Scribd
Upload
90 views3 pages

How To Interpret An Anova Table

Uploaded by

alyzatorinueva
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
90 views3 pages

How To Interpret An Anova Table

Uploaded by

alyzatorinueva
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

How to interpret an ANOVA table?

ANOVA involves the calculation and interpretation of a number of parameters


which are all summarized in a table. In practice, the calculations are best
carried out using statistical software or a MS Excel® spreadsheet. Before
going into details of calculations from the first principle, let’s take a look at a
typical output from a one-way ANOVA in a results table form, whether
manual or using software. We will also see how the results are interpreted.

The general form of a results table from a one-way ANOVA, for a total of N
observations in k groups is shown in Table 1 below.

Table 1: Results table from one-way analysis of variance

Source of Sum of Degrees of


Mean square MS F
variation squares SS freedom, v
Between-group SSb k-1 MSb=SSb/(k-1) MSb/MSw
Within-group SSw N-k MSw=SSw/(N-k)
Total SST=SSb+SSw N-1

The above table shows 3 rows relating to different sources of variation and a
number of columns containing calculated values related to each source of
variance.

Row 1 relates to variation between the means of the groups; the values are
almost always either referred to as “between group” terms or are identified
by the ‘grouping factor’. For example, data from different analysts are
grouped by ‘analyst’. We can also be more specific and label it as an
“analyst effect” or “between-analyst”.

Row 2 refers to variation within each group (or analyst, for example),
meaning the sample standard deviation of each group as compared with the
overall performance of other group members. Several different terms may be
used to describe this within group variation, “within-group”, “residual”,
“error”, or “measurement” being the most popular.

Row 3 on “Total” is not always given by software, but it is fairly consistently


labelled “Total” when present.
1
Under the column section of the ANOVA table, we have the following
subjects:

Sum of squares (SS)

The SS terms are calculated by adding a series of squared error terms.

In the case of within-group SS term SSw, we are interested in the differences


between the individual data points and the mean of the group to which they
belong.

For the Total SS term, it is the difference between the individual data points
and the mean of all the data (the “grand or overall mean”) that is of interest.

The between-group SS term is conveniently the difference between the Total


SS and the within-group SS.

Degrees of freedom v

For the one-way ANOVA in the table 1, the total number of data points is N
and the number of groups of data is k. The total number of degrees of
freedom is N – 1, just as for a simple data set of size N.

There are k different groups and therefore, k – 1 degrees of freedom for the
between-group effect.

The degrees of freedom associated with the within-group SS term is the


difference between the two values, N – k.

Of course, if each group of data contains the same number of replicates, n,


then the degrees of freedom for the within-group SS can be simplified to be
equal to k(n – 1) and the total number of observations is N – kn.

Mean squares MS

The mean squares are the key term in classical ANOVA. They are variances,
calculated by dividing the between- and within-group sum of squares by the
appropriate number of degrees of freedom. In Table 1, Mb represents the
between-group MS term (sometimes denoted M1) and Mw represents the
within-group MS term (sometimes denoted M0).

The mean squares are the values used in the subsequent test for significant
difference between the group means.

2
These mean squares also allow estimation of the variance components, i.e.
the separate variances for each different effect that contributes to the overall
dispersion of the data. We shall discuss how this is being done in the
coming blog.

F (or F Ratio)

The mean squares are compared using an F-test as indicated in the above
table. The hypothesis for the F -test in Table 1 are:

Ho : MSb = MSw
H1 : MSb > MSw

If all means are ‘equal’ (or not significantly different), the two mean squares
should also be not significantly different and hence Ho is true.

We expect MSb to be equal or greater than MSw as it has included an extra


element of variance between group and variance cannot be negative.
Therefore this F-test is a one-tailed test for whether MSb is greater than MSw
with formula MSb/MSw. This is the value shown in the column F in Table 1.

No value for F is given for the residual mean square, as there is no other
effect with which it can usefully be compared.

Fcrit and p-Value

Many ANOVA tables include one or two further columns, containing the
critical value Fcrit against which the calculated value F is to be compared for a
chosen significance level, and a p-value including the significance of the test.
These are important information for interpreting the outcome. For example if
a chosen significance level is set at 0.05, then a calculated F value which is
less than Fcrit at this significance level or a calculated p-value which is more
than 0.05 would indicate that the MSb is not significantly different from MSw
and hence, the mean values given by the group are not significantly different
from each other.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy