Q1. What age should a person be - to be assured of a high degree of accuracy in their MBTI and Strong results?

the quick answer is 10th grade in both cases. Yes, the MBTI is written at the 7th grade reading level and the Strong is
written at the 9th grade reading level, but that's for easy understanding of the questions. Maturity and knowing yourself is an
entirely different matter. For all Step I MBTIs, officially, the MBTI researchers say "14+ years", but off the record, they will tell
you that the closer a person is to her/his 16th birthday, the more accurate the results will be. Why? you ask.

The operative words, here are validity and reliability. To digress a bit… the MBTI is just one example of what’s called a 'self-
report inventory' ...otherwise known as a projective test. This means a person fills out a questionnaire without the help of
anyone else.  Like all self-report inventories, the MBTI asks questions about situations in life, and a person’s individual
preferences and behaviors.  In general, the biggest problem when very young teens are given self-report inventories is they
often do not yet know who they are - or in the case of the Strong, know what they are interested in. Therefore, they project
whatever is on their mind (at that given moment) onto the the answer form. Re-stated in a light-hearted way, many of the
youngest teens are not yet “completely connected with themselves”. In short, their personality preferences haven’t yet
crystallized. And toward this end, this is exactly the same reason why the majority of psychologists recommend that self-report
inventories and projective tests not be given to 6th-7th and 8th graders at all - unless it is in a highly supervised and
individualized situation. As anyone who has ever been around these "tweens" knows, they can be quite volatile and on some
days, finding consistent patterns in their personality development can be a task.     

As far as the MBTI -
Step II assessment is concerned, because of the great depth of information it gives a person about their
personality, the researchers at CPP (Consulting Psychologists Press) in Palo Alto, California will tell you a
minimum age of 18 is

Q2. Speaking of accuracy, just how accurate is the MBTI for the average person?

the quick (and very unofficial) answer: the newest forms, Forms M and Q, of the MBTI are, on average, about 93%
accurate. Before Form M and Q, there was Form G - which was retired in 1998. It had a "lifetime batting average" of being
approximately 89% accurate.

Now...before proceeding further in this discussion, however, you should know that psychometricians recoil in horror when words
such as "accurate" and "accuracy" are bandied about. To them, to use a word like "accurate" to describe an assessment would
be similar to a mental health professional using the word "insane" (a legal term) to characterize someone who should properly
be described in highly specific medical terms. To use another analogy...audiologists would be extremely reluctant to say that you
have (for example) an overall 15% hearing loss. Instead, they would want to tell you in highly specific terms how much you've
lost at high pitches, low pitches, medium tones, etc. For them to clump all of these measurement together in one "meatball" so-
to-speak, is something that goes against their professional grain. In the same manner, MBTI statisticians understandably resist
taking all of these (very separate) reliability and validity statistics....for very different parts of one's personality and lump them
together. And so, re: the MBTI, where do these numbers (93% for the current version and 89% from the recently retired Form
G) come from? Answer: they are gleaned from some very informal conversations with some very highly respected people, shall-
we-say. Along these lines, even the untrained eye, upon looking at the very public research stats, can surmise that it is quite
uncommon for any of these measurements for both reliability and validity - not to mention for each facet of personality being
gaged - ever drops below the above-cited figures. In short, these instruments are rigorously validated, and to use the term
"accuracy", please understand (to perhaps over-do the point) this would be the equivalent to asking an air force general or naval
admiral, "How good are those boats?" or "how excellent are those aircraft?" MBTI psychometricians would definitely not approve
of this word. Of course, in part, that's makes them the professional statisticians they are.        

We return now, to these words validity and reliability, the two words that the world of testing lives and breathes on. Validity
means that an assessment measures what it is supposed to measure and/or predict. As well, an assessment that is high in
reliability yields consistent results on a repeated basis - no matter what the circumstances of testing situation are (different
cultures, etc.). For example, if the SAT (Scholastic Aptitude Test) started slipping in its reliability and validity, the ETS folks
(Educational Testing Service) would quickly be out of business. This of course, would also true with the MBTI and Consulting
Psychologists Press.  In the testing world, if your validity and reliability are compromised, as today's youth would say, "you're
cooked". As testament to that fact, about 15 years ago, the MBTI placed on the market, what has irreverently been labeled by
insiders as a "quick & dirty" 70 question MBTI. That is, as opposed to the normal 126 questions on Form G; 93 questions on
Form M; and 144 questions on Form Q. Alarm bells went off when on-going studies showed that its reliability and validity (on
average) had dropped into the low .8s and high .7s (where 1.0 represents a perfect correlation). As such, that "quick & dirty"
version was immediately pulled off the market. Mind you, this is in a testing world where
any test that has a reliability and
validity factor of .7 or above is considered excellent.

To continue the story: regarding today's MBTI, the validity and reliability are slightly different for each of the eight letter
dichotomies, i.e., E/I; S/N, T/F, and J/P. That said, over 60 years of research, hundreds of studies and many doctoral
dissertations, have documented that the MBTI has a degree of validity and reliability that is almost unheard of in the testing
world.  MBTI levels of validity and reliability have consistently been in the high .8s and low .9s. So, now that you are reading this
in context, we can repeat the statement from above. Today, Forms M and Q have an exceptionally high "accuracy" of 93%, on

Q3. Your explanations of the concepts of validity and reliability are most welcomed and the over-all statistics
are quite impressive. However, can you give
a concrete example of this so that I can understand validity and
reliability better?

: Sure. About 15 years ago, on Form G, the MBTI folks were experimenting with the following question:

Choice A: I can play at any time.
Choice B: I must finish my work before I can play.  

Up to that point, the MBTI psychometricians (a real word that describes psychologists who specialize in behavioral statistics)
knew exactly what direction as well as how heavily that question pulled on the J/P dichotomy. Then a "tweaking" experiment
began. They changed just three words in choice B.

Choice A: I can play at any time (stayed the same).
Choice B (the newer version read): I would like to finish my work before I can play.

To no one's surprise, by taking out the word "must" and inserting the words "would like to", it changed the entire chemistry of
the question - and thus, changed how heavily that question pulled in either direction.       

Little did anyone know (until the MBTI folks decided on what its final version of the question should look like), that this particular
question had been quietly included in the MBTI - but as an experimental question only. That is, it did not count toward an
individual's results. Moreover, they studied this one question for over a decade to be sure that it was a valid question
(measuring what it was supposed to) as well as reliable (showed consistency in all circumstances it was administered in). To
beg the point, if this is what happened when just three words of a question were changed in the MBTI's "mother tongue" of
English - then how would changing these words effect responses in the other 20 languages it is published in? For example, how
would this one seemingly small change in wording play-out in MBTIs in Norwegian; Korean; French; Greek; Chinese; Arabic and

All of this, by the way, is standard operating procedure in the world of testing. No matter what form of MBTI or Strong you are
taking, you can rest assured, there are almost guaranteed to be 7 or 8 questions - the responses to which - do not count toward
your results, but that are quietly being researched to see how they pull.

Examples of this from the educational world are the SAT and ACT tests for high schoolers. The statisticians from both of these
organizations will freely tell you that on each of their tests, there are a minimum of 20 questions which are part of quiet on-
going mini-experiments. Again, the answers don't count toward a student's score, because the question has not yet been
"standardized". That is, the SAT and ACT people are busily engaged in adjusting the wording and of course, watching closely
how many students got the question right or wrong. Too many right or wrong answers in either direction means the question
was either too easy or too difficult. With these academic tests, assuming they are aiming for a medium degree of difficulty, in
turn, they standardize that question to reflect exactly that. As well, if they are purposely trying to structure the question so that
it is difficult or easy to answer, then their standardization process will march in tandem to that as well. In short, that is what
standardized tests are all about. There's a lot to it.         

Q4. And to (perhaps) ask the obvious, based on what you said above...would this be the main reason why it is
not wise to depend upon, as you imply, these "short and sloppy" downloadable (for free) personality
inventories that seem to be everywhere in cyberspace?

: Precisely. Generally, they have not been validated and definitely not reliable. These "drive-thru window", "express check-
out" devices you see for free on the internet do not have six decades of research behind them and are usually concoctions of
those who have studied the highly impulsive media and consumer culture we live in - not much more.    

Q5. And what about the validity and reliability of the Strong Interest Inventory?

: As for the Strong Interest Inventory, the news has been - and continues to be - as good, indeed even a bit better than for
the MBTI...if you can believe that. Simply put, the accuracy rate for the Strong, especially the 2004 revision, is hovering around
an astonishing 95%. To augment that statement - and if you're the type that likes to read about empirical studies that involve
boatloads of number crunching - you are invited to click on the hyperlink below to read first-hand the actual figures on reliability
& validity for the newly revised Strong. Pay special attention to the bottom of page 4. The stats are pretty impressive.


Q6. Okay, the actual MBTI and the Strong Assessments have an extremely high reliability and validity rating as
instruments themselves. But even when allowing for that, is it possible that I could get my results back and
they would be "totally not me"? And if so...how could that happen?

Yes, but this would be very-very uncommon. How could such a thing happen at all, you ask? Remember those above-
mentioned very young teenagers...teenagers who might not know who they are yet? Well...how do we say this without offending
a few people out there in cyberland? Ummm...upon occasion, this happens to adults as well. That is, despite the image some
adults have of their own personalities, they still don't know who they are yet. As an example, Terry Marselle, the founder of this
web-site, will be quick to tell you that for much of his own life, he mistakenly pictured himself as an extravert - when in fact he
is quite an introvert. So why did he go all these years thinking that he was an extravert? Because, like many people, when he is
in his "comfort zone", i.e. in the company of close friends, he is quite out-going. And thus, in this one area of his personality, he
had a false read on himself. That is, until he took the MBTI - Step ll in 1997 and found out that, yes, he is indeed an introvert. In
the closer look that the Step ll gives, one of the sub-scales in the E/I dichotomy reported that he is an "expressive introvert". At
first glance, an expressive introvert sounds like a conflict in terms. This is because expressiveness is a sub-scale of extraversion
- and by being an expressive introvert, Terry is considered "out-of-polarity" for a typical introvert. Nothing wrong with that, of
course, but that just goes to show you how (in this case) what the deeper-drilling of the MBTI - Step ll can reveal.

A different example might be when a person has come to believe that they make most decisions based upon objectivity
(the T / F dichotomy) - when in fact, most of their decisions are based on personal / emotional values. And here-in lies the
beauty of MBTI - Step ll. Just when you think you know yourself through-and-through...you find out even more! In the final
analysis, most people understand that personality is a fluid concept that on certain days, can be difficult to define. And so we
return to our original point and that is: the incidents of receiving your MBTI personality description and the results being "totally
not me" are very-very uncommon.         
This entire page is
dedicated to the two
words the world of
testing lives and
breathes on:

reliability and validity
reliability and validity of the   MBTI®  - Strong®  &  MMTIC®