Transcript Types of Inferential Statistics

Research Design & Analysis 1
Class 17
Announcements
Introduction to experimental designs
– examples from memory research
– Single factor multigroup designs
– Matched group designs
Review for midterm
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
1
Converting Question to an Analytical
Experiment
• operational definitions
• subject selection
• subject assignment
– within
– between
– matched
• variety of “mechanical details”
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
2
Experimental Hypothesis
• An emotional shock will disrupt memory for
events that occur immediately prior to
occurrence.
• Which words require operational definitions?
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
3
Experimental Hypothesis
• An emotional shock will disrupt memory
for events that occur immediately prior to
occurrence.
• Final, operationally valid version:
An unexpected 15-second scene that portrays
a mutilated body at the end of a 10 minute
travel film will disrupt a person's memory of
the price of 10 items listed in the film during
the one-minute period immediately
preceding the final scene.
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
4
Why Operational Definitions?
• avoid confusion
• makes study reproducible
• makes measurements reliable
But
trade off between precision of an
operational definition and it’s
construct validity
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
5
Converting Question to an Analytical
Experiment
• operational definitions
• subject selection
• subject assignment
• variety of “mechanical details”
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
6
Final Experiment
20 intro psychology subjects, 10 randomly
assigned to each group
•
•
•
•
•
•
•
Experimental group
Control group
Male experimenter
• Male experimenter
written instructions
• written instructions
10 min travel film
• 10 min travel film
1 o’clock testing
• 1 o’clock testing
group testing
• group testing
15 s final scene version ‘a’ • 15 s final scene version
- mutilated body seated in
‘b’ - craftsperson
chair (operational defn’ of
weaving a basket - I.V.
emotional shock) - I.V.
• Measure memory Measure memory - D.V.
D.V.
Statistically Compare differences in measured memory
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
7
Loftus and Burns 1982
Loftus, E., & Burns, T. Mental shock can
produce retrograde amnesia. Memory and
Cognition, 10, 318-323.
Model of human memory
Sensory Memory
Unattended
information
Attention
Working memory
Forgetting
Incoming
information
Retrieval
Rehearsal
Encoding
Long term memory
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
9
Free recall memory task
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
10
Model of human memory
Sensory Memory
Unattended
information
Attention
Working memory
Forgetting
Retrieval
Rehearsal
Encoding
Long term memory
7/21/2015
Incoming
information
Psyc2013 Class #17 (c) Peter
McLeod
False
Memories?
11
Multivalent Designs
Why use designs with more than two levels of
the independent variable?
• When you need multiple control conditions
– multiple control group design
• When you are interested in more than two
qualitatively different groups,
– Single factor non-parametric designs
• When you want to map out the function of
the effect of the IV.
– Single factor parametric designs
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
12
Mapping the Function of an I.V.’s
Effect
No bivalent study can show non-linear functions.
9
8
7
Single factor,
parametric
design
D.V.
6
5
4
3
2
1
0
0
1
2
3
4
5
6
7
8
9
10
Level of I.V. treatment
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
13
Stress and Cancer
93 lab rats randomly assigned from a defined population
Random
Random
Random
assignment 31 rats assignment 31 rats assignment 31 rats
Escapable shock
Inescapable
No shock
treatment
shock treatment
treatment
Ad lib food & water Ad lib food & water Ad lib food & water
14-10 LD cycle
14-10 LD cycle
14-10 LD cycle
Percent tumour
rejection 63%
Percent tumour
rejection 27%
Percent tumour
rejection 54%
Yoked control
Based on Visintainer, Volpicelli, & Seligman, 1982
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
14
Results
Multiple control
group design
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
15
Matched group designs
• Match participants on one or more
characteristics
• Randomly assign pairs to the groups
(one to each)
Advantages
– allows some control over subject variables
– can make experiment more sensitive
Disadvantages
– if matched variables not important, less
powerful statistics
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
16
Error variance
• Sources (extraneous variables)
Handling error variance
• Reducing error variance
• Increasing effectiveness of IV
• Randomizing error variance across
groups
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
17
Midterm
Everything since last midterm
• using non-experimental methods
• quantifying behaviour
• sequential analysis (discrete & continuous)
• reliability (% agreement & Cohen’s Kappa)
• sources of distortion
• descriptive statistics
• central tendency, spread, measurement scales
• frequency distributions & graphic presentation
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
18
Midterm
• survey research
• types of questions
• reliability & validity
• methods
• sampling methods
• correlations
• designs and statistics
• Pearson’s r, when to use, concerns
• Spearman’s rho and other correlations
• statistical inference - reading SPSS output
• scatterplots
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
19
Midterm
• Experimental designs
• types
• terminology
• converting question to experiment
• error variance
7/21/2015
Psyc2013 Class #17 (c) Peter
McLeod
20