WHAT A CORRELATION CAN TELL YOU(AND WHAT IT DOESN'T)

A correlation tells you that two variables are related. For example, if you know that income and reading achievement are positively related (which they are, in fact), then knowing that an area has a lot of very low income families will let you reasonably assume that there will be more children with low reading achievement, a greater need for programs that are designed for students with reading difficulties, more need in the library for low reading-level high interest books for youth, etc.

A correlation does NOT tell you that one thing causes another.
A CORRELATION DOES NOT TELL YOU THAT ONE VARIABLE CAUSES ANOTHER!

The above statement is one of the most frequently misunderstood in all of social science. Let me give an example:

This is a fact. Trust me, I have won money betting people on this. So, given this fact, what should we do? If we were stupid, which we are not, we might run out and buy "Billy-Bob Schwartz sure-fire guaranteed foot growth program" at a considerable cost to our school district because, after all, who could possibly be against helping our children read better?

Whenever you have a correlation, you have three possibilities:

1. X causes Y. In this case, having bigger feet causes you to be a better reader.
2. Y causes X. It could be that reading better makes your feet grow. Maybe all of that time spent off your feet while reading allows them to grow better. This explanation doesn't seem very likely.
3. Z (some other variable) causes both X and Y.

The most likely explanation for this correlation is that AGE is a factor in both shoe size and reading achievement. Kindergarteners, who generally don't read at all, have very small feet, third-graders, who, on the average read at the third-grade level, have bigger feet than kindergarteners, and, sixth-graders, who have even bigger feet, read at the sixth grade level.
This is important to keep in mind, because psychology involves making judgements about people, and I happen to think that we owe it to the clients, students and families we work with to treat them fairly. An example of misinterpretation of correlation occurs later in your textbook, when they discuss intelligence test scores. The book does wimp out, though, and avoid mentioning the major controversy involving intelligence tests, which is the argument whether minorities (African-Americans, Latinos and Native Americans) are genetically less intelligent because they tend to score lower on intelligence tests. So, it could be that being born African-American, etc. makes you less intelligent, or, it could be that minority children tend more often to be poor and that poor children do worse on intelligence test because they go to worse schools, have fewer books in the home, less access to computers, parents with less education who are less able to help them with their schoolwork, etc. etc. The second explanation seems more plausible to me.

So, what good is correlation if it does not tell you that one thing causes another? There are two uses. Remember, one of the three possibilities above is that there COULD be a causal relationship. So, if you find that children who are read books every day become better readers, it is worth a try to read to your children every day, especially if you find a THEORETICAL relationship explaining why reading to children might help them. (Note that this is a sneaky way of hinting that the next chapter on theories could be important.)

Second, knowing that there is correlation between two variables does help you predict. If you know that child abuse is correlated with behavior problems, and you will be working with a group of children who have been abused, then you can prepare behavior modification plans (also discussed in the next chapter, which is on theories ) read up on behavior problems and recommended treatment programs, etc. Knowing that there is a correlation between child abuse and behavior problems doesn't mean that a specific student, John Red Horse, who was abused by his parents will have behavior problems, but it does mean that more of the students in a class of children who had been abused will have behavior problems than in the average class. Don't worry that it doesn't tell you which ones, you'll know when you come in to class the first day and see Nicky setting the guinea pig on fire. (I 'm just KIDDING!)

Two questions on correlation:

1. Which of the following are most likely to show a negative correlation?
        (a) family income and school grades
        (b) number of children in a classroom and achievement
        (c) frequency of spanking and behavior problems
        (d) weight and self-esteem
        (e) both b and d

2. There is a positive correlation between mothers' educational level and the reading achievement of their children. This tells you that:
    (a) children who have mothers with college degrees tend to read better than children of high school dropouts.
    (b) if you complete your college degree, your own children will read better.
    (c) children's reading achievement is partly caused by their mothers' years of education.
    (d) absolutely nothing.

Click here to send me the answers to the questions above, or just to provide general information on how your life is going.

Okay, I'm excited, I can't wait to go to the next page on theories of development, which I will prove by
clicking right here!