ANOVA Revision

Have the assumptions What is being compared?

been met? Means Relationships between
Variables
How many samples are being compared? How many DVs?
1 2 Samples/Treatments/Measurements >2 Samples/1 sample 1 IV, 1DV 1 IV, >1 DV
Sample measured more than twice (2
variables)
What is the nature of the samples? How many IVs? Associat Prediction
ion
Related (1 sample, 2 Independent 1 IV >1 IV
measurements)/ Samples
Matched Samples (2
samples)
Yes: σ known: Two related samples t- 2 Independent Independent Two- Pearson Regression
Parametric z- test sample t-test measures: 1- Factor correlation
(Interval) test way ANOVA ANOVA
σ
unknown: Repeated
t- test Measures:
repeated-
measures
ANOVA
No: Ordinal: Ordinal: Wilcoxon signed Ordinal: Mann Ordinal for Ordinal: Ordin Nominal:
Non-Parametric Wilcoxo ranks test Whitney U independent Friedman al: CHAID (Chi
n measures: (for main Spearm Square
signed Nominal: Chi- Kruskal- effects only) an Automati
ranks squared test of Wallis c
test independence Nominal: Interactio
(2 samples) Nominal: Chi- n
Nominal: Chi- squared squared Detection
Chi- test of test of )
squared independenc independ
One way ANOVA
• What is a one-way independent measures ANOVA used for?

• To test for statistically significant differences between three or more independent sample means.

• What questions can you answer with a one way ANOVA?

1. Are there differences in reaction times based on the drink (orange juice, coffee and
water) someone consumes?
• In this example, we’re asking whether the three independent (separate) groups produce different reaction times.
• Each participant is a member of just one group (participants have only one of the drinks, they cannot have more than
one at different times).

• 2. Does the amount of television watched by South African children vary from province to
province (Gauteng, Western Cape, Eastern Cape and Kwazulu Natal)?
One way ANOVA
formulaes
Factorial ANOVA
• What is the purpose of factorial ANOVA?
• To test hypotheses about mean differences when there are two or more between-
subjects independent variables.
• Factorial designs provide:
• A main effect for each IV in isolation.
• An interaction effect, which examines the influence of the IVs in combination

• What are examples?

• Do the class evaluation scores of male and female students differ if the
lecturer is male versus female?
• Do the effects of different hormones (A, B, and C) on rats’ maze running
speed depend on whether they were raised in an enriched or impoverished
environment?
Repeated Measures ANOVA
What is the purpose of repeated measures ANOVA?
• To test for statistically significant differences between three or more related
sample means.
• They are comprised of the same individuals who provided data on multiple
occasions
• If each individuals is connected/matched with another individual in the sample,
this is known as a matched design.
• What are examples?
1. Does counselling alter teacher’s attitudes towards their jobs over
time? (pre-counselling, two month follow-up, one year follow-up?)
2. Do students prefer Mentos, chocolate or Smarties?
Repeated Measures
ANOVA formulaes
Question 1
• An analysis of variance is used to evaluate the mean differences for a
research study comparing four treatments with a separate sample of
n = 8 in each treatment. If the data produce an F-ratio of F = 5.50,
then which of the following is the correct statistical decision?
a. Reject the null hypothesis with α = .05 but not with α = .01.
b. Reject the null hypothesis with either α = .05 or α = .01.
c. Fail to reject the null hypothesis with either α = .05 or α = .01.
d. There is not enough information to make a statistical decision.
Answer
• An analysis of variance is used to evaluate the mean differences for a
research study comparing four treatments with a separate sample of
n = 8 in each treatment. If the data produce an F-ratio of F = 5.50,
then which of the following is the correct statistical decision?
b. Reject the null hypothesis with either α = .05 or α = .01.
Question 2 Step 1: state the
hypothesis
Step 2: set criteria for
decision
Step 3: calculate
Step 4: make a
decision
• Pg406 of gravetter
Question 3 Use data set:
data_2_1

• A researcher wants to investigate the impact of education support services on

the attitudes of teachers towards the mainstreaming (i.e. inclusion) of children
with disabilities into their classrooms. 30 teachers were selected to take part in
the study and divided into three groups of 10 participants each:
• •No additional support
• •In class teacher aide
• •Weekly support group.
• At the end of the term, all the teachers were asked to complete the attitude
towards mainstreaming scale. This scale consists of 18 items and a high score
reflects more positive attitudes towards mainstreaming.
Were the results significant?
• CPA 1

• It was significant
Question 4 Step 1: state the
hypothesis
Step 2: set criteria for
decision
Step 3: calculate
Step 4: make a
decision
• The critical region value is 3.88

• Therefore there is no difference and we fail to reject the null

hypothesis.
Question 5
 Pg 476

• Make a decision and state a conclusion

• All three F-ratios have df = 1, 76. With α = .05, the critical F value is 3.98 for all three tests.
• For these data, factor A (weight) has no significant effect; F(1, 76) = 0.28. Statistically, there is no
difference in the number of crackers eaten by normal vs. obese participants.
• Similarly, factor B (fullness) has no significant effect; F(1, 76) = 2.54. Statistically, the number of
crackers eaten by full participants is no different from the number eaten by hungry participants.
(Note: This conclusion concerns the combined group of normal and obese par- ticipants. The
interaction concerns these two groups separately.)
• These data produce a significant interaction; F(1, 76) = 4.51, p < .05. This means that the effect of
fullness does depend on weight. A closer look at the original data shows that the degree of
fullness did affect the normal participants, but it had no effect on the obese participants.