PSYC 317 EXAM 3 QUESTIONS AND
ANSWERS GRADED A+
So What is a Correlation? - •The relationship between two or more variables
•One variable changes when the other variable changes= If it does, it means the variables covary
•Ex. The OLDER the person is, perhaps the MORE slowly the person drives
•You receive a HIGHER grade the MORE that you study
•As age INCREASES, your ability to hear high pitched noise DECREASES
•As one varies, so does the other
•CORRELATION DOES NOT MEAN CAUSATION
From Descriptive to Correlational Research - Who will experience Alzheimer's?
How can you take descriptive research and turn it into something correlational and or predictive?
Characterized 10 lipids that were different in patients with and without Alzheimer's - found this pattern
thru descript. research
1. Decided what to examine: Alzheimer's
2. Decided how we to examine it
3. Decided what type of method to use: Physiological (Blood Samples)
Question: Can we use these lipids to predict who will experience Alzheimer's? - this is going from
descriptive to correlational research
--Answer: Yes, with more than 90 percent accuracy the onset of the disease.
Correlational Research - •Examines whether variables are related to one another
,•We measure the two variables usually through observational methods (e.g., surveys, observing, asking
in person)
•We examine their correlation coefficient
Correlation Coefficient (r) - •Tells us the strength and direction of a relationship (of at least two
variables)
•Direction: ranges from -1.00 to +1.00
-Negative correlations mean that as one variable increases, the other decreases
-Positive correlations mean that as one variable increases, the other increases = going the same direction
--We can tell the direction by a scatterplot
•Magnitude: size of the relationship
-Larger values = stronger relationship
•0.48 is a smaller correlation than 0.79
•-0.23 is a smaller correlation than -0.49 - remember to use absolute value
-Larger values = stronger relationship
-Size
•.1 = Small
•.3 = Medium
•.5 = Large
Other types of correlations (3) - -Perfect correlations usually only exist when you are correlating
something with itself
-No correlation - does not look like a line at all
--gives you r = 0.0, but there is no relationship
-Curvlinear relationship - there is a relationship present
--gives you 0.0, but there is a relationship
, -- Yurks Dotson law = optimal performance at moderate levels of anxiety = you have to look at the
scatterplot to see this relationship, r doesn't tell you
correlation coefficient and coefficient of determination - •Remember we can also square it (coefficient of
determination) to see how much variance of one variable is accounted for by another
•Example: jAlzheimer's jpatients j- jabnormal jlipids jaccount jfor j90% jof jwhether jsomeone jwould
jexperience jAlzheimer's. jThis jis jr2
Interpreting jCorrelation jCoefficients j- j•We jcan jinterpret jdirection jand jmagnitude
•BUT j- jour jcorrelation jcoefficient jis jnot jon ja jratio jscale, jso jwe jcan't jmultiply, jsubtract, jetc.
•Also jmeans jthat jour jvalues jare jnot jlinear
-E.g., j0.80 jis jnot jtwice jas jlarge jas j0.40 j= jnot jon ja jratio jscale
Statistical jSignificance jMatters jtoo! j- j•Just jbecause jthere jis ja jrelationship jin jyour jsample jdoesn't
jmean jit jwill jbe jtrue jfor jthe jpopulation!
•Therefore, jwe jrely jon jstatistics j(the jp-value) jto jtell jus jwhether jour jcorrelation jcan jbe jgeneralized
•Our jp-value jrefers jto jthe jprobability jof jthe jcorrelation joccurring jby jchance
-We jusually jwant jthis jto jbe j5% jor jlower j(p j≤ j0.05)
**look jat joutput**
--P jhs jto jbe jless jthan jor jequal j.05 j= jthis jmeans jthat jthere jis jless jthan ja j5% jchance jthat jyour jfindings
jwere jdue jto jchance
--Statistical jsignificance jexists jwhen ja jcorrelation jcoefficient jcalculated jon ja jsample jhas ja jvery jlow
jprobability jof jbeing jzero jin jthe jpopulation
What jAffects jStatistical jSignificance j(p j< j.05)? j- j•Sample jsize j- jthe jlarger jthe jsample, jthe jmore jlikely
jwe jfind jsignificance. jIf jyou jfind jsig, jyou jalso jneed jto jlook jat jeffect jsize
•Magnitude jof jthe jcorrelation j- jvery jlarge jcorrelation jwill jlikely jalso jbe jsignificant. jLarger jthe
jmagnitude, jthe jmore jlikely jthe jresults jwill jbe jsignificant
ANSWERS GRADED A+
So What is a Correlation? - •The relationship between two or more variables
•One variable changes when the other variable changes= If it does, it means the variables covary
•Ex. The OLDER the person is, perhaps the MORE slowly the person drives
•You receive a HIGHER grade the MORE that you study
•As age INCREASES, your ability to hear high pitched noise DECREASES
•As one varies, so does the other
•CORRELATION DOES NOT MEAN CAUSATION
From Descriptive to Correlational Research - Who will experience Alzheimer's?
How can you take descriptive research and turn it into something correlational and or predictive?
Characterized 10 lipids that were different in patients with and without Alzheimer's - found this pattern
thru descript. research
1. Decided what to examine: Alzheimer's
2. Decided how we to examine it
3. Decided what type of method to use: Physiological (Blood Samples)
Question: Can we use these lipids to predict who will experience Alzheimer's? - this is going from
descriptive to correlational research
--Answer: Yes, with more than 90 percent accuracy the onset of the disease.
Correlational Research - •Examines whether variables are related to one another
,•We measure the two variables usually through observational methods (e.g., surveys, observing, asking
in person)
•We examine their correlation coefficient
Correlation Coefficient (r) - •Tells us the strength and direction of a relationship (of at least two
variables)
•Direction: ranges from -1.00 to +1.00
-Negative correlations mean that as one variable increases, the other decreases
-Positive correlations mean that as one variable increases, the other increases = going the same direction
--We can tell the direction by a scatterplot
•Magnitude: size of the relationship
-Larger values = stronger relationship
•0.48 is a smaller correlation than 0.79
•-0.23 is a smaller correlation than -0.49 - remember to use absolute value
-Larger values = stronger relationship
-Size
•.1 = Small
•.3 = Medium
•.5 = Large
Other types of correlations (3) - -Perfect correlations usually only exist when you are correlating
something with itself
-No correlation - does not look like a line at all
--gives you r = 0.0, but there is no relationship
-Curvlinear relationship - there is a relationship present
--gives you 0.0, but there is a relationship
, -- Yurks Dotson law = optimal performance at moderate levels of anxiety = you have to look at the
scatterplot to see this relationship, r doesn't tell you
correlation coefficient and coefficient of determination - •Remember we can also square it (coefficient of
determination) to see how much variance of one variable is accounted for by another
•Example: jAlzheimer's jpatients j- jabnormal jlipids jaccount jfor j90% jof jwhether jsomeone jwould
jexperience jAlzheimer's. jThis jis jr2
Interpreting jCorrelation jCoefficients j- j•We jcan jinterpret jdirection jand jmagnitude
•BUT j- jour jcorrelation jcoefficient jis jnot jon ja jratio jscale, jso jwe jcan't jmultiply, jsubtract, jetc.
•Also jmeans jthat jour jvalues jare jnot jlinear
-E.g., j0.80 jis jnot jtwice jas jlarge jas j0.40 j= jnot jon ja jratio jscale
Statistical jSignificance jMatters jtoo! j- j•Just jbecause jthere jis ja jrelationship jin jyour jsample jdoesn't
jmean jit jwill jbe jtrue jfor jthe jpopulation!
•Therefore, jwe jrely jon jstatistics j(the jp-value) jto jtell jus jwhether jour jcorrelation jcan jbe jgeneralized
•Our jp-value jrefers jto jthe jprobability jof jthe jcorrelation joccurring jby jchance
-We jusually jwant jthis jto jbe j5% jor jlower j(p j≤ j0.05)
**look jat joutput**
--P jhs jto jbe jless jthan jor jequal j.05 j= jthis jmeans jthat jthere jis jless jthan ja j5% jchance jthat jyour jfindings
jwere jdue jto jchance
--Statistical jsignificance jexists jwhen ja jcorrelation jcoefficient jcalculated jon ja jsample jhas ja jvery jlow
jprobability jof jbeing jzero jin jthe jpopulation
What jAffects jStatistical jSignificance j(p j< j.05)? j- j•Sample jsize j- jthe jlarger jthe jsample, jthe jmore jlikely
jwe jfind jsignificance. jIf jyou jfind jsig, jyou jalso jneed jto jlook jat jeffect jsize
•Magnitude jof jthe jcorrelation j- jvery jlarge jcorrelation jwill jlikely jalso jbe jsignificant. jLarger jthe
jmagnitude, jthe jmore jlikely jthe jresults jwill jbe jsignificant
