What Does the Analysis of Variance Tell You?

 The analysis of variance test is the initial step in

analyzing factors that affect a given data set. Once the
analysis of variance test is finished, an analyst performs
additional testing on the methodical factors that
measurably contribute to the data set's inconsistency. The
analyst utilizes the ANOVA test results in an f-test to
generate additional data that aligns with the proposed
regression models.

 The ANOVA test allows a comparison of more than two

groups at the same time to determine whether a
relationship exists between them. The result of the ANOVA
formula, the F statistic (also called the F-ratio), allows for
the analysis of multiple groups of data to determine the
variability between samples and within samples.
When to use ANOVA
It is applied when data needs to be
experimental. Analysis of variance is
employed if there is no access to statistical
software resulting in computing ANOVA by
hand. It is simple to use and best suited for
small samples. With many experimental
designs, the sample sizes have to be the same
for the various factor level combinations.
One Way ANOVA
The one-way analysis of variance (ANOVA)
is used to determine whether there are
any statistically significant differences
between the means of three or more
independent (unrelated) groups. This
guide will provide a brief introduction to
the one-way ANOVA, including the
assumptions of the test and when you
should use this test.
When One Way Anova is used ?
You have a group of individuals randomly
split into smaller groups and completing
different tasks.

The individuals are split into groups based on

a attribute they possess.
Limitations of the One Way ANOVA
One Way ANOVA will tell you that at
least two groups were different from
each other. But it won’t tell you what
groups were differents.
Assumptions
 The dependent variable is normally distributed in each
group that is being compared in the one-way ANOVA
(technically, it is the residuals that need to be normally
distributed, but the results will be the same).

 There is homogeneity of variances. This means that the

population variances in each group are equal.

 Independence of observations. This is mostly a study

design issue and, as such, you will need to determine
whether you believe it is possible that your observations
are not independent based on your study design (e.g.,
group work/families/etc).
One way ANOVA
 Hypothesis
 The null hypothesis (H0) states that the population mean are
equal. The alternative hypothesis (H a) states that all population mean
differ.
 H0 : µ 1 = µ 2 = µ 3 = . . . = µ k
 Ha : all mean differ
 ONE-WAY ANOVA
 In the techniques of one-way ANOVA, random samples of size
n are selected from each of k populations. The k different
populations are classified on the basis of single criterion such as
different treatments or groups. In using this technique, one would
formulate opposing hypotheses such as:
 H0 : µ 1 = µ 2 = µ 3 = . . . = µ k
 Ha : at least one of the population means is different from the
others.
How to use One Way ANOVA
Results
Two Way ANOVA
 The two-way ANOVA compares the mean differences
between groups that have been split on two
independent variables (called factors).

 The primary purpose of a two-way ANOVA is to

understand if there is an interaction between the two
independent variables on the dependent variable.

 Two-way ANOVA informs you whether the effect of

one of your independent variables on the dependent
variable is the same for all values of your other
independent variable (and vice versa).
Assumptions
 Your dependent variable should be measured at
the continuous level (i.e., they
are interval or ratio variables).

 Your two independent variables should each consist

of two or more categorical, independent groups

 You should have independence of observations, which

means that there is no relationship between the
observations in each group or between the groups
themselves.
Assumptions
 There should be no significant outliers. Outliers are data
points within your data that do not follow the usual pattern
(e.g., in a study of 100 students' IQ scores, where the mean
score was 108 with only a small variation between students,
one student had a score of 156, which is very unusual, and
may even put her in the top 1% of IQ scores globally).

 Your dependent variable should be approximately normally

distributed for each combination of the groups of the two
independent variables.

 There needs to be homogeneity of variances for each

combination of the groups of the two independent variables.
Two Way ANOVA
Research question type: Explaining a continuous variable with 2 categorical
variables

What kind of variables? Continuous (scale/interval/ratio) and 2 independent

categorical variables (factors)

Common Applications: Comparing means of a single variable at different

levels of two conditions (factors) in scientific experiments.

Example:
The effective life (in hours) of batteries is compared by material type (1, 2 or 3) and
operating temperature: Low (-10˚C), Medium (20˚C) or High (45˚C). Twelve batteries
are randomly selected from each material type and are then randomly allocated to
each temperature level. The resulting life of all 36 batteries is shown below:
Example
Research question: Is there difference in mean life of the batteries for differing material type and
operating temperature levels?

In analysis of variance we compare the variability between the groups (how far apart are the
means?) to the variability within the groups (how much natural variation is there in our
measurements?). This is why it is called analysis of variance, abbreviated to ANOVA.
This example has two factors (material type and temperature), each with 3 levels.
Hypotheses:
The 'null hypothesis' might be:
 H0: There is no difference in mean battery life for different combinations of material type
and temperature level

And an 'alternative hypothesis' might be:

 H1: There is a difference in mean battery life for different combinations of material type and
temperature level

If the alternative hypothesis is accepted, further analysis is performed to explore where the
individual differences are.
Working in SPSS
Steps in SPSS (PASW):

Data need to be arranged in SPSS in a particular way to perform a two-way

ANOVA. The dependent variable (battery life) values need to be in one column, and
each factor needs a column containing a code to represent the different levels.

In this example Material has codes 1 to 3 for material type in the first column and
Temp has codes 1 for Low, 2 for Medium and 3 for High operating temperatures.

The battery life (Life) is entered in the third column – see screen to the left.

Note carefully how the data are entered.

Then choose:
Analyze > General Linear Model > Two-Way ANOVA…
 Transfer the outcome variable (Life in this example) into the Dependent Variable box, and
the factor variables (Material and Temp in this case) as the Fixed Factor(s)
 Click on Model… and select Full factorial to get the 'main effects' from each of the two
factors and the 'interaction effect' of the two factors. [It is possible to build a Custom
model, if you prefer]
 Continue

 Click on Plots…, and choose Temp for Horizontal Axis and Material in Separate Lines (see
right)
 Click Add and Continue

 Click on Post Hoc… and select Material and Temp

 Check Tukey (or post hoc test of choice)
 Continue

 Click on Options… and choose to Display Means for Material, Temp and Material*Temp
 Check Descriptive statistics and Homogeneity tests (see right)
 Continue and OK
the SPSS output there is a table
showing the descriptive statistics
for the main variable (battery life)
at each of the levels for each factor
(9 in this example), plus Totals.

Check the result of Levene's test for

a p-value (Sig.) > 0.05, so that
similar variances for each group of
measurements can be assumed
(otherwise the ANOVA is probably
invalid).

In the example, p = 0.529, so the

two-way ANOVA can proceed.
The Tests of Between Subjects Effects table gives the
results of the ANOVA. Table 2 below shows the output for the
battery example with the important numbers emboldened.

Table 2: Tests of Between-Subjects Effects

Dependent Variable: Battery life (in hours)
 Results:
 From the Descriptive Statistics table, it can
be seen that, overall, battery life decreases
with higher operating temperature,
although battery life remains high for
material 3 at medium
 temperature. This pattern is more obvious
when looking at the plot – see right.

 Since the lines representing the three

materials in the plot are not parallel, this
implies there is an interaction effect
between material and operating
temperature. [The lines would be
approximately parallel if there were no
interaction.] So, how battery life changes
with temperature depends on the material,
and vice versa.
The ANOVA table gives F statistics = 7.91, p=0.002; 28.97, p<0.001 and
3.56, p=0.019, for material, operating temperature and
material*temperature, respectively [NEVER write p = 0.000]. So, both
material and temperature are needed, as well as their interaction, to
explain battery life.

The nature of these differences can be explored further by looking at

the SPSS output from the 'post hoc' tests. These suggest that mean
battery life overall is statistically significantly longer for material 3 than
1 (p=0.001), and at lower compared to higher temperature levels.
However, at low operating temperatures material 2 appeared to give a
longer life than for materials 2 and 3, but lasted least at high
temperatures.
Conclusion:
From the results it can be said that there is strong evidence that the
mean battery life varies with material used and operating temperature
(p=0.002 and p<0.001). The presence of interaction between material
and temperature means that the way battery life changes for different
materials depends on the temperature. Similarly, the way battery life
changes for different temperatures depends on material. Overall,
material 3 performs best.

The three tables of estimated marginal means give details of mean

battery lives by factor, plus 95% CIs, giving more detail concerning the
accuracy of these battery life estimates from the sample to the overall
population. In this example, the CIs are all fairly 'wide', so results should
be used with caution.
Validity of two-way ANOVA:

ANOVA is based on two assumptions:

 the observations are random samples from normal distributions
 the populations have the same variance [variance = (standard
deviation)2]
 observations are independent of each other

So, before carrying out any tests the data must be examined in
more detail to determine whether these assumptions are satisfied.
Comments:

 Multiple t-tests should not be performed

 It is possible to perform two-way ANOVA with
different sample sizes per group. Select Type IV Sum
of squares in the Univariate: Model dialog box.
Results
Food Preferences and Dietary Intakes of Filipino
Adolescents in Metro Manila, The Philippines
Author: Magbuhat, Rizza Marie T, Borazon, Elaine Q
& Villarino, Blanca J

Purpose: This study examined differences in food

preferences and dietary intake among male and
female Filipino adolescents of different nutritional
status as measured by body mass index (BMI).

Results: Resulting values were analysed using one-

way ANOVA. Results suggest that food preferences
should be considered in the nutritional care
management of malnourished adolescents.