Correlation Research
Summary
TLDRProfessor Doug Crawl from Northern Kentucky University discusses correlational research, which explores the association between two variables. He explains that correlation aids in prediction, using ACT scores and college performance as an example. Crawl clarifies that correlation values range from -1 to 1, indicating both the direction and strength of the relationship. He emphasizes that correlation does not imply causation, highlighting the importance of not confusing correlation with causative relationships. The lecture also touches on the direction problem and the third variable problem, using self-esteem and GPA, as well as ice cream sales and assaults, to illustrate these concepts.
Takeaways
- đ Correlational research aims to discover if two variables are associated or related.
- đ A variable is a characteristic that varies among individuals, such as age or self-esteem levels.
- đ§ Knowing the correlation between two variables can help in making predictions about one variable when the other is known.
- đ Correlations range from -1 to 1, with -1 indicating a perfect negative correlation, 1 indicating a perfect positive correlation, and 0 indicating no correlation.
- đŒ Positive correlations suggest that as one variable increases, the other also increases, while negative correlations suggest an inverse relationship.
- đ The magnitude of the correlation coefficient indicates the strength of the relationship between the variables.
- đ« Correlation does not imply causation; just because two variables are correlated does not mean one causes the other.
- â The direction problem in correlations arises when it's unclear whether variable A causes B or B causes A.
- đ The third variable problem occurs when two variables are correlated not because they directly affect each other, but because they are both influenced by a third variable.
- đ§Ș To determine causality, experiments are necessary rather than relying solely on correlational data.
Q & A
What is correlational research?
-Correlational research is a type of study that seeks to discover if two variables are associated or related in some way.
What is a variable in the context of research?
-A variable is a characteristic that everyone has but with different values among individuals, such as age or levels of self-esteem.
Why is it important to understand correlations?
-Understanding correlations is important for prediction. If two variables are correlated, knowing one allows us to make an educated guess about the other.
Give an example of a correlation mentioned in the script.
-An example of a correlation given in the script is between ACT scores and college performance.
What is the range of correlation coefficients?
-Correlation coefficients range from -1 to 1, where -1 indicates a perfect negative correlation, 1 indicates a perfect positive correlation, and 0 indicates no correlation.
What does the direction of a correlation indicate?
-The direction of a correlation indicates whether the variables increase or decrease together. A positive correlation means they rise and fall together, while a negative correlation means they have a reciprocal relationship.
What does the magnitude of a correlation tell us?
-The magnitude of a correlation, ignoring the sign, tells us how strongly related the variables are and informs us about the confidence we can have in our predictions.
Why can't we infer causality from correlation?
-We cannot infer causality from correlation because correlation only indicates an association between variables, not a cause-and-effect relationship.
What is the 'direction problem' in correlation studies?
-The 'direction problem' refers to the issue of determining whether variable A causes variable B, variable B causes variable A, or if there is a bidirectional relationship.
What is the 'third variable problem' in correlation studies?
-The 'third variable problem' occurs when two variables are correlated not because they directly cause each other, but because a third variable influences both of them.
What does the example of ice cream sales and assaults illustrate about correlations?
-The example of ice cream sales and assaults illustrates that two variables can be correlated due to a third variable (heat), and this correlation does not imply causation between the two variables themselves.
Outlines
đ Understanding Correlational Research
Doug Craw, a professor at Northern Kentucky University, introduces correlational research, which aims to identify associations between two variables. Variables are characteristics that vary among individuals, such as age or self-esteem. The purpose of studying correlations is to make predictions; knowing one variable can help us estimate another. For instance, there's a correlation between ACT scores and college performance, suggesting that higher ACT scores are generally indicative of better college performance. However, this is not a perfect rule, as exceptions exist. Craw emphasizes that while correlations can range from -1 to 1, indicating the strength and direction of the relationship, they do not imply causation. A positive correlation means variables increase or decrease together, while a negative correlation implies an inverse relationship. The magnitude of the correlation indicates the strength of the relationship and the reliability of predictions based on it. Despite their usefulness, correlations must be interpreted with caution, as they do not provide insight into causality.
đ€ The Limitations of Correlations in Determining Causality
This paragraph delves into the limitations of using correlations to determine causality. Craw discusses how correlations can suggest a relationship between variables but do not prove that one causes the other. He uses the example of self-esteem and GPA, which may be correlated, but it's unclear whether self-esteem leads to higher GPA or vice versa. The 'direction problem' arises because it's challenging to determine the cause-and-effect relationship solely from correlation data. The 'third variable problem' further complicates matters, as two variables might be correlated not because they directly influence each other but because a third, unseen variable influences both. Craw gives the example of ice cream sales and assault rates, which correlate positively across months but are actually both influenced by temperature. This illustrates how correlations can be misleading without considering other potential factors. The paragraph concludes by stating that to understand causality, one must conduct experiments rather than relying solely on correlational data.
Mindmap
Keywords
đĄCorrelational Research
đĄVariables
đĄPrediction
đĄCorrelation Coefficient
đĄPositive Correlation
đĄNegative Correlation
đĄMagnitude
đĄCausation
đĄDirection Problem
đĄThird Variable Problem
Highlights
Correlational research aims to discover if two variables are associated or related.
A variable is a characteristic that everyone has but with different values, such as age or self-esteem.
Correlations are used for prediction, allowing educated guesses about one variable based on knowledge of another.
ACT scores and college performance are an example of correlated variables.
Correlations range from -1 to 1, indicating the strength and direction of the relationship between variables.
A positive correlation indicates variables rise and fall together, while a negative correlation suggests a reciprocal relationship.
The magnitude of correlation, ignoring the sign, indicates the strength of the relationship.
Highly correlated variables allow for more confident predictions, whereas weak correlations provide less certainty.
A correlation of zero indicates no relationship between variables.
Correlation does not imply causation; knowing two variables are correlated does not reveal the causal relationship.
The direction problem in correlations arises when it's unclear whether one variable causes the other or vice versa.
The third variable problem suggests that two correlated variables might be related due to a third, underlying variable causing both.
An example of the third variable problem is the correlation between ice cream sales and assaults, which both peak in the summer due to heat.
To determine causality, an experiment is needed rather than relying solely on correlational data.
Transcripts
my name is Doug crawl I'm a professor in
the department of psychological science
at Northern Kentucky University and I'm
going to talk to you about correlational
research correlational research seeks to
discover if two variables are associated
or related in some way now a variable is
a characteristic that everyone has but
different people have different values
so for example everyone has an age but
different people have different ages
everyone has a level of self-esteem but
different people have different levels
of
self-esteem now one of the main reasons
that we want to know about correlations
is for prediction that if two variables
are correlated then that means knowing
one allows us to take an educated guess
about what the other one is likely to be
for example something like ACT scores
and college performance there's a
correlation between ACT scores and
college performance so if we know a
person's ACT score that gives us a clue
about what the person's College
performance it's likely to be now it's
not perfect sometimes people may have a
relatively low ACT score and end up
doing brilliantly in college and
sometimes people have a really high ACT
score and end up doing not so well in
college but as a general rule ACT score
tells us something about what their
college performance is likely to be all
right well correlations range from one
to minus one got to be a number
somewhere in that range so if we
calculate the correlation it turns out
to be you know uh 227 or a platypus or
by low cell High then something's wrong
because it can't be that it's got to be
a number somewhere from one to minus one
now we could break that down into a
couple of aspects one is the direction
is it a positive number or a negative
number a positive correlation like
positive. 53 or positive. 27 that
indicates that the variables rise and
fall together so as one variable
increases then the other variable also
increases if one of them is falling then
the other one uh tends to be low as well
so for example something like uh hour
studying and exam performance we would
expect that uh as the number of ours
studied goes up that exam performance
would also tend to go up if uh hour
studying is low then exam performance
would also tend to be low on the other
hand a negative correlation means that
they have a reciprocal relationship
whereas one of them increases the other
one tends to fall or if one of them is
low then the other one tends to be high
maybe something like uh classes skipped
and exam performance we would expect
that as the number of classes skipped
goes up that exam performance would tend
to be lower to the degree that people
don't skip classes then we would expect
exam performance to be higher so that's
the direction but then there's also the
magnitude that is how strongly uh
related are they ignoring the sign so
two variables could be very strongly
correlated like positive point8 or
negative .8 or they could be only weekly
correlated like positive .1 or NE .1 and
the magnitude kind of is it tells us
something about how strongly related
they are and so it tells us something
about how confident we can be in our
prediction if two things are highly
correlated The Knowing one really tells
us a lot about what the other one is
likely to be if they're only weakly
correlated then it tells us relatively
little about what the other one is
likely to be and of course if they're
correlated zero then it tells us nothing
about what the other one is likely to
be but probably the most important thing
to keep in mind with regard to
correlations is that correlation is not
causation that is if we know that two
variables are correlated that doesn't
tell us anything about what the causal
relationship between those variables
might be uh so for example let's suppose
we do a study here at NKU and uh we
discover that self-esteem and GPA are
positively correlated maybe positive. 35
and so if you as you might recall that
means that as one of them Rises the
other one tends to rise if one of them
is low than the other one tends to be
low and let's suppose someone hears
about that and maybe writes an article
in a newspaper somewhere and says study
reveals that self-esteem and GPA are
positively correlated therefore if we
want people's grades to be higher what
we need to do is build up everyone's
level of self-esteem now that would work
if the reason that GPA and self-esteem
are correlated is that esteem is driving
GPA then that would mean that any
changes that occur on self-esteem those
should cause changes to occur on GPA
but that might not necessarily be the
reason that they're correlated there's
at least a couple of problems that we
need to address uh one of them is the
direction problem the direction problem
is that it could be that self-esteem is
causing GPA but it could also be that
GPA is causing self-esteem and if that
second one is true then that would mean
that raising self-esteem won't do
anything to GPA because GPA is the one
causing self-esteem and uh we really
have no way of determining which of
those two things might be it could be
either one now some of you might be
thinking that sometimes we could
probably be fairly confident in ruling
out one of those two directions so for
example before we knew that smoking
caused lung cancer we probably knew that
they were correlated so you can imagine
a a researcher sitting there thinking
about this like let's see H smoking and
lung cancer are positively correlated
could it be that getting lung cancer
causes people to start smoking and
doesn't seem very likely and so it's
probably the other way and that sort of
reasoning of if we could just eliminate
One Direction then we would know that it
was the other one that sort of reasoning
would work if this was the only problem
but it's not the only problem we also
have the third variable problem that is
if two variables are correlated it could
be that neither one causes the other but
rather they're correlated because some
third variable causes both so for
example with regard to self-esteem and
GPA maybe uh good nutrition
tends to cause people to feel good about
themselves have higher esteem and also
tends to cause them to have higher
grades except it might not be nutrition
maybe it's U getting enough sleep maybe
when people get enough sleep then that
tends to cause them to have higher
esteem and also to have higher grades
except it might not be that maybe it's a
a supportive family to the degree that
people have a supportive family maybe
that causes them to have a higher esteem
and also higher grades and of course we
could you know sit here all day trying
to think about different things it could
possibly be and there's really no way to
know let me give you my favorite example
of the third variable problem I read
this years ago I'm not sure exactly
where it was but apparently ice cream
sales vary according to the month of the
year they tend to be relatively low in
January and then they go up and up and
up and they tend to Peak somewhere
around August and then they tend to
decline and be relatively low again in
December same sort of thing for assaults
like aggravated assaults tend to be
relatively low in January and then they
go up and up and up Peak somewhere
around August and then they tend to
decline and be relatively low in
December and so as a result those two
things are positively correlated they
tend to rise and fall together now can
imagine someone thinking about the
direction of that they might think let's
see is it the case that when people uh
beat somebody up they think you know it
would be a great way to top off this
evening big bowl of ice
cream I don't think that's it maybe it's
that when they uh eat a big bowl of ice
cream then they get that brain freeze
thing and that makes them mad and so
they decide to haul up and punch someone
in the nose no that doesn't make sense
to me either and you could probably
guess what the third variable probably
is it's probably heat that to the degree
that it gets hot then that tends to
cause ice cream sales to go up that also
tends to cause assaults to go up as well
so with a correlation we can't tell
anything about causality and yet we
often want to know something about cause
and so correlations are not going to
help us if we want to know about cause
we're going to have to do an experiment
o
Voir Plus de Vidéos Connexes
5.0 / 5 (0 votes)