Scribd
Upload
52 views3 pages

Q Test

Statistics Q test

Uploaded by

Karen Tagarisa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
52 views3 pages

Q Test

Statistics Q test

Uploaded by

Karen Tagarisa
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

The Q-test for the Outlying Data Point

Sometimes you collect data by repeated measurements that seem to cluster around a certain value,
except for one odd data point. You want to be honest; you really did get that measurement, and
you don’t want to cheat. However, it does seem to be from another universe. For example, suppose
these are the readings from your experiment:
25.1 21.2 27.5 22.7 23.8 26.3 40.6 22.9
That 40.6 not only looks out of place, but it has an effect on the mean and standard deviation of
your data set:
mean standard deviation
including maximum value 26.1 6.3
excluding maximum value 24.0 2.1
It may also have an effect on statistical significance when you compare this distribution to another
one by a t-test.
Is it a mistake? Is it noise? Did your recording device hiccup? What is the likelihood of getting
an outlier this far from the next value when you are picking from a population with these
characteristics?
Chemists (and other scientists) often test for outliers with the Q-test. This test calculates the ratio
between the putative outlier’s distance from its nearest neighbor and the range of values:
Notice that as the distance between the potential outlier and its nearest neighbor increases, so does
Q.
potential outlier − nearest neighbor
Q=
maximum value − minimum value

For our hypothetical data listed above,


potential outlier − nearest neighbor 40.6 − 27.5
Q€= = = 0.675
maximum value − minimum value 40.6 − 21.2

The significance test consists of comparing your calculated Q to the theoretical Q that is expected
to occur 5% of the time if you were sampling from a population with this mean and standard
deviation. The € null hypothesis is that all your data points come from the same pool. If your Q is
as big as or bigger than the critical Q, then you can reject that null hypothesis with 95% confidence,
and safely exclude the one odd data point.
Table 1, on page 2, lists critical values of Q at the 5% significance level. Notice that the smaller
the number of data points, the larger Q must be for you to reject a data point. For a set of 8 data
points, according to the table, Q would be greater than 0.526 just by luck of the draw 5% of the
time or less. Since our experimental Q (0.675) is larger than 0.526, we may safely omit the 40.6
value from our data set, because such a large value is unlikely to occur by chance alone. We can
conclude that that data point measures something other than what all the others are measuring. To
maintain honesty, we can report something like this: “One very large (or small) value was rejected
from the data set by the Q-test with 95% confidence.”
You may only reject one data point from a data set by this method.
2 Q-Test for the Outlying Data Point

Excel Tricks to help you calculate Q


Sort
Select your data, then choose Sort from the Data menu. This makes visual inspection for outliers
easier. Note: if there are identifiers adjacent to the data values, be sure to include them in the
sort, or you will lose the connection between the labels and the values!
MAX
From the fx menu, under statistical. This returns the maximum value from a list of values. The
result goes in the cell where you typed the equation. You can specify the data array either by
typing the range of cells where your data are found, or by clicking and dragging over them.
=MAX(data array)
MIN
This function works just like MAX, but tells you the minimum value in a series.
=MIN(data array)
COUNT
This tells you how many numerical values there are in an array. (If you use COUNTA, it will
return the number of non-empty cells, rather than the number of data points.)
=COUNT(data array).
LARGE
This function tells you the kth largest value in a series. Set k = 2 to find the second largest value.
Set k = n-1 to find the next-to-last value in a set of n data points.
=LARGE(data array, k)
SMALL
This function tells you the kth smallest value in a series. Set k = 2 to find the second smallest
value.
=SMALL(data array, k)
ABS
This function returns the absolute value of a number. The number may be the result of a
calculation, or may refer to a particular cell in which a calculation is done.
=ABS(number)
=ABS(number1+number2)
Q-Test for the Outlying Data Point 3

Table 1. Critical values of Q


at the 95% confidence ( a = 0.05) level, for data sets up to n = 30
number of
data points1 Q0.05
3 0.970
4 0.829
5 0.710
6 0.625
7 0.568
8 0.526
9 0.493
10 0.466
11 0.444
12 0.425
13 0.410
14 0.396
15 0.384
16 0.374
17 0.365
18 0.356
19 0.349
20 0.342
21 0.337
22 0.331
23 0.326
24 0.321
25 0.317
26 0.312
27 0.308
28 0.305
29 0.301
30 0.298

1
Table from David B. Rorabacher, 1991. Statistical treatment for rejection of deviant values: critical values of
Dixon's "Q" parameter and related subrange ratios at the 95% confidence level. Anal. Chem., 63 (2): 139–146
http://pubs.acs.org/doi/abs/10.1021/ac00002a010, accessed 9/10/09.

You might also like