Correlation to Conditional = Implied Causation?

[secretad22]

I have practiced above 170 numerous times, including a 174. However, I came across something I want Manhattan people to help answer for me.

Premise: Correlation of A & B

Conclusion: If A ---> Then B

My thoughts: Of course it is a flaw that the argument concludes a conditional statement from the correlation. However, a very prominent LSAT poster told me that a causal assumption is involved in that hypothetical problem. Is this the case in your opinion?

Must there be a causal assumption when you go from a correlation in the premise to a conditional in the conclusion?

[ManhattanPrepLSAT1]

Can you tell me the specific question you're looking at, because I think there's an issue with the translation from the LSAT question to yours? Here's the issue, the relationship "if A... then B" is a correlation. And we never want to assume causation from a mere correlation, which I'm sure you know.

So in effect, saying that A and B are correlated, therefore, if A occurs than B occurs, seems pretty good. You're saying that there's a correlation between A and B, therefore there's a correlation between A and B. The only question I would have at that point is, "is the evidence suggesting a perfect correlation between A and B?" If so, we're good. But if the correlation is anything other than perfect, then the conclusion is no longer proven by the evidence.

But maybe there's something else in the question, that I could use to get a better insight to the specific situation...

[secretad22]

Thank you for the response.

This conversation stems from the Manhattan LR Guide. The question from PT 30 S2 Q15, about the recent economic downturn.

I posted this in its proper place, but it has not received an answer, so I will post it here to show you the example.

PT 30 S2 Q15:

The very first statement in this stimulus has me wondering if I can even say that the banks CAUSED the decline by loaning less money. I would say that it is not the case that we know what caused the decline. If the banks contributed to it, do we know that it is (the) / (a) cause?

If you negate choice (A) you get:

The downturn did cause a significant decrease in the total amount of money on deposit with banks which is the source of funds for banks to lend.

Does that destroy the conclusion of this argument?

The conclusion is: if standards are relaxed ---> banks will lend more money.

My thoughts: Even if the downturn caused a significant decrease in the total amount of money on deposit with banks, couldn't the banks still loan more money?

I think a necessary assumption is that the downturn did not cause a complete depletion of deposit bank money. However, a significant decrease of deposit money does not necessarily affect banks lending more money.



My issue in a more general sense is that lets say there is a correlation of ADHD and high blood pressure. There are some people with both ADHD and high blood pressure, so there is some relationship between those two variables.

However, to conclude, a conditional, as to say If ADHD ---> then high blood pressure is not valid.

Is the argument's author assuming causality is my question.

I assume this example I am discussing is what can be described as an imperfect correlation. I have not heard of the perfect/imperfect correlation terms. I assume a perfect correlation would be something like dog and animal?

[ManhattanPrepLSAT1]

Great questions!

My issue in a more general sense is that lets say there is a correlation of ADHD and high blood pressure. There are some people with both ADHD and high blood pressure, so there is some relationship between those two variables.

However, to conclude, a conditional, as to say If ADHD ---> then high blood pressure is not valid.

It depends on the strength of the correlation. If the argument says they are perfectly correlated, then it would be valid (meaning the two variables are mutually dependent - either you have both or else you have neither). However, the correlation you described, would not permit the conditional conclusion.

One of the major flaws we see time and again in LR is "generalizing from a limited source." Yours is a great example of this, "some people with ADHD have high blood pressure, therefore everyone with ADHD has high blood pressure.

Is the argument's author assuming causality is my question.

In your question, no causality is implied. However in PT30, S2, Q15 causality is implied with the language "contributed to."

Here's are some other language cues you can use to help spot "causality."

due to
because of
as a result of
contributes to
leads to
has led to
stimulates
causes
induces
produces
has the byproduct of
is a factor of
has the effect of
effected by
as a consequence of
if you want to _____, then you should ________.*
(*implying the second blank will cause the first to occur)

Hope that helps! And it looks like we have an answer to your question on PT30, S2, Q15 up already.

[goriano]

Great questions!

My issue in a more general sense is that lets say there is a correlation of ADHD and high blood pressure. There are some people with both ADHD and high blood pressure, so there is some relationship between those two variables.

However, to conclude, a conditional, as to say If ADHD ---> then high blood pressure is not valid.

It depends on the strength of the correlation. If the argument says they are perfectly correlated, then it would be valid (meaning the two variables are mutually dependent - either you have both or else you have neither). However, the correlation you described, would not permit the conditional conclusion.

One of the major flaws we see time and again in LR is "generalizing from a limited source." Yours is a great example of this, "some people with ADHD have high blood pressure, therefore everyone with ADHD has high blood pressure.

Is the argument's author assuming causality is my question.

In your question, no causality is implied. However in PT30, S2, Q15 causality is implied with the language "contributed to."

Here's are some other language cues you can use to help spot "causality."

due to
because of
as a result of
contributes to
leads to
has led to
induces
stimulates
causes
has the effect of
effected by
as a consequence of
if you want to _____, then you should ________.*
(*implying the second blank will cause the first to occur)

Hope that helps! And it looks like we have an answer to your question on PT30, S2, Q15 up already.

Thank you for the list of "trigger words" that indicate causality. Very helpful. Are the following trigger words considered correlation:

-People who do X tend to do Y
-People who do X are more likely to do Y
-X is associated with Y
-X is a byproduct of Y

If possible, could you create a similar list for correlation as well? Thank you!

[ManhattanPrepLSAT1]

Thank you for the list of "trigger words" that indicate causality. Very helpful. Are the following trigger words considered correlation:

-People who do X tend to do Y
-People who do X are more likely to do Y
-X is associated with Y
-X is a byproduct of Y

If possible, could you create a similar list for correlation as well? Thank you!

The first three are correlations, the last one (byproduct) is causation.

Let me first add to "causation" indicators the following two:
is a factor of
produces

*I'll edit the list above to include them (and yours!) for future readers.

For correlation:
associated with
co-occurs
coincides

But the majority of words that indicate correlation actually imply either sufficiency or necessity. So they don't only tell you that a relationship exists, but what sort of relationship it is.

Sufficient Condition indicators:
if
if only
the only
all
any
each
every
when
whenever
wherever
whoever
whatever
no*

*No A's are B's is notated: A ---> ~B

Necessary Condition Indicators:
only
only when
only if
needs
must
requires
depends on
relies on
necessitates
unless*
until*
except*
cannot... without*

*A must occur unless B occurs is notated: ~A ---> B

Conditional relationships are perfect correlations (at least in one direction). But there are other, less perfect correlations - likelihoods or frequencies.

The Frequency of Most:
more than half
a majority
nearly all
almost all
typically
usually
generally
most
tend to
few*

*Few A's are B's is notated: A -most-> ~B, which reads "most A's are not B's"

The Frequency of Some:
some
many
several
a few
occasionally
sometimes
regularly
often
not all*

*Not all A's are B's is notated: A <-some-> ~B, which reads, "some A's are not B's"

Hope that helps!

[wj097]

I have question regarding the correlation language "tend to".
In PT 20 S2 Q25, the correct answer choice refers "tend to" in the stimulus as some kind of causation...would be great if someone can resolve this issue, or point me where I am not seeing it right.

Thx

[noah]

I have question regarding the correlation language "tend to".
In PT 20 S2 Q25, the correct answer choice refers "tend to" in the stimulus as some kind of causation...would be great if someone can resolve this issue, or point me where I am not seeing it right.

Thx

I just replied there. (It's actually PT30, S2, Q25).

It looks like you misunderstood Matt's lists. "Tend to" is in his "Most" list, not his correlation one.