Correlation

  Shows the relationship between two different sets of scores

  Answers the question:   "Are high scores on one variable (thing) predictive of high scores on some other measure (or realm)?"

  Shows the degree to which two things change together (e.g., height & weight; distance from favorite person & happiness;
 

  It's "going-togetherness"

  Degree of correspondence between two values

  The tendency of one score to vary or change when another one does so.
 
 

Is there a correlation (relationship) between (from birth) Physical growth and knowledge of the world? (Yes, as you grow with age, you tend to get smarter.)

The amount you drink and how good the other person in the room looks? (Yes.  Many pub goers have reported that alcohol can make others look more attractive)
 
 

Level of literacy and number of books read? (The more books you read, the more literate you become)
 
 

Does a strong correlation mean that a change in one thing causes the other thing to change?
(Correlation is NOT causation.  There is just a "tendency" of the two to occur together.)
 

A high correlation between two sets of scores can be used to prove that two tests measure the same thing

The Smith Test of Intelligence                            The Jones Test of Intelligence
Rank order (among a group who took test together)                  Rank order
Student 1: 1st                                                                        Student 1: 1st
Student 2: 4th                                                                        Student 2: 3th
Student 3: 9th                                                                        Student 3: 9th
Student 4: 7th                                                                        Student 4: 7th
Student 5: 6th                                                                        Student 5: 6th
Student 6: 2nd                                                                       Student 6: 2nd
Student 7: 5th                                                                        Student 7: 5th
Student 8: 3rd                                                                        Student 8: 4rd
Student 9: 8th                                                                        Student 9: 8th


More on Correlation

  Shows the relationship between two different sets of scores

  Answers the question:   "Are high scores on one variable (thing) predictive of high scores on some other measure (or realm)?"

  Shows the degree to which two things change together
 (e.g., height & weight, distance from favorite person & happiness,
  Northerness in USA (between Mississippi and Atlantic Ocean and educational level)

  It's "going-togetherness"

  Degree of correspondence between two values

  The tendency of one score to vary or change when another one does so.
 
 

Is there a correlation (relationship) between
 (from birth) Physical growth and knowledge of the world?
(Yes.  As people grow, they tend to gain more information about the world)

 
 

The amount you drink and how good the other person in the room looks?
(So they say.)

 

 
Level of literacy and number of books read?
(Yep. One may not cause the other, but we certainly see this trend.)

 
 
 

Does a strong correlation mean that a change in one thing causes the other thing to change?
(Correlation is NOT causation.  There is just a "tendency" of the two to occur together.)

 

A high correlation between two sets of scores can be used to prove that two tests measure the same thing

The Smith Test of Intelligence         The Jones Test of Intelligence
Rank order (among a group who took test together)   Rank order
Student 1: 1st        Student 1: 1st
Student 2: 4th        Student 2: 3th
Student 3: 9th        Student 3: 9th
Student 4: 7th        Student 4: 7th
Student 5: 6th        Student 5: 6th
Student 6: 2nd        Student 6: 2nd
Student 7: 5th        Student 7: 5th
Student 8: 3rd        Student 8: 4rd
Student 9: 8th        Student 9: 8th
 
 


Quiz on Norms and Correlations
 

1. List three things that must be considered in determining if the "norms" (data on comparison groups) provided by a test are useful.
 

2. You are told that there is a correlation coefficient of +1.50    What can you tell from this information?
 
 

3. Which is the strongest connection/correlation coefficient?
+.46
+.93
-.1
-.96
 
 

4. The "line of best fit" might also be called the _____________ line.
 

5. Most alcoholics drank milk as children.  The correlation between the two
traits is very strong.  Therefore drinking milk places one at risk for
alcoholism.  Right?  Most heroin addicts smoked marijuana previously.
Marijuana use causes future heroin use.  Right?

(Nope: Correlation does not prove causation)
 
 
 
 
 
 Tom Mcintyre at www.BehaviorAdvisor.com