Chapter 8: Experimental Design simple designs with 2 levels of IV

1
Chapter 8Experimental Design(simple designs
with 2 levels of IV)
2
Heres a possible research question

• Does crowding affect learning ability?
• How can we design an experiment to test this once
  we define crowding and learning?
• Any ideas, folks?

3
Heres a possible research question

• Possibilities
• 1. Half of people exposed to crowding, half
  exposed to no crowding, then measure differences
• 2. Same people first exposed to crowding then no
  crowding, then measure differences
• Other ideas?

4
More about our research question

• Is there a problem if the room is hotter for the
  more crowded situation than the less crowded
  one?
• Any ideas, folks?
• How can we CONTROL for these kinds of problems?

5
Confounding and Internal Validity

• Experimental method has advantage of allowing a
  relatively unambiguous interpretation of the
  results.
• Manipulated independent variable
• - Create groups (or conditions) that differ in
  levels of IV
• Compares groups (or conditions) in terms of their
  scores on DV
• - All other variables kept constant through
  direct experimental control and/or randomization
  of subjects

6
Confounding and Internal Validity (cont)

• An experiment has STRONG internal validity when
  results (differences in dependent variable) can
  confidently be attributed to the effect of the
  independent variable.
• In other words, little or no possibility
  something ELSE (a CONFOUND) caused the differences

7
Basic Experiments

• Simplest possible experimental design
• Uses ONE independent variable with only two
  levels (an experimental or treatment group or
  condition and a control or placebo group or
  condition)
• Uses randomization of assignment of participants
  to groups or conditions
• Experimental control can be seen by holding
  everything constant EXCEPT for IV manipulation

8
Basic Experiments (cont)

• The basic, simple experimental design with 2
  groups can take one of two forms
• Posttest-only design
• 2. Pretest-posttest design

9
Basic Experiments (cont)

• Posttest-only design the manipulation is done to
  the participants (randomly assigned to different
  groups) and then you measure if there are any
  differences between the groups afterwards

R random assignment to conditions
10
Basic Experiments (cont)

• Pretest-posttest design
• Same as a posttest-only design but adds a pretest
  before the experimental manipulation to the 2
  groups
• This allows the researcher to tell if the groups
  are equivalent at the beginning of the experiment
• Whats the problem if they arent?

11
Basic Experiments (cont)

• Advantages and Disadvantages of the two basic
  2-group designs
• Advantages of the pretest-posttest design
• You can assess equivalency between the two groups
  before you do manipulation

12
Basic Experiments (cont)

• Disadvantages of the pretest-posttest design
• Time consuming and awkward to administer
• Sensitizes participants to what is being
  studied
• Demand characteristics
• Possible solutions to control for demand
  characteristics disguise pretest, embed
  the pretest in another measure,
• concealed observation

13
Another TWO names for a 2-group post-test only
design are

• Independent groups design (aka between-subjects
  design)
• Just like we already talked about, participants
  are randomly assigned to 2 conditions to make 2
  groups (Half of participants are in one group
  half in the other group)

This is NOT what we did in Stroop study but we
could have
14
Can you do a 1-group design? YES!

• Repeated-measures design (aka within-subjects
  design)
• The same participants are observed in BOTH
  conditions, one after the other
• How does this differ from pre-test post-test?

This IS what we did in Stroop study
15
Assigning Participants to Experimental
Conditions (cont)

• Advantages and disadvantages of
  repeated-measures designs
• Advantages
• Fewer participants (more economical)
• Extremely sensitive to statistical differences
• (more likely to detect an effect of the IV
  on the DV ?this is called POWER)

16
Assigning Participants to Experimental
Conditions (cont)

• Advantages and disadvantages of repeated-measures
  designs
• Disadvantages
• Order effects
• - Practice effects (scores often INCREASE
  with time)
• - Fatigue effects (scores often DECREASE
  with time)

17
Assigning Participants to Experimental
Conditions (cont)

• How do we CONTROL for possible confound of order
  effects?
• Counterbalancing techniques
• Complete counterbalancing all possible orders
  of presentation of conditions are used
• Latin square used in experiments with many
  conditions (not just 2)
• Randomized blocks used when conditions are
  repeated many times

18
Developmental Research Designs

• Developmental psychologists often study the ways
  that individuals change as a function of age.
• 1. Cross-sectional method measured at one
  point in time
• 2. Longitudinal method same group observed at
  different times as they get older (best internal
  validity)
• 3. Sequential method combination of 1 2

19
Developmental Research Designs (cont)
20
Developmental Research Designs (cont)

• Design considerations
• Cohort effects possible confound in
  cross-sectional design
• Mortality effects possible confound in
  longitudinal design
• Monetary costs longitudinal costs more
• Difficulty level harder to do longitudinal

21
The End