact. Though I'm sure he has a lot of data that doesn't mean anything (assuming he didn't wipe the stuff from when it was non-operational). I think that something like that would be at least potentially helpful. If not, then, heh, oh well.

[Strutopolis]

ILE Your result

IEE These types might also be considered
LIE
EIE

Wohoo, Im a fucking intuitive extrovert! Or wiat, that was not a such a big suprise, actually. :evilgrin:

[Slacker]

But if he's doing this



then he's validating through self typings anyway, so I thought maybe if you recorded the self-typing together with the other demographic info it would provide a more precise way to assess the effectiveness of the test

He isn't necessarily validating through self-typing. I bet he has a pretty good idea of what type each of us is regardless of what we think we are ourselves. He might be comparing our results with that.

[hellothere]

yeah I thought that might be what he's doing.

[Ezra]

Apparently, I'm an LII. LIE, LSE and LSI may also be considered i.e. they are likely. SLE is unlikely.

Personally, I think it was a good test, and even though I saw the workings of it, I answered as honestly as possible, because I see no point in taking a test if you already know what you're going to get, because you answer questions as an idealised version of yourself e.g. "I want to be ESI, so I will answer as if I am an ESI".

The flaws I think are that it appears to work based on something other than functions. Otherwise ILE would be likely instead of unlikely, based on the fact that I'm an LII. And LSI would be unlikely if SLE was. It seems that it's basically determined that I'm a Logical type and a Rational type; i.e. a TJ. If it were an MBTT test, this would be nothing new, because I know I'm a TJ (and if I'm an IxTJ, then I am an INTJ, because I am not an ISTJ).

Rick, what do you make of this test now you've got some feedback from a lot of people?

[Rick]

Rick, what do you make of this test now you've got some feedback from a lot of people?

I'm waiting very impatiently for my partners to provide me access to the database so that I can study how people are answering the questions and figure out what's going on. Until then I can only guess which questions need to be phrased better or substituted with better ones.

I do think the test has promise, and I'm crossing my fingers that a little bit of tweaking and improvement will double its accuracy.

I also wish I had more technical capabilities so that I could see if the dichotomy part alone was producing better results than the test as a whole, or whether the function part alone was producing better results. So I'm a bit frustrated with my technical limitations and dependence on programmers.

[Mariano Rajoy]

LII Your result

LIE These types might also be considered
EII
EIE
LSI These types are not very likely
LSE
ILI
ILE
ESI These types are quite unlikely
ESE
IEI
IEE
SLI these types are extremely unlikely
SLE
SEI
SEE

[marooned]

I do think the test has promise, and I'm crossing my fingers that a little bit of tweaking and improvement will double its accuracy.

I think it has promise as well.

[xyz]

"Surrender the algorithm?" Do you even know what you are talking about?
Since when is providing all personal socionics work for your scrutiny required to have "credibility" at this site?
Nobody here is interested in the "algorithm" but you, and I have no idea what your intentions are for digging into the test mechanism, so why would I want to give you a 38 page document explaining how to create the test? What are the chances you would use it to help me improve my test? I think about zero. It would just give you another reason to go on ranting and raving for another ten pages in this thread or copy it and make ten new tests of your own, but worse.

Here is a sample excerpt of the "algorithm."

(...removed...)

At last, somebody freaking said it. Thank you.

[Creepy-Cyclops]

Maybe we're all untestable. Ha!

[normal]

[stevENTj]

Test results
ILE Your result
ILI These types might also be considered
LIE
LII
SLE These types are not very likely
IEE
SLI
LSE
IEI These types are quite unlikely
EIE
LSI
EII
SEE these types are extremely unlikely
SEI
ESE
ESI


I'm definitely not an ILE, but ILI was mentioned second, so not too bad. I think I saw one or two other ILIs getting ILE also. Hmmm.... I think I answered one or several of the questions incorrectly. They sounded overly positive so you just have to say yes, lol.

[Cuddly McFluffles]

Maybe we're all untestable. Ha!

*raises hand* I'm one of those.

[Rick]

Rick, I have a question about the algorithm, the first page seems fairly straight forward, but was the 2nd page related to the first page or the third page?

As for additional commentary on the algorithm. Since you've said that it is a 2 stage test, what one has to do is factor the information on each of the sections down to a common "product" in order for them to blend together. With both sections being subjective it can be diffucult in find this common "product". Some data mining techniques take a single set of answers and run the answers thru multiple algorithms and then blend the results together rather then multiple set of answers thru the same algorithm. Each "snapshot" or algorithm that produce a result will be given a ratio. So basically take the first section of the test or the 2nd section and run it thru multiple algorithms, weigh each set of results via a ratio and generate a final prediction. Unfortunately this method requires calibration in order train the ratios of the various algorithms you will use. So you will need a large dataset of self-typings in order to train the engine.

This is approaching the problem from a data mining perspective, which I think you are trying to do.

The 2nd and 3rd pages are all dichotomy based, so their results are factored in with the first page's results.

I do have a common product that allows the two sections to be blended seemlessly. After the first section the likelihood of each type is calculated, and each answer from the last set of questions (function related) multiplies certain type probabilities by certain coefficients. In other words, if my top score from the first section is LSI and I rate introverted logic as "this is totally not how I am," that coefficient for LSI will be far below 1 (but > 0), dropping the probability of that type severely (and also other types with Ti in the Ego).