Example of testing for an outlier
main topic     interpreting results     session command     see also 

You receive data for the breaking strength of broom handles. You notice that one of the values in the sample seems unusually small. Before you analyze the data further, you use Outlier Test to determine whether the smallest value is an outlier.

1    Open the worksheet OUTLIER.MTW.

2    Choose Stat > Basic Statistics > Outlier Test.

3    In Variables, enter BreakStrength.

4    Click Options.

5    In What do you want to determine, choose Smallest data value is an outlier.

6    Click OK in each dialog box.

Session window output

Outlier Test: BreakStrength

 

 

Method

 

Null hypothesis         All data values come from the same normal population

Alternative hypothesis  Smallest data value is an outlier

Significance level      α = 0.05

 

 

Grubbs' Test

 

Variable        N   Mean  StDev   Min    Max     G      P

BreakStrength  14  123.4   46.3  12.4  193.1  2.40  0.044

 

 

Outlier

 

Variable       Row  Outlier

BreakStrength   10    12.38

Graph window output

 

Interpreting the results

The results for the broom handle data show that the mean of the sample is 123.4. The G statistic indicates that  the smallest data value, 12.38, is 2.4 standard deviations less than the mean. The p-value indicates that, if all values are truly from the same, normally distributed population, then the probability of obtaining a minimum value that small is only 0.044. This p-value is less than your chosen significance level, so you can reject the null hypothesis and conclude that the smallest value is an outlier.

You investigate further and discover that the person who entered the data accidentally entered 12.38 instead of 123.8.