ANOVA
Introduction
Questions
- What null hypothesis is tested by ANOVA?
- What are the uses of ANOVA?
What is ANOVA?
- Analysis of Variance (ANOVA) is a statistical method used to compare two or more means
- Inferences about means are made by analyzing variance (therefore it is not called analysis of mean)
- ANOVA is used to test general rather than specific differences among means
Smiles and Leniency Example
- Let us investigate types of smiles on the leniency
- Types of smiles: neutral, false, felt, miserable
The results from the Tukey hsd test (Six Pairwise Comparisons):
Comparison | Mi-Mj | Q | p |
---|---|---|---|
False - Felt | 0.46 | 1.65 | 0.649 |
False - Miserable | 0.46 | 1.65 | 0.649 |
False - Neutral | 1.25 | 4.48 | 0.010 |
Felt - Miserable | 0.00 | 0.00 | 1.000 |
Felt - Neutral | 0.79 | 2.83 | 0.193 |
Miserable - Neutral | 0.79 | 2.83 | 0.193 |
- Notice that the only significant difference is between the False and Neutral conditions.
- ANOVA tests the non-specific null hypothesis that all four populations means are equal
μfalse = μfelt = μmiserable = μneutral
- This non-specific null hypothesis is sometimes called the omnibus null hypothesis
- When the omnibus null hypothesis is rejected, the conclusion is that at least one population mean is different from at least one other mean
- ANOVA does not reveal which means are different from which
- It offers less specific information than the Tukey hsd test
- The Tukey hsd is therefore preferable to ANOVA in this situation
Why to use ANOVA instead of HSD Tukey
- There are complex types of analyses that can be done with ANOVA and not with the Tukey test
- ANOVA is by far the most commonly-used technique for comparing means
- Is important to understand ANOVA in order to understand research reports.
Questions
ANOVA designs
Questions
- What are factors and levels of each factor?
- What is a between-subjects or a within-subjects factor
- What is factorial design
- What kind of types of experimental designs can be analyzed by ANOVA?
Factors and Levels
- In the case study Smiles and Leniency, the effect of different types of smiles on the leniency showed to a person was investigated
- Type of smile (neutral, false, felt, miserable) is independent variable or factor
- "Type of Smile" has four levels
Single or Multi factor ANOVA
- An ANOVA conducted on a design in which there is only one factor is called a one-way ANOVA
- If an experiment has two factors, then the ANOVA is called a two-way ANOVA
Suppose an experiment on the effects of age and gender on reading speed were conducted using three age groups (10 yr, 15 yr, and 20 yr) and the two genders (males and females)
What are the factors and levels in this experiment?
Answer >>
The factors would be age and gender. Age would have three levels and gender would have two levels
Between- and Within-Subject Factors
Between-subjects
- In the Smiles and Leniency study, the four levels of the factor "Type of Smile" were represented by four separate groups of subjects
- When different subjects are used for the levels of the factor, the factor is called a between-subjects factor or a between-subjects variable
- The term "between subjects" reflects the fact that comparisons are between different groups of subjects.
Within-subjects
- In the ADHD Treatment Study, every subject was tested with each of four dosage levels (0, 0.15, 0.30, 0.60 mg/kg) of a drug
- Therefore there was only one group of subjects and comparisons were not between different groups of subjects but between conditions within the same subjects
- When the same subjects are used for the levels of the factor, the factor is called a within-subjects factor or within-subjects variable
- Within-subjects variables are sometimes referred to as repeated-measures variables since there are repeated measurements of the same subjects.
Multi-Factor Designs
- Consider a hypothetical study of the effects of age and gender on reading speed in which males and females from the age levels of 8 years, 10 years, and 12 years were tested.
- There would be a total of six different groups:
Group | Gender | Age |
---|---|---|
1 | Female | 8 |
2 | Female | 10 |
3 | Female | 12 |
4 | Male | 8 |
5 | Male | 10 |
6 | Male | 12 |
- This design has two factors: age and gender
- age has three levels
- gender has two levels
- When all combinations of the levels are included (as they are here) the design is called a factorial design
- A concise way of describing this design is as a Gender (2) x Age (3) factorial design where the numbers in parentheses indicate the number of levels
- Complex designs frequently have more than two factors and may have combinations of between- and within-subject factors.
Questions
One-Factor ANOVA (Between-Subjects)
Questions
- What does the Mean Square Error estimates when the null hypothesis is true and when the null hypothesis is false?
- What does the Mean Square Between estimates when the null hypothesis is true and when the null hypothesis is false?
- What are the assumptions of a one-way ANOVA
- How to compute MSB
- How to compute MSE
- How to compute F and the two degrees of freedom parameters
- Describe the shape of the F distribution
- Why is ANOVA best thought of as a two-tailed test even though literally only one tail of the distribution is used
- What is the relationship between the t and F distributions
- How to partition the sums of squares into conditions and error
One-factor between-subjects design
- In Smiles and Leniency case study there were four conditions with 34 subjects in each condition
- There was one score per subject
- The null hypothesis tested by ANOVA is that the population means for all conditions are the same.
H0: μfalse = μfelt = μmiserable = μneutral
If the null hypothesis is rejected, then it can be concluded that at least one of the population means is different from at least one other population means.
Analysis of variance
- Analysis of variance is a method for testing differences among means by analyzing variance
The test is based on two estimates of the population variance (σ2)
Mean Square Error (MSE)
- based on differences among scores within the groups
- MSE estimates σ2 regardless of whether the null hypothesis is true (the population means are equal)
Mean Square Between (MSB)
- based on differences among the sample means
- MSB only estimates σ2 if the population means are equal
- If the population means are not equal, then MSB estimates a quantity larger than σ2
MSB and MSE
- If the MSB is much larger than the MSE, then the population means are unlikely to be equal
- If the MSB is about the same as MSE, then the data are consistent with the hypothesis that the population means are equal.
ANOVA Assumptions
- The populations have the same variance (homogeneity of variance)
- The populations are normally distributed
- Each value is sampled independently from each other value
- The last assumption requires that each subject provide only one value
- If a subject provides two scores, then the value are not independent
- The analysis of data with two scores per subject is shown in the section on within-subjects ANOVA
These assumptions are the same as for a t test of differences between groups except that it applies to two or more groups, not just to two groups.
- The means and variances of the four groups in the Smiles and Leniency case study are shown below
- There are 34 subjects in each of the four conditions (False, Felt, Miserable, and Neutral).
Means and Variances from Smiles and Leniency Study.
Condition | Mean | Variance |
---|---|---|
False | 5.376 | 3.3380 |
Felt | 4.9118 | 2.8253 |
Miserable | 4.9118 | 2.1132 |
Neutral | 4.1176 | 2.3191 |
Sample Sizes
- The first calculations in this section all assume that there is an equal number of observations in each group
- Unequal sample size calculations are covered later
- We will refer to the number of observations in each group as n
- The total number of observations as N
- For these data there are four groups of 34 observations
- Therefore n = 34 and N = 136
Computing MSE
- The assumption of homogeneity of variance states that the variance within each of the populations (σ2) is the same
- This variance, σ2, is the quantity estimated by MSE and is computed as the mean of the sample variances
- For these data, the MSE is equal to 2.6489
Computing MSB
The formula for MSB is based on the fact that the variance of the sampling distribution of the mean is
where n is the sample size. Rearranging this formula we have
- Therefore, if we knew the variance of the sampling distribution of the mean, we could compute σ2 by multiplying by n.
- Although we do not know the variance of the sampling distribution of the mean, we can estimate it with the variance of the sample means. For the leniency data, the variance of the four sample means is 0.270
- To estimate σ2, we multiply the variance of the sample means (0.270) by n (the number of observations in each group, which is 34)
- We find that MSB = 9.179.
To sum up these steps:
- Compute the means
- Compute the variance of the means
- Multiply by the variance of the means by n
Recap
- If the population means are equal, then both MSE and MSB are estimates of σ2 and should therefore be about the same
- Naturally, they will not be exactly the same since they are just estimates and are based on different aspects of the data:
- MSB is computed from the sample means
- MSE is computed from the sample variances.
- If the population means are not equal, then MSE will still estimate σ2 because differences in population means do not affect variances
- However, differences in population means affect MSB since differences among population means are associated with differences among sample means
- It follows that the larger the differences among sample means, the larger the MSB
- In short, MSE estimates σ2 whether or not the populations means are equal whereas MSB estimates σ2 only when the population means are equal and estimates a larger quantity when they are not equal.
Comparing MSE and MSB
- MSB estimates a larger quantity than MSE only when the population means are not equal
- Therefore finding of a larger MSB than an MSE is a sign that the population means are not equal
- MSB could be larger than MSE by chance even if the population means are equal
- MSB must be much larger than MSE in order to justify the conclusion that the population means differ
- But how much larger must MSB be?
- For the Smiles and Leniency data, the MSB and MSE are 9.179 and 2.649 respectively
- Is that difference big enough?
- To answer, we would need to know the probability of getting this big a difference or a bigger difference between if the population means were all equal
- The mathematics necessary to answer this question were worked out by the statistician Ronald Fisher
F Ratio
- Although Fisher's original formulation took a slightly different form, the standard method for determining the probability is based on the ratio of MSB to MSE
- This ratio is named after Fisher and is called the F ratio.
F = MSB/MSE
For these data, the F ratio is
F = 9.179/2.649 = 3.465
- MSB is 3.465 times higher than MSE
- Would this have been likely to happen if all the population means were equal?
- That depends on the sample size
- With a small sample size, it would not be too surprising because small samples are unreliable
- However, with a very large sample, the MSB and MSE are almost always about the same, and an F ratio of 3.465 or larger would be very unusual
Sampling distribution of F for the sample size in the Smiles and Leniency study
- As you can see, it has a positive skew
- For larger sample sizes, the skew is less
F distribution and interpretation
- From the figure above you can see that F ratios of 3.465 or above are unusual occurrences
- The area to the right of 3.465 represents the probability of an F that large or larger and is equal to 0.018
- In other words, given the null hypothesis that all the population means are equal, the probability value is 0.018 and therefore the null hypothesis can be rejected
- Therefore, the conclusion that at least one of the population means is different from at least on of the others is justified
F distribution and sample size
- The shape of the F distribution depends on the sample size
- More precisely, it depends on two degrees of freedom (df) parameters:
- one for the numerator (MSB)
- one for the denominator (MSE)
Recall that the degrees of freedom for an estimate of variance is equal to the number of scores minus one. Since the MSB is the variance of k means, it has k-1 df. The MSE is an average of k variances each with n-1 df. Therefore the df for MSE is k(n-1) = N-k where N is the total number of scores, n is the number in each group, and k is the number of groups. To summarize:
dfnumerator = k-1 dfdenominator = N-k
For the Smiles and Leniency data,
dfnumerator = k-1 = 4-1 = 3 dfdenominator = N-k = 136-4 = 132 F = 3.465
The F distribution calculator shows that p = 0.018
One-Tailed or Two?
- Is the probability value from an F ratio a one-tailed or a two-tailed probability?
- In the literal sense, it is a one-tailed probability since, as you can see in figure above, the probability is the area in the right-hand tail of the distribution
- However, the F ratio is sensitive to any pattern of differences among means
- It is therefore a test of a two-tailed hypothesis and is best considered a two-tailed test.
Relationship to the t test
- Both ANOVA and an independent-group t test can test the difference between two means
- Results will always be the same
- When there are only two groups the following relationship between F and t will always hold:
F(1,dfd) = t2(df) dfd is the degrees of freedom for the denominator of the F test and df is the degrees of freedom for the t test dfd will always equal df.
Sources of Variation
- Why do scores in an experiment differ from one another?
- Consider the scores of two subjects in the Smiles and Leniency study:
- one from the "False Smile" condition
- one from the "Felt Smile" condition
Possible reasons that the scores could differ:
- the subjects were treated differently (they were in different conditions and saw different stimuli)
- the two subjects may have differed with regard to their tendency to judge people leniently
- one of the subjects was in a bad mood after receiving a low grade on a test
- innumerable other reasons
Unexplained Variance
All of these reasons except the first (subjects were treated differently) are possibilities that were not under experimental investigation and therefore all of differences (variation) due to these possibilities are unexplained
It is traditional to call unexplained variance error even though there is no implication that an error was made
Therefore, the variation in this experiment can be thought of as being either:
- variation due to the condition the subject was in
- due to error (the sum total of all reasons subjects's scores could differ that were not measured).
SSQ and GM
- ANOVA partitions the variation into its various sources
- The term sums of squares is used to indicate variation
- The total variation is defined as the sum of squared differences from the mean of all subjects
- The mean of all subjects is called the grand mean' and is designated as GM
= Sum of Squares Total
- The total sum of squares (SSQtotal or SST) is defined as
which means simply to take each score, subtract the grand mean from it, square the difference, and then sum up these squared values.
- For the Smiles and Leniency study, SSQtotal = 377.19.
Sum of Squares Conditions
The sum of squares conditions is calculated as shown below:
- n is the number of scores in each group
- k is the number of groups
- M1 is the mean for Condition 1
- M2 is the mean for Condition 2
- Mk is the mean for Condition k
Smiles and Leniency study, the values are:
SSQcondition = 34(5.37-4.83)2 + (4.91-4.83)2 + (4.91-4.83)2 + (4.12-4.83)2 = 27.5
If there are unequal sample sizes, the only change is that the following formula is used for the sum of squares for condition:
where ni is the sample size of the ith condition. SSQtotal is computed the same way as shown above.
Sum of Squares error
The sum of squares error is the sum of the squared deviations of each score from its group mean. This can be written as
where Xi1 is the ith score in group 1 and M1 is the mean for group 1, Xi2 is the ith score in group 2 and M2 is the mean for group 2, etc
For the Smiles and Leniency study, the means are: 5.38, 4.91, 4.91, and 4.12. The SSQerror is therefore:
(2.5-5.38)2 + (5.5-5.38)2 + ... + (6.5-4.12)2 = 349.66
The sum of squares error can also be computed by subtraction:
SSQerror = SSQtotal - SSQcondition SSQerror = 377.19 - 27.53 = 349.66
Therefore, the total sum of squares of 3771.9 can be partitioned into SSQcondition (27.53) and SSQerror (349.66).
Once the sums of squares have been computed, the mean squares (MSB and MSE) can be computed easily. The formulas are:
MSB = SSQcondition/dfn
where dfn is the degrees of freedom numerator and is equal to k-1.
MSB = 27.5/3 = 9.17
which is the same value of MSB obtained previously (except for rounding error). Similarly,
MSE = SSQerror/dfd
where dfd is the degrees of freedom for the denominator and is equal to N-k
dfd = 136 - 4 = 132 MSE = 349.66/132 = 2.65
which is the same as obtained previously (except for rounding error) Note that the dfd are often called the dfe for degrees of freedom error.
The Analysis of Variance Summary Table
- The table is a convenient way to summarize the partitioning of the variance
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Condition | 3 | 27.5349 | 9.1783 | 3.465 | 0.0182 |
Error | 132 | 349.6544 | 2.6489 | ||
Total | 135 | 377.1893 |
- Mean squares (MS) are always the sums of squares divided by degrees of freedom
Questions
One-Way Demo
simulations/anova1/anova1.html Template:Statistics Links
Multi-Factor ANOVA (Between-Subjects)
Learning Objectives
- Define main effect, simple effect, interaction, and marginal mean
- State the relationship between simple effects and interaction
- Compute the source of variation and df for each effect in a factorial design
- Plot the means for an interaction
- Define three-way interaction
Basic Concepts and Terms
Bias Against Associates of the Obese Example
- In the Bias Against Associates of the Obese case study, the researcher was interested in whether the weight of a companion of a job applicant would affect judgments of a male applicant's qualifications for a job
- Two independent variables were investigated:
- whether the companion was obese or of typical weight
- whether the companion was a girl friend or just an acquaintance
- One approach would have been to conduct two separate studies, one with each independent variable
- However, it is more efficient to conduct one study that includes both independent variables
- Moreover, there is a much bigger advantage than efficiency for including two variables in the same study:
- it allows a test of the interaction between the variables.
Interation
- There is an interaction when the effect of one variable differs depending on the level of a second variable
- For example, it is possible that the effect of having an obese companion would differ depending on the relationship to the companion
- Perhaps there is more prejudice against a person with an obese companion if the companion is a girl friend than if she is just an acquaintance
- If so, there would be an interaction between the obesity factor and the relationship factor
Main Effect
There are three effects of interest in this experiment:
- Weight: Are applicants judged differently depending on the weight of their companion?
- Relationship: Are applicants judged differently depending on their relationship with their companion?
- Weight x Relationship Interaction: Does the effect of weight differ depending on the relationship with the companion?
- The first two effects (Weight and Relationship) are both main effects.
- A main effect of an independent variable is the effect of the variable averaging over the levels of the other variable(s)
Marginal Means
- It is convenient to talk about main effects in terms of marginal means
- A marginal mean for a level of a variable is the mean of the means of all levels of the other variable
- For example, the marginal mean for the level "Obese" is the average of "Girl-Friend-Obese" and "Acquaintance-Obese."
- Table 1 shows that this marginal mean is equal to the average of 5.65 and 6.15 which is 5.90
- Similarly, the marginal mean for Typical is the average of 6.19 and 6.59 which is 6.39
- The main effect of Weight is based on a comparison of the these two marginal means
- Similarly, the marginal means for Girl Friend and Acquaintance are 5.92 and 6.37.
Companion Weight | ||||
---|---|---|---|---|
Obese | Typical | Marginal Mean | ||
Relationship | Girl Friend | 5.65 | 6.19 | 5.92 |
Acquaintance | 6.15 | 6.59 | 6.37 | |
Marginal Mean | 5.90 | 6.39 |
Simple Effect
- The simple effect of a variable is the effect of the variable at a single level of another variable
- The simple effect of "Weight" at the level of "Girl Friend" is the difference between the "Girl-Friend Typical" and the "Girl-Friend Obese" conditions
- The difference is 6.19-5.65 = 0.54
- Similarly, the simple effect of "Weight" at the level of "Acquaintance" is the difference between the "Acquaintance Typical" and the "Acquaintance Obese" conditions
- The difference is 6.59-6.15 = 0.44.
Interaction and Simple Effect
- There is an interaction when the effect of one variable differs depending on the level of another variable.
- This is equivalent to saying that there is an interaction when the simple effects differ
- In this example, the simple effects are 0.54 and 0.44
- As shown below, these simple effects are not significantly different.
Tests of Significance
- The important questions are not whether there are main effects and interactions in the sample data.
- Instead, what is important is what the sample data allow you to conclude about the population
- This is where Analysis of Variance comes it
- ANOVA tests main effects and interactions for significance
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Weight | 1 | 10.4673 | 10.4673 | 6.214 | 0.0136 |
Relation | 1 | 8.8144 | 8.8144 | 5.233 | 0.0234 |
W x R | 1 | 0.1038 | 0.1038 | 0.062 | 0.8043 |
Error | 172 | 289.7132 | 1.6844 | ||
Total | 175 | 310.1818 |
Weight Analysis
- Consider first the effect of "Weight."
- The degrees of freedom (df) for "Weight" is 1
- The degrees of freedom for a main effect is always equal to the number of levels of the variable minus one
- Since there are two levels of the "Weight" variable (typical and obese) the df is 2 -1 = 1
- The mean square (MS) is the sum of squares divided by the df
- The F ratio is computed by dividing the MS for the effect by the MS for error (MSE)
- For the effect of "Weight," F = 10.4673/1.6844 = 6.214
- The last column, p, is is the probability of getting an F of 6.214 or larger given that there is no effect of weight in the population
- The p value is 0.0136 which is quite low and therefore the null hypothesis of no main effect of "Weight" is rejected
- The conclusion is that the weight of the companion lowers judgments of qualifications
Relation Analysis
- The effect "Relation" is interpreted the same way
- The conclusion is that being accompanied by a girl friend leads to lower ratings than being accompanied by an acquaintance.
Weight x Relation interaction
- The df for an interaction is the product of the df of variables in the interaction
- For the Weight x Relation interaction (W x R), the df = 1 since both Weight and Relation have one df: 1 x 1 = 1
- The p value for the interaction is 0.8043 which is the probability of getting an interaction as big or bigger than the one obtained in the experiment if there were no interaction in the population
- Therefore, these data provide no evidence for an interaction
- Always keep in mind that the lack of evidence for an effect does not justify the conclusion that there is no effect
- In other words, you do not accept the null hypothesis just because you do not reject it.
Error analysis
- For "Error," the degrees of freedom is equal to the total number of observations minus the total number of groups
- The sample sizes for this experiment are shown below
- The total number of observations is 40 + 42 + 40 + 54 = 176
- Since there are four groups, df = 176 - 4 = 172.
Companion Weight | |||
---|---|---|---|
Obese | Typical | ||
Relationship | Girl Friend | 40 | 42 |
Acquaintance | 40 | 54 |
Total Analysis
- The final row is "Total."
- The degrees of freedom total is equal to the sum of all degrees of freedom
- It is also equal to the number of observations minus 1, or 176 -1 = 175
- When there are equal sample sizes, the sum of squares total will equal the sum all other sums of squares
- However, when there are unequal sample sizes, as there are here, this will not generally be true
- The reasons for this are complex and are discussed in the section Unequal n .
Plotting Means
- Although the plot shown below illustrates the main effects and the interaction (or lack of interaction) clearly, it is called an interaction plot
- It is important to carefully consider the components of this plot
- First, the dependent variable is on the Y axis
- Second, one of the independent variables is on the X axis
- In this case, it is the variable is "Weight."
- Finally, a separate line is drawn for each level of the other independent variable
- It is better to label the lines right on the graph as shown here than with a legend.
- If you have three or more levels on the X axis, you should not use lines unless there is some numeric ordering to the levels
- If your variable on the X axis is a qualitative variable, you can use a plot such as the one below
- However, as discussed in the section on bar charts, it would be better to replace each bar with a box plot.
- Line graphs are a common option with more than two levels if the variable is numeric
- A line graph has the advantage of showing the pattern of interaction clearly
- Its disadvantage is that it does not convey the distributional information contained in box plots.
An Example with Interaction
The following example was presented in the section on specific comparisons among means. It is also relevant here.
- Twelve subjects were selected from a population of high-self-esteem subjects and an additional 12 subjects were selected from a population of low-self-esteem subjects
- Subjects then performed on a task and (independent of how well they really did) half were told they succeeded and the other half were told they failed
- Therefore there were six subjects in each esteem/success combination and 24 subjects altogether.
After the task, subjects were asked to rate (on a 10-point scale) how much of their outcome (success or failure) they attributed to themselves as opposed to being due to the nature of the task.
Esteem | |||
---|---|---|---|
High | Low | ||
Outcome | Success | 7 | 6 |
8 | 5 | ||
7 | 7 | ||
8 | 4 | ||
9 | 5 | ||
5 | 6 | ||
Failure | 4 | 9 | |
6 | 8 | ||
5 | 9 | ||
4 | 8 | ||
7 | 7 | ||
3 | 6 |
The ANOVA Summary Table for these data is shown in Table below.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Outcome | 1 | 0.0417 | 0.0417 | 0.0256 | 0.8744 |
Esteem | 1 | 2.0417 | 2.0417 | 1.2564 | 0.2756 |
O x E | 1 | 35.0417 | 35.0417 | 21.5641 | 0.0002 |
Error | 20 | 32.5000 | 1.6250 | ||
Total | 23 | 69.6250 |
As you can see, the only significant effect is the Outcome x Esteem (O x E) interaction. The form of the interaction can be seen in Figure 5.
- Clearly the effect of "Outcome" is different for the two levels of Esteem: For subjects high in self esteem, failure led to less attribution to oneself then did success.
- By contrast, for subjects low in self esteem, failure led to more attribution to oneself than did success. Notice that the two lines in the graph are not parallel.
Nonparallel lines indicate interaction.
- The significance test for the interaction determines whether it is justified to conclude that the lines in the population are not parallel.
- Lines do not have to cross for there to be interaction.
Three-Factor Designs
- Three-factor designs are analyzed in much the same way as two-factor designs
- The table below shows the analysis of a study described by Franklin and Cooley investigating three factors on the strength of industrial fans:
- Hole Shape (Hex or Round)
- Assembly Type (Stake or Spun)
- Barrel Shape (knurled or smooth)
- The dependent variable, breaking torque, was measured in foot-pounds
- There were eight observations in each of the eight combinations of the three factors.
As you can see in table below there are:
- three main effect
- three two-way interactions
- one three-way interaction
- The degrees of freedom for the main effects are, as in a two-factor design, equal to the number of levels of the factor minus one
- Since all the factors here have two levels, all the main effects have one degree of freedom
- The interaction degrees of freedom are always equal to the product of the degrees of freedom of the component parts
- This holds for the three-factor interaction as well as the two-factor interactions
- The error degrees of freedom is equal to the number of observations (64) minus the number of groups (8) and is 56.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Hole | 1 | 8258.27 | 8258.27 | 266.68 | <0.0001 |
Assembly | 1 | 13369.14 | 13369.14 | 431.73 | <0.0001 |
H x A | 1 | 2848.89 | 2848.89 | 92.00 | <0.0001 |
Barrel | 1 | 35.0417 | 35.0417 | 21.5641 | <0.0001 |
H x B | 1 | 594.14 | 594.14 | 19.1865 | <0.0001 |
A x B | 1 | 135.14 | 135.14 | 4.36 | 0.0413 |
H x A x B | 1 | 1396.89 | 1396.89 | 45.11 | <0.0001 |
Error | 56 | 1734.12 | 30.97 | ||
Total | 63 | 221386.91 |
- A three-way interaction means that the two-way interactions differ as a function of the level of the third variable
- The usual way to portray a three-way interaction is to plot the two-way interactions separately
- The Figure below shows the Barrel (knurled or smooth) x Assembly (Staked or Spun) separately for the two levels of Hole Shape (Hex or Round)
- For the Hex Shape, there is very little interaction with the lines being close to parallel with a very slight tendency for the effect of Barrel to be bigger for Staked than for Spun
- The two-way interaction for the Round Shape is different: The effect of Barrel is bigger for Spun than for Staked
- The finding of a significant three-way interaction indicates that this difference in two-way interactions is not due to chance.
Questions
Data used to most of the questions:
Age | Recall | Score |
---|---|---|
1 | 1 | 9.0 |
1 | 2 | 7.0 |
1 | 3 | 11.0 |
1 | 1 | 8.0 |
1 | 2 | 9.0 |
1 | 3 | 13.0 |
1 | 1 | 6.0 |
1 | 2 | 6.0 |
1 | 3 | 8.0 |
1 | 1 | 8.0 |
1 | 2 | 6.0 |
1 | 3 | 6.0 |
1 | 1 | 10.0 |
1 | 2 | 6.0 |
1 | 3 | 14.0 |
1 | 1 | 4.0 |
1 | 2 | 11.0 |
1 | 3 | 11.0 |
1 | 1 | 6.0 |
1 | 2 | 6.0 |
1 | 3 | 13.0 |
1 | 1 | 5.0 |
1 | 2 | 3.0 |
1 | 3 | 13.0 |
1 | 1 | 7.0 |
1 | 2 | 8.0 |
1 | 3 | 10.0 |
1 | 1 | 7.0 |
1 | 2 | 7.0 |
1 | 3 | 11.0 |
2 | 1 | 8.0 |
2 | 2 | 10.0 |
2 | 3 | 14.0 |
2 | 1 | 6.0 |
2 | 2 | 7.0 |
2 | 3 | 11.0 |
2 | 1 | 4.0 |
2 | 2 | 8.0 |
2 | 3 | 18.0 |
2 | 1 | 6.0 |
2 | 2 | 10.0 |
2 | 3 | 14.0 |
2 | 1 | 7.0 |
2 | 2 | 4.0 |
2 | 3 | 13.0 |
2 | 1 | 6.0 |
2 | 2 | 7.0 |
2 | 3 | 22.0 |
2 | 1 | 5.0 |
2 | 2 | 10.0 |
2 | 3 | 17.0 |
2 | 1 | 7.0 |
2 | 2 | 6.0 |
2 | 3 | 16.0 |
2 | 1 | 9.0 |
2 | 2 | 7.0 |
2 | 3 | 12.0 |
2 | 1 | 7.0 |
2 | 2 | 7.0 |
2 | 3 | 11.0 |
Unequal Sample Sizes
Learning Objectives
- State why unequal n can be a problem
- Define confounding
- Compute weighted and unweighted means
- Distinguish between Type I and Type III sums of squares
- Describe why the cause of the unequal sample sizes makes a difference for the interpretation
The Problem of Confounding
- Whether by design, accident, or necessity, the number of subjects in each of the conditions in an experiment may not be equal.
- For example, the sample sizes for the Obesity and Relationships case study are shown in table below
- Although the sample sizes were approximately equal, the most subjects were in the Acquaintance/Typical condition
- Since n is used to refer to the sample size of an individual group, designs with unequal samples sizes are sometimes referred to as designs with unequal n.
Companion Weight | |||
---|---|---|---|
Obese | Typical | ||
Relationship | Girl Friend | 40 | 42 |
Acquaintance | 40 | 54 |
An absurd design
- We consider an absurd design to illustrate the main problem caused by unequal n.
- Suppose an experimenter were interested in the effect of diet and exercise on cholesterol
Exercise | |||
---|---|---|---|
Moderate | None | ||
Diet | Low Fat | 5 | 0 |
High Fat | 0 | 5 |
What makes this example absurd, is that there are no subjects in either the Low Fat/No exercise condition or the High Fat/Moderate exercise condition
Exercise | ||||
---|---|---|---|---|
Moderate | None | Mean | ||
Diet | Low Fat | -20 | 0 | -25 |
-25 | ||||
-30 | ||||
-35 | ||||
-15 | ||||
High Fat | 0 | -20 | -5 | |
6 | ||||
-10 | ||||
-6 | ||||
5 | ||||
Mean | -25 | -5 | -15 |
- The last column shows the mean change in cholesterol for the two Diet conditions whereas the last row shows the mean change for the two Exercise conditions.
- The value of -15 in the lower-right most cell in the table is the mean of all subjects.
- We see from the last column that those on the low-fat diet lowered their cholesterol an average of 25 units whereas those on the high-fat diet lowered theirs by only 5 units
- However, there is no way of knowing whether the difference is due to diet or to exercise since every subject in the low-fat condition was in the moderate-exercise condition and every subject in the high-fat condition was in the no-exercise condition
- Therefore, Diet and Exercise are completely confounded.
The problem with unequal n is that it causes confounding.
Weighted and Unweighted Means
The difference between weighted and unweighted means is critical for understanding how to deal with the confounding resulting from unequal n.
- Weighted and unweighted means will be explained using the data shown in Table 4
- Here, Diet and Exercise are confounded because 80% of the subjects in the low-fat condition exercised as compared to 20% of those in the high-fat condition
- However, there is not complete confounding as there was with the data in the Absurd Example
- The weighted mean for "low fat" is computed as the mean of the low-fat moderate-exercise condition and the low-fat no-exercise mean, weighted in accordance with sample size
- To compute a weighted mean, you multiply each mean by its sample size and divide by N, the total number of observations
- Since there are four subjects in the low-fat moderate-exercise condition and one subject in the low-fat no-exercise condition, the means are weighted by factors of 4 and 1 as shown below where Mw is the weighted mean.
- The weighted mean for the low-fat condition is also the mean of all five scores in this condition.
- Thus if you ignore the factor "Exercise," you are implicitly computing weighted means.
The unweighted mean for the low-fat condition (Mu) is simply the mean of the two means.
Exercise | |||||
---|---|---|---|---|---|
Moderate | None | Weighted Mean | Unweighted Mean | ||
Diet | Low Fat | -20 | -20 | -26 | -23.75 |
-25 | |||||
-30 | |||||
-35 | |||||
M=-27.5 | M=-20.0 | ||||
High Fat | -15 | 6 | -4 | -8.125 | |
-6 | |||||
5 | |||||
-10 | |||||
M=-15.0 | M=-1.25 | ||||
Weighted Mean | -25 | -5 | |||
Unweighted Mean | -21.25 | -10.625 |
- One way to evaluate the the main effect of Diet is to compare the weighted mean for the low-fat diet (-26) with the weighted mean for the high-fat diet (-4)
- This difference of -22 is called "the effect of diet ignoring exercise" and is misleading since most of the low-fat subjects exercised and most of the high-fat subjects did not
- However, the difference between the unweighted means of -15.5 (-23.75 minus -8.25) is not affected by this confounding and is therefore a better measure of the main effect
In short:
the weighted means ignore the effects of other variables (exercise in this example) and result in confounding; unweighted means control for the effect of other variables and therefore eliminate the confounding.
Statistical analysis programs use different terms for means that are computed controlling for other effects. SPSS calls them "estimated marginal means" whereas SAS and SAS JMP call them least squares means.
Types of Sums of Squares
- The section on Multi-factor ANOVA stated that the sum of squares total was not equal to the sum of the sum of squares for all the other sources of variation when there is unequal n.
- This is because the confounded sums of squares are not apportioned to any source of variation
- For the data in Table 4:
- the sum of squares for Diet is 390.625
- the sum of squares for Exercise is 180.625
- the sum of squares confounded between these two factors is 819.375
Type III Sums of Squares
- In the ANOVA Summary Table shown in Table 5, this large portion of the sum of squares is not apportioned to any source of variation and represents the "missing" sums of squares
- That is, if you add up the sums of squares for Diet, Exercise, D x E, and Error, you get 904.625
- If you add the confounded sum of squares of 819.375 to this value you get the total sum of squares of 1722.00
- When confounded sums of squares are not apportioned to any source of variation, the sums of squares are called Type III sums of squares
- Type III sums of squares are, by far, the most common and if sums of squares are not otherwise labelled, it can safely be assumed that they are Type III
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Diet | 1 | 390.625 | 390.625 | 7.43 | 0.034 |
Exercise | 1 | 180.625 | 180.625 | 3.423 | 0.113 |
D x E | 1 | 15.625 | 15.625 | 0.2969 | 0.605 |
Error | 6 | 315.750 | 52.625 | ||
Total | 9 | 1722.000 |
Type I Sums of Squares
- When confounded sums of squares are apportioned to sources of variation, the sums of squares are called Type I sums of squares
- The order in which the confounded sums of squares is apportioned is determined by the order in which the effects are listed
- The first effect gets any sums of squares confounded between it and any of the other effects
- The second gets the sums of squares confounded between it and subsequent effects, but not confounded with the first effect, etc
- The Type I sums of squares are shown in Table 6
- As you can see, with Type I sums of squares, the sum of all sums of squares is the total sum of squares.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Diet | 1 | 1210.000 | 390.625 | 7.43 | 0.034 |
Exercise | 1 | 180.625 | 180.625 | 3.423 | 0.113 |
D x E | 1 | 15.625 | 15.625 | 0.2969 | 0.605 |
Error | 6 | 315.750 | 52.625 | ||
Total | 9 | 1722.000 |
Type II Sums of Squares
- In Type II sums of squares, sums of squares confounded between main effects is not apportioned to any source of variation whereas sums of squares confounded between main effects and interactions is apportioned to the main effects
- In our example, there is no confounding between D x E interaction and either of the main effects
- Therefore, the Type II sums of squares are equal to the Type III sums of squares.
Unweighted Mean Analysis
- Type III sums of squares are tests of difference in unweighted means
- However, there is an alternative method to testing the same hypotheses tested using Type III sums of squares
- This method, unweighted means analysis, is computationally simpler than the standard method but is an approximate test rather than an exact test
- It is, however, a very good approximation in all but extreme cases
- Moreover, it is exactly the same as the traditional test for effects with one degree of freedom
Causes of Unequal Samples
- None of the methods for dealing with unequal sample sizes are valid if the experimental treatment is the source of the unequal sample sizes
- Imagine an experiment seeking to determine whether publicly performing an embarrassing act would affect one's anxiety about public speaking
- In this imaginary experiment, the experimental group is asked to reveal to a group of people the most embarrassing thing they have ever done
- The control group is asked to describe what they had at their last meal
- Twenty subjects are recruited for the experiment and randomly divided into two equal groups of 10, one for the experimental treatment and one for the control
- Following the description, subjects are given an attitude survey concerning public speaking
- This seems like a valid experimental design
- However, of the 10 subjects in the experimental group, four withdrew from the experiment because they did not wish to publicly describe an embarrassing situation
- None of the subjects in the control group withdrew
- Even if the data analysis shows a significant effect, it would not be valid to conclude that the treatment had an effect because a likely alternative explanation cannot be ruled out
- Namely, subjects who were willing to describe an embarrassing situation differed from those who were not even before the experiment began
- Thus, the differential drop-out rate destroyed the random assignment of subjects to conditions, a critical feature of the experimental design
- No amount of statistical adjustment can compensate for this flaw.
Questions
Tests Supplementing ANOVA
Learning Objectives
- Compute Tukey HSD test
- Describe an interaction in words
- Describe why one might want to compute simple effect tests following a significant interaction
Tests Supplementing ANOVA
- The null hypothesis tested in a one-factor ANOVA is that all the population means are equal
- Stated more formally,
H0: μ1 = μ2 = ... μk k is the number of conditions
- When the null hypothesis is rejected, then all that can be said is that at least one population mean is different from at least one other population mean
- The methods described in the sections on All Pairwise Comparisons and on Specific Comparisons for doing more specific tests apply here
- Keep on mind that these tests are valid whether or not they are preceded by an ANOVA.
Main Effects
- As shown below, significant main effects in multi-factor designs can be followed up in the same way as significant effects in a one-way designs
- Table below shows the data from an imaginary experiment with three levels of Factor A and two levels of Factor B.
A1 | A2 | A3 | Marginal Means | |
---|---|---|---|---|
B1 | 5 | 9 | 7 | 7.08 |
4 | 8 | 9 | ||
6 | 7 | 9 | ||
5 | 8 | 8 | ||
Mean = 5 | Mean = 8 | Mean = 8.25 | ||
B2 | 4 | 8 | 8 | 6.50 |
3 | 6 | 9 | ||
6 | 8 | 7 | ||
8 | 5 | 6 | ||
Mean = 5.25 | Mean = 6.75 | Mean = 7.50 | ||
Marginal Means | 5.125 | 7.375 | 7.875 | 6.79 |
- Table below shows the ANOVA Summary Table for these data
- The significant main effect of A indicates that, in the population, at least one of the marginal means for A is different from at least one of the others.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
A | 2 | 34.333 | 17.17 | 9.29 | 0.0017 |
B | 1 | 2.042 | 2.04 | 1.10 | 0.3070 |
AB | 2 | 2.333 | 1.167 | 0.63 | 0.5431 |
Error | 18 | 33.250 | 1.847 | ||
Total | 23 | 71.958 | 3.129 |
Tukey HSD test
The Tukey HSD can be used to test all pairwise comparisons among means in a one-factor ANOVA as well as comparisons among marginal means in a multi-factor ANOVA.
The formula for the equal-sample-size case is shown below.
Mi and Mj are marginal means MSE is the mean square error from the ANOVA n is the number of scores each mean is based upon
- For this example, MSE = 1.847 and n= 8 because there are eight scores at each level of A
- The probability value can be computed using the Studentized Range Calculator
- The degrees of freedom is equal to the degrees of freedom error
- For this example, df = 18
- The results of the Tukey HSD test are shown in Table 3
- The mean for A1 is significantly lower than the mean for A2 and the mean for A3
- The means for A2 and A3 are not significantly different.
Comparison | Mi - Mj | Q | p |
---|---|---|---|
A1 - A2 | -2.25 | -4.68 | 0.0103 |
A1 - A3 | -2.75 | -5.73 | 0.0021 |
A2 - A3 | -0.50 | -1.04 | 0.7456 |
Specific comparisons among means are also carried out much the same way as shown in the relevant section on testing means
The formula for L is
ci is the coefficient for the ith marginal mean Mi is the ith marginal mean
For example, to compare A1 with the average of A2 and A3, the coefficients would be 1, -0.5, -0.5
L = (1)(5.125) + (-0.5)(7.375) + (-0.5)(7.875) = -2.5
To compute t, use:
= -4.25 MSE is the mean square error from the ANOVA n is the number of scores each marginal mean is based on (eight in this example)
- The degrees of freedom is the degrees of freedom error from the ANOVA and is equal to 18
- Using the Online Calculator, we find that the two-tailed probability value is 0.0005
- Therefore, the difference between A1 and the average of A2 and A3 is significant.
Important issues concerning multiple comparisons and orthogonal comparisons are discussed in the Specific Comparisons section in the Testing Means chapter.
Interactions
- The presence of a significant interaction makes the interpretation of the results more complicated.
- Since an interaction means that the simple effects are different, the main effect as the mean of the simple effects does not tell the whole story
- This section discusses how to describe interactions, proper and improper uses of simple effects tests, and how to test components of interactions
Describing Interactions
- A crucial step first step in understanding a significant interaction is constructing an interaction plot
- Figure 1 shows an interaction plot from data presented in the section on Multi-factor ANOVA.
- The second step is to describe the interaction in a clear and understandable way
- This is often done by describing how by describing how the simple effects differed
- Since this should be done using as little jargon as possible, the word "simple effect" need not appear in the description
- An example is as follows:
The effect of Outcome differed depending on the subject's self esteem. The difference between the attributions to self following success and attributions to self following failure was larger for high-self-esteem subjects (mean difference = 2.50) than for low-self-esteem subjects (mean difference = -2.33).
- No further analyses are helpful in understanding the interaction since the interaction means only that the simple effects differ
- The interaction's significance indicates that the simple effects differ from each other, but provides no information about whether they differ from zero.
- If neither simple effect is significant, the conclusion should be that the simple effects differ, and that at least one of them is not zero.
- However, no conclusion should be drawn about which simple effect(s) is/are not zero.
- Another error that can be made by mistakenly accepting the null hypothesis is to conclude that two simple effects are different because one is significant and the other is not
- Consider the results of an imaginary experiment in which the researcher hypothesized that addicted people would show a larger increase in brain activity following some treatment than would non-addicted people
- In other words, the researcher hypothesized that addiction status and treatment would interact
- The results shown in Figure 2 are very much in line with the hypothesis
- However, the test of the interaction resulted in a probability value of 0.08, a value not quite low enough to be significant at the conventional 0.05 level
- The proper conclusion is that the experiment supports the researcher's hypothesis, but not strongly enough to allow a firm conclusion.
Simple Effect Tests
- It is not necessary to know whether the simple effects differ from zero in order to understand an interaction because the question of whether simple effects differ from zero has nothing to do with interaction except that if they are both zero there is no interaction
- It is not uncommon to see research articles in which the authors report that they analyzed simple effects in order to explain the interaction
- However, this is not a correct since an interaction does not depend on the analysis of the simple effects.
- However, there is a reason to test simple effects following a significant interaction
- Since an interaction indicates that simple effects differ, it means that the main effects are not general
- In the made-up example, the main effect of Outcome is not very informative, and the effect of outcome should be considered separately for high- and low-self-esteem subjects.
- As will be seen, the simple effects of Outcome are significant and in opposite directions: Success significantly increases attribution to self for high-self-esteem subjects and significantly lowers attribution to self for low-self-esteem subjects
- This is a very easy result to interpret.
- What would the interpretation have been if neither simple effect had been significant?
- On the surface, this seems impossible: How can the simple effects both be zero if they differ from each other significantly as tested by the interaction?
- The answer is that a non-significant simple effect does not mean that the simple effect is zero: the null hypothesis should not be accepted just because it is not rejected
- If neither simple effect is significant, the conclusion should be that the simple effects differ, and that at least one of them is not zero.
- However, no conclusion should be drawn about which simple effect(s) is/are not zero.
Another error that can be made by mistakenly accepting the null hypothesis is to conclude that two simple effects are different because one is significant and the other is not. Consider the results of an imaginary experiment in which the researcher hypothesized that addicted people would show a larger increase in brain activity following some treatment than would non-addicted people. In other words, the researcher hypothesized that addiction status and treatment would interact. The results shown in Figure 2 are very much in line with the hypothesis. However, the test of the interaction resulted in a probability value of 0.08, a value not quite low enough to be significant at the conventional 0.05 level. The proper conclusion is that the experiment supports the researcher's hypothesis, but not strongly enough to allow a firm conclusion.
Unfortunately, the researcher was not satisfied with such a weak conclusion and went on to test the simple effects. It turned out that the effect of Treatment was significant for the Addicted group (p = 0.02) but not significant for the Non-Addicted group (p = 0.09). The researcher then went on to conclude that since there is an effect of Treatment for the Addicted group but not for the Non-Addicted group, the hypothesis of a greater effect for the former than for the latter group is demonstrated. This is faulty logic, however, since it is based on accepting the null hypothesis that the simple effect of Treatment is zero for the Non-Addicted group just because it is not significant.
Components of Interaction (optional)
Figure 3 shows the results of an imaginary diet on weight loss. A control group and two diets were used for both overweight teens and overweight adults.
The difference between Diet A and the Control diet was essentially the same for teens and adults whereas the difference between Diet B and Diet A was much larger for the Teens than it was for the Adults. Over one portion of the graph the lines are parallel whereas over another portion they are not. It is possible to test these portions or components of interactions using the method of specific comparisons discussed previously. The test of the difference between Teens and Adults on the difference between Diets A and B could be tested with the coefficients shown in Table 4. Naturally, the same consideration regarding multiple comparisons and orthogonal comparisons apply to comparisons involving components of interaction that apply to other comparisons among means.
Age Group | Diet | Coefficient |
---|---|---|
Teen | A | 1 |
Teen | B | -1 |
Adult | A | -1 |
Adult | B | 1 |
Questions
Within-Subjects ANOVA
Learning Objectives
- Define a within-subjects factor
- Explain why a within-subjects design can be expected to have more power than a between-subjects design
- Be able to create the Source and df columns of an ANOVA summary table for a one-way within-subjects design
- Explain error in terms of interaction
- Discuss the problem of carry-over effects
- Be able to create the Source and df columns of an ANOVA summary table for a design with one between-subjects and one within-subjects variable
- Define sphericity
- Describe the consequences of violating the assumption of sphericity
- Discuss courses of action that can be taken if sphericity is violated
Within-Subjects
- Within-subjects factors involve comparisons of the same subjects under different conditions
- For example, in the ADHD Treatment Study, each child's performance was measured four times, once after being on each of four drug doses for a week
- Therefore, each subject's performance was measured at each of the four levels of the factor "Dose."
- Note the difference from between-subjects factors for which each subject's performance is measured only once and the comparisons are among different groups of subjects
- A within-subjects factor is sometimes referred to as a repeated measures factor since repeated measurements are taken on each subject
- An experimental design in which the independent variable is a within-subjects factor is called a within-subjects design.
One-factor Designs
ADHD Example
- Let's consider how to analyze the data from the ADHD treatment case study
- These data consist of the scores of 24 children with ADHD on a delay of gratification (DOG) task
- Each child was tested under four dosage levels
- For now we will be concerned only with testing the difference between the mean in the placebo condition (the lowest dosage, D0) and the mean in the highest dosage condition (D60)
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Subjects | 23 | 5781.98 | 251.39 | ||
Dosage | 1 | 295.02 | 295.02 | 10.38 | 0.004 |
Error | 23 | 653.48 | 28.41 | ||
Total | 47 | 630.48 |
Results Interpretation
- The first source of variation, "Subjects," refers to the differences among subjects
- If all the subjects had exactly the same mean (across the two dosages) then the sum of squares for subjects would be zero; the more subjects differ from each other, the larger the sum of squares subjects.
- Dosage refers to the differences between the two dosage levels
- If the means for the two dosage levels were equal, the sum of squares would be zero
- The larger the difference between means, the larger the sum of squares.
- The error reflects the degree to which the effect of dosage is different for different subjects
- If subjects all responded very similarly to the drug, then the error would be very low
- For example, if all subjects performed moderately better with the high dose than they did with the placebo, then the error would be low
- On the other hand, if some subjects did better with the placebo while others did better with the high dose, then the error would be high
- It should make intuitive sense that the less consistent the effect of the drug, the larger the drug effect would have to be in order to be significant
- The degree to which the effect of the drug differs depending on the subject is the Subjects x Drug interaction
- Recall that an interaction occurs when the effect of one variable differs depending on the level of another variable
- In this case, the size of the error term is the extent to which the effect of the variable "Drug" differs depending on the level of the variable "Subjects."
- Note that each subject is a different level of the variable "Subjects."
- Other portions of the summary table have the same meaning as in between-subjects ANOVA. The F for dosage is the mean square for dosage divided by the mean square error
- For these data, the F is significant with p = 0.004. Notice that this F test is equivalent to the t-test for correlated pairs, with F = t2.
Design for four dosage level
- Table 2 shows the ANOVA Summary Table when all four doses are included in the analysis
- Since there are now four dosage levels rather than two, the df for dosage is three rather than one
- Since the error is the Subjects x Dosage interaction, the df for error is the df for "Subjects" (23) times the df for Dosage (3) and is equal to 69.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Subjects | 23 | 9065.49 | 394.15 | ||
Dosage | 3 | 557.61 | 185.87 | 5.18 | 0.003 |
Error | 69 | 2476.64 | 35.89 | ||
Total | 95 | 12099.74 |
Carry-over effects
- Often performing in one condition affects performance in a subsequent condition in such a way to make a within-subjects design impractical
- For example, consider an experiment with two conditions
- In both conditions subjects are presented with pairs of words
- In Condition A subjects are asked to judge whether the words have similar meaning whereas in Condition B subjects are asked to judge whether they sound similar
- In both conditions, subjects are given a surprise memory test at the end of the presentation
- If condition were a within-subjects variable, then the there would be no surprise after the second presentation and it is likely that the subjects would have been trying to memorize the words.
- Not all carry-over effects cause such serious problems
- For example, if subjects get fatigued by performing a task, then they would be expected to do worse on the second condition they were in
- However, as long as half the subjects are in Condition A first and Condition B second, the fatigue effect itself would not invalidate the results, although it would add noise and reduce power
- The carryover effect is symmetric in that having Condition A first affects performance in Condition B to the same degree that having Condition B first affects performance in Condition A.
- Asymmetric carryover effects cause more serious problems
- For example, suppose performance in Condition B were much better if preceded by Condition A whereas performance in Condition A was approximately the same regardless of whether it was preceded by Condition B
- With this kind of carryover effect it is probably better to use a between-subjects design.
One between and one-within-subjects factor
- In the Stroop Interference case study, subjects performed three tasks: naming colors, reading color words, and naming the ink color of color words
- Some of the subjects were males and some of the subjects were females
- Therefore this design had two factors: gender and task
- The ANOVA Summary Table for this design is shown in Table 3.
Source | df | SSQ | MS | F | p |
---|---|---|---|---|---|
Gender | 1 | 83.32 | 83.32 | 1.99 | 0.165 |
Error | 45 | 1880.56 | 41.79 | ||
Task | 2 | 9525.97 | 4762.99 | 228.06 | <0.001 |
Gender x Task | 2 | 55.85 | 27.92 | 1.34 | 0.268 |
Error | 90 | 1879.67 | 20.89 |
- First notice that there are two error terms: one for the between-subjects variable Gender and one for both the within-subjects variable Task and the interaction of the between-subjects variable and the within-subjects variable
- Typically, the mean square error for the between-subjects variable will be higher than the other mean square error
- In this example, the mean square error for Gender is about twice as large as the other mean square error.
- The degrees of freedom for the between-subjects variable is equal to the number of levels of the between subjects variable minus one
- In this example it is one since there are two levels of gender. Similarly, the degrees of freedom for the within-subjects variable is equal to the number of levels of the variable minus one
- In this example, it is two since there are three tasks
- The degrees of freedom for the interaction is the product of the degrees of freedom of the two variables.
- For the Gender x Task interaction, the degrees of freedom is the product of degrees of freedom Gender (which is 1) and the degrees of freedom Task (which is 2) and is equal to 2.
Assumption of Sphericity
- Within-subjects ANOVA makes a restrictive assumption about the variances and the correlations among the dependent variables
- Although the details of the assumption are beyond the scope of this book, it is approximately correct to say that it is assumed that all the correlations are equal and all the variances are equal
- Table 4 shows the correlations among the three dependent variables in the Stroop Interference case study.
word reading | color naming | interference | |
---|---|---|---|
word reading | 1 | 0.7013 | 0.1583 |
color naming | 0.7013 | 1 | 0.2382 |
interference | 0.1583 | 0.2382 | 1 |
- Note that the correlation between the word reading and the color naming variables of 0.7013 is much higher than the correlation between either of these variables with the interference variable
- Moreover, as shown in Table 5, the variances among the variables differ greatly.
Variable | Variance |
---|---|
word reading | 15.77 |
color naming | 13.92 |
Interference | 55.07 |
- Naturally, the assumption of sphericity, like all assumptions, refers to populations not samples
- However it is clear from these sample data, the assumption is not met here in the population.
Consequences of Violating the Assumption of Sphericity
- Although ANOVA is robust to most violations of its assumptions, the assumption of sphericity is an exception: Violating the assumption of sphericity leads to a substantial increase in the Type I error rate.
- Moreover, this assumption is rarely met in practice
- Although violations of this assumption had at one time received little attention, the current consensus of data analysts is that it is no longer considered acceptable to ignore them.
Approaches to Dealing with Violations of Sphericity
- If an effect is highly significant, there is a conservative test that can be used to protect against an inflated Type I error rate
- This test consists of adjusting the degrees of freedom for all within subject variables as follows: The degrees of freedom numerator and denominator are divided by the number of scores per subject minus one.
- Consider the effect of Task shown in Table 3
- There are three scores per subject and therefore the degrees of freedom should be divided by two
- The adjusted degrees of freedom are:
(2)(1/2) = 1 for the numerator and (90)(1/2)= 45 for the denominator
- The probability value is obtained using the F probability calculator with the new degrees of freedom parameters
- The probability of an F of 228.06 or larger with 1 and 45 degrees of freedom is less than 0.001
- Therefore, there is no need to worry about the assumption violation in this case.
- Possible violation of sphericity does make a difference in the interpretation of the analysis shown in Table 2.
- The probability value of an F or 5.18 with 1 and 23 degrees of freedom is 0.032, a value that would lead to a more cautious conclusion than the p value of 0.003 shown in Table 2.
- The correction described above is very conservative and should only be used when, as in Table 3, the probability value is very low
- A better correction, but one that is very complicated to calculate is to multiply the degrees of freedom by a quantity called ε
- There are two methods of calculating ε
- The correction called the Huynh-Feldt (or H-F) is slightly preferred to the called the Geisser Greenhouse (or G-G) although both work well
- The G-G correction is generally considered a little too conservative
- A final method for dealing with violations of sphericity is to use a multivariate approach to within-subjects variables
- This method has much to recommend it, but it is beyond the score of this text.
Questions