Causation
Form of logic. Implies correlation between A + B; Implies chronology - A before B and it strongly suggests no competing cause. There are two types of causal event.
- Correlation and
- Co-Incidence
Premises give us either correlation or co-incidence
Either will be a competing explanation, Third common cause or no relationship.
Correlation
Two things (phenomena) that happen / change together (can take place over time).
“Correlation” is a “complex” phenomenon as correlation itself is a relationship between two “simpler” phenomena.
Empirically observed co-variance.
Empirically observed - data; tests; trials; what out see out in the world.
Co-variance - change that happens together.
When there are more than a few instances of phenomenon co-occurring, it ceases to be a coincidence and becomes a correlation. (It’s unclear how many instances that it takes, but it happens together a lot.)
1. A causes B
2. B causes A
3. C causes both A and B
4. No relationship
Co-Incidence
Few instances (one or two) of a phenomenon co-occurring. They happen together. But co-incidence doesn’t imply correlation.
Causation
Will either strengthen or weaken.
What is Causal Logic? Is it ever valid?
- Causal logic is not formal. It’s informal.
- Arguments that use causal logic will never be valid. Their premises, even when true, will never guarantee the truth of their conclusions. i.e. There can always exist some scenario where the cause is NOT true.
But, that doesn’t mean causal arguments are necessarily weak. In fact, causal arguments can be very strong.
KEY: Causation is the relationship between phenomena.
For causal arguments specifically, requires evaluating other hypotheses to decide whether the proposed hypothesis is the true explanation.
Causal claims can be chained together.
What are Phenomena?
A phenomenon is a fact or event. Phenomena can also be fast or slow.
BUT, not all phenomena are equal.
1. There are Phenomena that are Premises (i.e. potential Causes).
- There are Phenomena that are Target Phenomena (i.e. Effects / the phenomenon that prompted the search for an explanation / the thing that we’re trying to explain).
TARGET PHENOMENA are also known as Hypothesis.
What are Explanations?
An explanation is our way of trying to understand. It’s our way of telling a CAUSAL story to make sense of the PHENOMENA
i.e. An explanation is just a set of causally linked phenomena.
An entire diagram / chain of causally linked phenomena, is the “explanation.”
i.e. Causal claims can be chained together.
What is a Hypothesis?
A hypothesis is a potential explanation. A hypothesis is a guess, a story that we make up that tries to make sense of the TARGET PHENOMENA.
How can we tell which hypothesis is the true explanation and which one isn’t?
1. The hypothesis has to be testable in order for it to potentially be true / qualify to be a hypothesis.
2. A hypothesis is itself a phenomenon that can be explained in terms of causes (in relation to ____(what?))
If the ALTERNATIVE hypothesis is declared to be true and can explain the original phenomena, then what does that do to the original argument (support for the original hypothesis)?
If the alternative hypothesis is declared to be true and can explain the original phenomena, then THAT WEAKENS the original argument (support for the original hypothesis) BECAUSE there will be no need anymore to favor it (the original hypothesis / argument).
What benefit is there to evaluating a potential hypothesis by asking “How?”
By asking “How?”, in so doing, you’re SEARCHING for a causal mechanism which is a more detailed causal story. If you are able to Identify a causal mechanism, then that STRENGTHENS the original argument (support for the original hypothesis).
True or False: Causes must precede their effects?
It is a principle of causal logic that causes must precede their effects. That principle gives us another method to evaluate causal hypotheses. We can preclude any hypothesis that fails to respect this principle.
If finding SIMILAR causal phenomena is the route you’re taking to evaluate a hypothesis, then you want to make sure ____(what?)
So, if finding similar causal phenomena is the route you’re taking to evaluate a hypothesis, then you want to make sure THE PHENOMENA YOU”RE EXAMINING ARE ACTUALLY SIMILAR.
The more similar the better. Recognize that this is the LOGIC OF ANALOGOUS ARGUMENTS.
How might you decipher 2 different hypothesis that tell different causal stories?
Because different hypotheses tell different causal stories, one way to decide between them is to CHECK FOR CORROBORATING OR CONFLICTING EVIDENCE.
What are Predictions? How do they relate to Hypotheses?
Predictions are claims about the future. As such, they’re claims about events that have not yet taken place.
All hypotheses make predictions (even if they’re implicit) and different hypotheses make different predictions.
H => P
A final way to check a hypothesis is to see if its predictions are true or false.
Define “Positive” Correlation.
Positive correlation - move in the same direction.
If A and B are positively correlated, then they both increase or decrease together, in the same direction.
Define “Negative” Correlation
Negative correlation - move in opposite directions.
If A and B are negatively correlated, then they move in the opposite direction. Note that they are still correlated. It’s just that when one moves in one direction, the other moves in the other direction.
Define “Statistical” Correlation
Technically, correlation is a statistic expressed as a coefficient between -1 to 1, where 1 is perfectly positively correlated and -1 is perfectly negatively correlated. The closer the coefficient is to -1 or 1, the stronger the correlation.
-1 (Perfectly Negatively Correlated) < x < 1 (Perfectly Positively Correlated)
When you encounter a correlation on the LSAT, ASSUME CORRELATIONS ARE IMPERFECT.
A and B do move together (either in the same direction - positive - or in the opposite direction - negative) but that doesn’t mean the movement is perfect. This will help you evade the LSAT’s attempt to induce you to confuse imperfect correlation with no correlation.
What then does a correlation coefficient of 0 mean?
i.e. on the spectrum of -1 < x < 1
A correlation coefficient of 0 means there is NO CORRELATION. i.e. Phenomena have no movement together.
What is the difference between “Correlation” and “Causation”?
Correlation phenomenon are just like “one-off” phenomenon in that they demand an explanation (a causal story). Just because 2 phenomena may or may not move together does not equal causation outright.
For example, there is no correlation between shoe size and lung cancer, AND that does not state what CAUSES lung cancer. The correlation there (even if it is zero) still DEMANDS an explanation: “What causes Lung Cancer?”
The danger here is that you conflate correlation with causation.
- Correlation is NOT identical to causation.
- Correlation does NOT imply causation.
Really pay attention to the word “imply.” Correlation does not imply causation, true, but that doesn’t mean correlation has no relationship to causation. Often, it does. - Correlation often CAN suggest the possibility of causation.
- Correlation CAN be evidence for causation.
What are the 4 Hypothesis to consider for Correlation?
When you encounter a correlation on the LSAT, consider four common hypotheses:
Hypothesis 1: A causes B
Hypothesis 2: B causes A
Hypothesis 3: C causes both A and B
Hypothesis 4: Just a coincidence, no causal relationship connects A and B
What are the different methods for testing or deciding which hypothesis is the ONE TRUE EXPLANATION?
i.e. The different methods apply to hypotheses about correlations as well.
- Chronology
- Causal mechanism
- Similar causal relationships
- Direct evidence
- Predictions
What makes up an Ideal Experiment? i.e. run through the Logic of an Experiment
- Large, random sample of the population.
It should be large enough so that whatever characteristics in the population in general should show up in equal proportion in the sample. The larger your sample size, the more dilute idiosyncrasies become.
Large and random samples help to establish representativeness. - Randomly assign sample into two groups.
Test group vs. Control Group
Random assignments into test and control groups help to cleanly isolate causal forces. - Compare the results.
- Reveal causation
- Isolate the cause by controlling for everything else
What impact does Self-Assignment have on causation?
Self-assignment OBSCURES causation. How do we know it was truly random?
What is the difference between:
1. “Because of” versus “In spite of” or “despite”?
This is what “in spite of” or “despite” means in the context of causal logic.
You don’t say “She’s running a fever Because she took Tylenol” as if Tylenol caused her fever. You say, “She’s running a fever IN SPITE OF (or despite) having taken Tylenol” because that captures the causal reality that Tylenol is helping to reduce the fever.
Similarly, you say that she’s feeling nauseous DESPITE having taken Dramamine. The implication is that had she not taken Dramamine, she’d probably be feeling even worse.
CAUTION: The exam will try to pull a fast one and make you think that A causes B to worsen where in fact, A causes B to improve. Often, this is achieved by presenting to you what LOOKS LIKE an experiment. But if you remember the principles of how to properly set up an experiment and look more closely, you’ll see through the ruse.
It’s not “because of” it’s “in spite of.”
