SPURIOUS CORRELATION AND IT'S DETECTION




CHURCH ATTENDANCE CAUSES DRINKING

In cross-sectional analysis, one would select a number of cities and for each city record the number of churches and the number of bars noting that the more churches that were in a city the more the number of bars. Curious, but true. # of # of city # churches bars 1 10 14 2 15 20 .... .... ..... n 23 36 In time series analysis, one would select one city and research the historical archives and determine what the number of churches and the number of bars were at different points in time. # of # of Year churches bars 1 20 29 2 29 41 .... .... ..... n 45 71 Perhaps a concomitant variable lurking in the back ground ? The concommitant variable could be population or even the ethnicity of the city. Perhaps the number of people of a certain type build more churches and bars ? If the true cause variable is brought into the model the apparaent association disappears.

IS VOCABULARY RELATED TO FOOT SIZE ?

Another example... the relationship between length of foot and number of words in vocabulary in the school-age (K-12) population. If you control for age, the correlation disappears...

WHY IS IT THAT STUDENTS WHO USE TUTORS HAVE LOWER

TEST SCORES THAN STUDENTS WHO DON'T ?

The answer is simply students with good test scores simply don't use tutors.


CLICK HERE:Home Page For AUTOBOX