Psy2005: Applied Research Methods & Ethics in Psychology

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 30

Tutor Led

Psy2005: Applied Research Methods &


Ethics in Psychology
Lab Week 7: Using a factorial ANOVA (between-
groups) to investigate the effects of type of
session and type of therapy on self-reported
drug use

1
Tutor Led

SPSS/PASW data files


Open Psy2005 folder in My Learning
Open Week 7
Double click on ‘drug treatments4.sav’
Fundamental principle
Each participant has their own row
Each different bit of data must go in a separate column /
variable
Data view vs. Variable View
Change via ‘tabs’ at bottom of window
 or keyboard combination ⌘T
Data view for viewing / editing data
Variable view for details of variables
Tutor Led

Aims & Outcomes


Provide an overview of research focusing on drug
treatments
Conduct an independent groups factorial ANOVA
Explain the key features of factorial ANOVA
Complete Workbook 1 (week 7)
Assessment deadline for Lab report 1 (Dec. 16th)

3
Tutor Led

Factorial ANOVA
What are the advantages of factorial ANOVA?
We can look at the effects of more than one independent
variable
We can look at how these variables interact
What is a Main effect?
“The effects of one independent variable (factor) summed
(averaged) over all levels of the other independent variable.”
What is an interaction effect?
“The combined effect of two or more predictor variables on a
outcome variable”
“When the effect of one factor is not constant across all levels
of the other factor/s”

4
Tutor Led

The Beer Goggles Theory!


We wanted to know whether it was true that people found
their preferred sex more attractive when they were drunk?
Independent variable 1: Sex (male/female)
Independent variable 2: Drink (sober/drunk)
Dependent variable average rating of attractiveness of
pictures of people
Main effect hypotheses
H1: That there is a significant difference in self-reported
attractiveness ratings from sex
H2: That there is a significant difference in self-reported
attractiveness ratings from drinking

5
Tutor Led

Beer Goggles Theory!


What if we believed that there was more to this than
the effect of one independent variable on the
dependent variable?
What if we believed that two independent variables
worked together to change the dependent variable?
We could for example believe that this theory was only
true for males or indeed only true for females
H0: That there is no significant interaction between sex
and drinking on self-reported attractiveness
H1: That there is a significant interaction between sex and
drinking on self-reported attractiveness

6
Factorial ANOVA

Total Variability

Variance
Unexplained
Explained by the
Variance
Experiment

Variance
Explained by
Variance Variance
interaction
explained by Sex explained by drink
between sex and
drink

7
Tutor Led

Result: No effects of sex or drinking


69

59

49

39
Drunk
29 Sober

19

-1
Male Female

This finding shows that the ratings of attractiveness were the same at all
Levels of both the independent variables. Males and females do not differ and
Drunk and sober show no differences
8
Main Effect for Drinking but no effects for Sex
and no interaction effects Tutor Led
70

65

60

55
Drunk
50 Sober

45

40
Male Female

This result shows that both males and females found the pictures more attractive
When they were sober. In this case we have no interaction between the variables
because both sexes responded in the same fashion. We do, however, have a significant
9 Main effect of drinking with attractiveness being rated higher when sober.
Main effect for Sex, No effect for Drinking &Led
Tutor

No interaction effect
70

65

60

55 Drunk
Sober
50

45

40
Male Female

This result shows that males found the pictures more attractive regardless of
drinking. In this case we have no interaction between the variables because
10 males and females responded in the same way when drunk or sober
Cross-Over Effect. Interaction effect but no
main effects for sex or drinking Tutor Led

65

60

55
Drunk
50 Sober

45

40
Male Female
This is a classic cross-over interaction effect whereby the opposite findings are
observed at different levels of each independent variable. Here the females found
the pictures more attractive when they were drunk and the males found the
11 pictures more attractive when they were sober.
Interaction Effect with two main effects for
drinking and sex Tutor Led
65

60

55
Drunk
50
Sober
45

40
Male Female

These results show both main and interaction effects. However, both main
effects can be largely explained by the interaction between the variables. In
summary the results show that whilst drinking has no effect on the ratings of the
Males. Being drunk leads to improved ratings in the females

12
Tutor Led
Two main effects but no interaction effect
75
70
65
60
55 Drunk
Sober
50
45
40
35
Male Female
These results show two main effects but no interaction effect. Firstly, there is a main
Effect of sex because males are rating the pictures as more attractive than the females
(an average of 10%}. Secondly, there is also an effect of drink because both sexes rate
the pictures as more attractive when they are drunk (an average of 20%).
13
Two main effects and an interaction effect
Tutor Led

90
80
70
60
50 Drunk
Sober
40
30
20
10
Male Female
This is a complicated one! There is a main effect for sex because the females are
Rating the pictures as more attractive. There is also a main effect for drinking
Because the the pictures are being rated as more attractive when the people are drunk.
Finally there is an interaction effect because the females amplify their attractiveness
Ratings when they are drunk.
14
Tutor Led

Drug Treatments in Current Study


Treatments (Weekly Sessions)
12-Step programme
A Cognitive–Behavioural Motivational Intervention
Treatment as Usual
Type of session
Group therapy
Individual therapy
Outcome Measures (1, 6, & 12 months)
Self-monitored logbooks: drugs taken
Recordings taken over a 28 day period prior to
measurement

15
Tutor Led

Participants & Therapists


Participants:
Prolific and other Priority Offender status and tested
positive for cocaine or heroin during their arrest.
Randomly allocated to one of the three groups.
Therapists: Twelve therapists ran the sessions.
All were qualified to degree standard and had a
minimum of three years experience

16
Tutor Led

Procedure
Participants took part as part of a voluntary
rehabilitation procedure
Participants had either:
90 minute weekly closed (nobody was allowed to join
after the first session) meetings in groups of 4-8 people
with two Counsellors.
45 minute weekly individual sessions with one
Counsellor
Participants attended the sessions for 1 year.
Treatment outcomes were measured at 1 month, 6
months & 12 months.
17
Tutor Led

About the Experiment!


Teasing apart the design
Independent variables (factors):
Type of drug treatment : 12 Step, CBT/MI,
Treatment as Usual
Type of session: group, individual
Dependent variable
Self-reported drug use

18
Self-reported drug use Tutor Led
Type of session:
at time: 6 months Group Vs Individual therapy
Situation. More on this later!

Type of therapy:
Self-reported drug
12 Step
use at time: 1 month
CBT with MI
Self-reported drug Treatment as Usual.

19
use at time: 1 Year
Student
Student Led
Led
Conducting a factorial ANOVA

20
Student
Student Led
Led

21
22
Tutor Led

The Output: Factors

Provides a simple breakdown of the factors


A non-significant Levene Test shows that group
variances are equal (homogeneity of variance)

23
Tutor Led
The Output: The ANOVA

Session (η2=.151, F(1,136)=24.24, p<0.001)


Therapy (η2=.252, F(2,136)=22.94, p<0.001)
Session x Therapy (η2=.113, F(2,136)=8.64 p<0.001)

24
The Output: Main Effect for Session Factor
Session (η2=.151, F(1,136)=24.24, p<0.001)

We can tentatively conclude that self-reported drug-


use was significantly lower in the group session
Remember! There is a significant interaction effect

Tutor Led
25
The Output: Main Effect for Treatment Factor
(η2=.252, F(2,136)=22.94, p<0.001)

We know from previous analyses that CBT/MI is


significantly more effective than the other two
measures.
Remember! There is a significant interaction effect

Tutor Led
26
The Output: The interaction effect
(η2=.113, 15.876, F(2,136)=8.64 p<0.001)

Tutor Led
27
Conclusions: This is your first Summative
Assessment
Following a Factorial ANOVA on Drug Use (1 month)
the following findings were observed
Main effect for Session Factor
Main effect for Treatment Factor
Interaction effect for Session x Therapy
What does this mean?
You will have to wait until Week 10 to get a better
understanding of the interaction.
Before then, look at the graph and try to make sense of
it yourself

Tutor Led
28
Student
Student Led
Led
NOW IT IS YOUR TURN!
Quit the current data set and open
drugtreatments2019final.sav
Use the slides for the two-way ANOVA to carry out an
identical analysis for drug-use after one year of
treatment (drug3).
This can be found in:
Drugtreatments2019final
Paste the results in to the Workbook
Please note! These are the results you will be writing
up for you first laboratory report

29
Student
Student Led
Led

Complete Work & Save your files


Data Set: ‘drugtreatment2017final.sav’
Cut and paste the graphs in to Workbook 1 (week 7)
Next week: Simple main effects analyses to
breakdown the interaction and main effects

30

You might also like