# Basic Stat- outlier detection

• Mar 18th 2010, 09:46 PM
tom6030
Basic Stat- outlier detection
Hi... im taking a course that is fundamental of statistic, so the class only taught us only 2 method to determine outlier.

1) IQR (IQR = 75th percentile - 25th percentile)
2) observation of a scatter-plot, is it close along the correlation line

In the test, I got this question:

Height of Couples:
Woman 64 65 65 66 66 70
Man 68 68 69 70 72 74

Is any height above 70 inches must be considered as an outlier?

Since its a 2-way table, so i drew a scatter-plot and adding a description saying that as long as the point falls along the correlation, the height does not matter as long as the height is correlated.

But i end up got no marks for that question, the teacher explain to me I should use IQR on each data to prove its not an outlier. Im doubt using this method because it cannot measure the correlation between the data sets. Could anyone give me some explanation proving Im wrong or the teacher is wrong.

Thanks!