Question 764:

Asked on June 23, 2009

Tags: Correlation

1

Answer:

No answer provided yet.

While there could many factors at play here, the major factor is that when dealing with individuals or individual observations there is no guarantee of an exact relationship unless the correlation between the two variables is extremely high (r >.95). For example, if we assume the correlation between education score on some standardized test and years attended in school, we might assume this correlation to be between .4 and .7. By squaring these correlation coefficients we get the coefficient of determination. (.16 and .49) and it's called R-Squared. R-squared tells you how much one variable accounts for the variation in the other. As you can see, even a correlation as high as .7 only accounts for 49% of the variation. That means 51% of the variation is the education score is affected by factors other than their years in a formal school. This might be learning at home, genetics etc.