- Joined
- Jun 23, 2010
- Messages
- 11,808
- Reaction score
- 2,807
Continuing a discussion between @Ace-Co-A, @sakabato93, and myself elsewhere on SDN, I'd like to consider the concept of intelligence and its relation to the MCAT. I can identify a number of core issues to be addressed:
A) What is intelligence? How can it be meaningfully defined?
B) What is the purpose of the MCAT? Does the test succeed in fulfilling its objectives?
C) What is the relationship between intelligence and performance on the MCAT?
D) What problems can be identified with the conclusions we might draw from answering the above?
Here's my take on these issues.
A) The concept of intelligence is complex, and often it can be useful to operationalize intelligence in the context with which one is concerned. For example, it is my opinion that intelligence in the most broad sense can be operationalized at least in part in terms of several characteristics, which fall into categories related to (1) memory and (2) processing.
(1) Characteristics of memory relevant to the concept of intelligence: memory capacity (long-term, short-term, and working memory stores; how much information one is able to retain in memory), memory encoding (how easily/quickly one can add new information to each of the memory stores), and memory persistence (the duration of time for which information can be retained in memory stores without retrieval or reinforcement).
(2) Characteristics of processing relevant to the concept of intelligence: processing speed (the rate at which manipulation of information can be carried out), and parallel processing (the extent to which distinct processes can be carried out simultaneously).
All of that is to say that I find it reasonable to say that between two individuals, the one who can retain more information in memory, add new information more quickly and easily into memory, retain it the longest, think the fastest, and think at the most complex level is (ignoring other differences) the more intelligent one. This approach of course leaves out important facets such as characteristics associated with attention and perception, but I'm really not qualified to flesh out the entire concept of intelligence here. Add to those such macro-scale abilities like creativity, social behavior, and others and you get a seriously complicated situation. The characteristics I've presented, however, I do feel are pretty core (though not comprehensive) to the idea of intelligence at least the way I think of it.
Open questions: What might intelligence mean in the context of physician performance? Which of my core characteristics do you agree or disagree with? What might you add? Would you approach defining the term differently, and if so, how?
B) According to the AAMC…
I'm not surprised that the word "intelligence" is nowhere to be found. From this statement, however, we can identify that the MCAT is designed to measure the following:
1) The amount of information retained in memory (this is the science content test takers are responsible for knowing), and by necessity the accessibility of this information as well.
2) The degree to which two behaviors can be performed: problem-solving and critical thinking.I'm not going to get into defining those behaviors, but anyone familiar with the MCAT would agree that the passage-style format involves, at least to some extent, the assimilation of new information into memory and the processing of that information in the context of background knowledge.
Further, because the MCAT is a timed test, I feel I can reasonably argue that the exam measures processing speed (it's not like the time limit is 3 hours per section; a decent pace is required and is frequently a limiting factor is test taker performance).
I'm actually going to steer clear of addressing whether the test succeeds or not. The MCAT has respectable validity with regard to predicting USMLE Step 1 performance, and is the product of decades of development by professionals with background in test design. Overall, it almost certainly works pretty well in measuring what it's trying to measure.
Open questions: Are problem-solving and critical thinking necessarily representative of intelligence? Given that the test explicitly attempts to measure test taker knowledge, does this necessarily represent intelligence? Why or why not?
C) One of the points @Ace-Co-A raised in prior discussion examined the relationship between MCAT performance and intelligence:
It's an interesting issue; to what extend can we say that a higher score represents greater intelligence? How much of one's score is attributable to the MCAT's measurement of factors that we would agree represent aspects of intelligence? How certain can be we that someone who scores a 40 is at least somewhat more intelligence that someone who scores a 24? Absolutely? Not at all? What influences our answer?
D) @Ace-Co-A further noted problems with the strength of an MCAT score-intelligence relationship:
These are legitimate issues that affect both the reliability and validity of the test itself. Standardization as a process acts to mitigate the various confounds in (2) to some extent, but the acknowledgement that they cannot be eliminated completely is part of why we have classical test theory to describe scores: one's score is always the sum of one's true, perfectly representative score and some unknown amount of error (X = T + E).
As an excellently reliable exam (r > 0.9) the MCAT has a standard error of measurement (SEM) of 2 composite points, which represents the typical degree to which one's reported score differs from one's true score. Examining the confidence intervals obtained when the SEM is known demonstrates that we should not distinguish between three scores of 34, 33, and 32, for example. All may be significantly representative of the test taker's true characteristics (as problematic as that may be for admissions purposes.)
My comment on (3) is simply that, depending on how relevant we think content knowledge is in terms of representing one's intelligence, the contribution of these external factors to the MCAT's measurement of one's intelligence may be smaller than would appear.
As a random thought, the MCAT is a multiple-choice test with no guessing penalty, so luck/chance/whatever you want to call it could account for significant reported score variance in some instances as well.
Conclusions: I personally think that the characteristics measured by the MCAT both directly and indirectly attempt to capture the test taker's intelligence (broadly defined), but acknowledge that there are certainly external factors that can affect both the ability of the exam to accurately measure what it attempts to and the representativeness of one's score. In general, however, I do think the exam does a pretty good job of measuring these intelligence-relevant characteristics. I would also emphasize that caution should be used in drawing strict conclusions and comparisons regarding intelligence when using MCAT scores as a measure, especially when scores are close or there are known external variables that may affect the reliability of the exam as a metric.
A) What is intelligence? How can it be meaningfully defined?
B) What is the purpose of the MCAT? Does the test succeed in fulfilling its objectives?
C) What is the relationship between intelligence and performance on the MCAT?
D) What problems can be identified with the conclusions we might draw from answering the above?
Here's my take on these issues.
A) The concept of intelligence is complex, and often it can be useful to operationalize intelligence in the context with which one is concerned. For example, it is my opinion that intelligence in the most broad sense can be operationalized at least in part in terms of several characteristics, which fall into categories related to (1) memory and (2) processing.
(1) Characteristics of memory relevant to the concept of intelligence: memory capacity (long-term, short-term, and working memory stores; how much information one is able to retain in memory), memory encoding (how easily/quickly one can add new information to each of the memory stores), and memory persistence (the duration of time for which information can be retained in memory stores without retrieval or reinforcement).
(2) Characteristics of processing relevant to the concept of intelligence: processing speed (the rate at which manipulation of information can be carried out), and parallel processing (the extent to which distinct processes can be carried out simultaneously).
All of that is to say that I find it reasonable to say that between two individuals, the one who can retain more information in memory, add new information more quickly and easily into memory, retain it the longest, think the fastest, and think at the most complex level is (ignoring other differences) the more intelligent one. This approach of course leaves out important facets such as characteristics associated with attention and perception, but I'm really not qualified to flesh out the entire concept of intelligence here. Add to those such macro-scale abilities like creativity, social behavior, and others and you get a seriously complicated situation. The characteristics I've presented, however, I do feel are pretty core (though not comprehensive) to the idea of intelligence at least the way I think of it.
Open questions: What might intelligence mean in the context of physician performance? Which of my core characteristics do you agree or disagree with? What might you add? Would you approach defining the term differently, and if so, how?
B) According to the AAMC…
MCAT Essentials 2013 said:The MCAT tests for mastery of basic concepts in biology, general chemistry, organic chemistry, and physics. Additionally, it serves to assess examinees’ capacity for problem solving and critical thinking.
I'm not surprised that the word "intelligence" is nowhere to be found. From this statement, however, we can identify that the MCAT is designed to measure the following:
1) The amount of information retained in memory (this is the science content test takers are responsible for knowing), and by necessity the accessibility of this information as well.
2) The degree to which two behaviors can be performed: problem-solving and critical thinking.I'm not going to get into defining those behaviors, but anyone familiar with the MCAT would agree that the passage-style format involves, at least to some extent, the assimilation of new information into memory and the processing of that information in the context of background knowledge.
Further, because the MCAT is a timed test, I feel I can reasonably argue that the exam measures processing speed (it's not like the time limit is 3 hours per section; a decent pace is required and is frequently a limiting factor is test taker performance).
I'm actually going to steer clear of addressing whether the test succeeds or not. The MCAT has respectable validity with regard to predicting USMLE Step 1 performance, and is the product of decades of development by professionals with background in test design. Overall, it almost certainly works pretty well in measuring what it's trying to measure.
Open questions: Are problem-solving and critical thinking necessarily representative of intelligence? Given that the test explicitly attempts to measure test taker knowledge, does this necessarily represent intelligence? Why or why not?
C) One of the points @Ace-Co-A raised in prior discussion examined the relationship between MCAT performance and intelligence:
My principal objection to equating or strongly relating MCAT score with 'intelligence' will be seen in the following inference: if it is always true that a higher MCAT score corresponds to a keener intellect, then we must say that someone with a higher MCAT score (let's say a 34) is smarter than someone with a lower MCAT score (let's say a 31), right?
It's an interesting issue; to what extend can we say that a higher score represents greater intelligence? How much of one's score is attributable to the MCAT's measurement of factors that we would agree represent aspects of intelligence? How certain can be we that someone who scores a 40 is at least somewhat more intelligence that someone who scores a 24? Absolutely? Not at all? What influences our answer?
D) @Ace-Co-A further noted problems with the strength of an MCAT score-intelligence relationship:
(1) MCAT scores vary from day-to-day. For example, Kim might score 34 one day, 33 the next, and 32 the day after. Which score is most representative of her intelligence? Is she smarter the first day (with her 34) than the third (with her 32)?
(2) Test scores vary depending on uncontrollable variables (the person might have a fever, her father might have died the night before, her untimely boyfriend proposed that morning, etc.)
(3) Test scores depend on the resources available to the individual (practice tests, books/information, problem sets, previous coursework, prep class, etc.)
These are legitimate issues that affect both the reliability and validity of the test itself. Standardization as a process acts to mitigate the various confounds in (2) to some extent, but the acknowledgement that they cannot be eliminated completely is part of why we have classical test theory to describe scores: one's score is always the sum of one's true, perfectly representative score and some unknown amount of error (X = T + E).
As an excellently reliable exam (r > 0.9) the MCAT has a standard error of measurement (SEM) of 2 composite points, which represents the typical degree to which one's reported score differs from one's true score. Examining the confidence intervals obtained when the SEM is known demonstrates that we should not distinguish between three scores of 34, 33, and 32, for example. All may be significantly representative of the test taker's true characteristics (as problematic as that may be for admissions purposes.)
My comment on (3) is simply that, depending on how relevant we think content knowledge is in terms of representing one's intelligence, the contribution of these external factors to the MCAT's measurement of one's intelligence may be smaller than would appear.
As a random thought, the MCAT is a multiple-choice test with no guessing penalty, so luck/chance/whatever you want to call it could account for significant reported score variance in some instances as well.
Conclusions: I personally think that the characteristics measured by the MCAT both directly and indirectly attempt to capture the test taker's intelligence (broadly defined), but acknowledge that there are certainly external factors that can affect both the ability of the exam to accurately measure what it attempts to and the representativeness of one's score. In general, however, I do think the exam does a pretty good job of measuring these intelligence-relevant characteristics. I would also emphasize that caution should be used in drawing strict conclusions and comparisons regarding intelligence when using MCAT scores as a measure, especially when scores are close or there are known external variables that may affect the reliability of the exam as a metric.