Randomization

1
Randomization

• C8-1. Know what Randomization is, what it
  accomplishes (mention 'unknown sources of
  variation')(264), and how to do it (exhibit 8.1,
  page 266). What does "representative" mean?
  (267).
• Randomization The only known technique for
  controlling unknown sources of variation every
  member of a population has an equal chance of
  being selected. (participant variables/extraneous
  variables)
• Representative (Random selection) The extent to
  which a sample is similar in composition to a
  population.
• Random Assignment should Distribute potential
  extraneous variables equally to various groups.
• --by chance may get bias
• --the larger the sample, the less likely bias
  will occur

2
Random Selection and Assignment

• C8-2. What does Random Selection (sampling)
  assure, and what does Random Assignment Assure?
  (267,0)
• Random Sampling (selection) assures a
  representative sample
• Random Assignment assures a equal distribution of
  extraneous variables (participant variables) thus
  making your control and experiment groups equal
  in those potential confounding variables.

3
More Extraneous vars

• C8-3. Be able to define matching and know the two
  benefits (273-274,0)
• Matching equating groups on one or more
  variables by measuring participants on those
  variables and assigning them in equal amounts to
  the various groups.
• ADV.
• increases sensitivity of experiment (the DV)
• More balanced groups
• DisADV time-consuming pre-measuring induces
  demand characteristics
• Precision Matching each participant is matched
  to another on some variable.
• Example
• Step 1 Measure on variable suspected of having
  impact
• Step 2 put in order
• Step 3 Block according to number of groups
• Step 4 RA participants from a block to one
  group repeat this for each block.

4
Matching

• C8-4. Be able to define matching and know the two
  benefits (273-274,0)
• Three groups IV Rock music, Classical
  music, no music
• DV performance on a math test
• Matching Variable IQ scores
• 112
• 101
• 132
• 117
• 104
• 124
• 136
• 118
• 99
• 113
• 140
• 106

5
Controlling Participant Effects

• C8-5 Be able to briefly describe each of the 3
  ways to control for participant effects

1. Double Blind Placebo Experimenter and
participant are unaware of the treatment
administered. 2. Deception Disguising the real
experiment by telling the participant a fake
version of the experiment -can use a little
deception or a lot of deception, but be
careful! -Satisfy curiosity so the P is less
likely to devise their own hypothesis 3.
Procedural Control An attempt to control for P
interpretation of the experiment, we also gain
insight into P perception. a. Exit Interview
(Retrospective verbal report) b. Concurrent
Verbal Report -think aloud technique -sacrifi
ce groups -concurrent probing
6
Controlling Experimenter Effects

• C8-6 Explain how to control for experimenter
  expectancy effects

• Dealing with expectancy
• a. The blind technique
• Experimenter is unaware of the conditions
• b. The Partial Blind Technique
• Knowledge of each research P's treatment
  condition is kept from the experimenter through
  as many stages of the experiment as possible
• c. Automation
• -Instructions are written, tape recorded,
  filmed, televised, or on computer.
• -Minimizes experimenter interaction with the
  participants

7
BAD RESEARCH DESIGNS

• C9-1 Be able to identify three faulty experiment
  designs and if given a scenario, be able to point
  out the faults.

• One group, Posttest only
• 1 group ? Treatment (IV) ? Posttest (DV)
• No way to know if treatment had an effect
• No control for extraneous variables

• 2. One group, Pretest-Posttest
• 1 group ? Pretest? Treatment (IV) ? Posttest (DV)
• History and maturation issues
• Pre Post test may not be equal
• Practice effects
• Tip off participants Hypothesis demand
  characteristics

8
BAD RESEARCH DESIGNS

• C9-1 (cont.) Be able to identify three faulty
  experiment designs and if given a scenario, be
  able to point out the faults.

• Nonequivalent groups/participants, Post-test
  only
• 2 or more groups ? various levels of IV? Posttest
  (DV)
• Difficult to ensure matching groups
• No random assignment is bad
• Too small of a sample size is bad

9
GOOD RESEARCH DESIGNS

• C9-2 Know the 3 requirements of true
  experimental design

How would you measure the effect of Cheesy Poofs
on Aggressive Behavior?

• The design answers the research question
• 2. Extraneous variables are controlled for
• 3. Generalizability Results can be applied to
  individuals outside of the experiment (it has to
  be relevant of course)

10
Experimental and Control Groups

• C9-3 Know what an experimental group is. Know
  what a control group is and the two functions
  that a control group serves

Experimental Group Group of participants that
receives the treatment (some level of the IV)
Control Group 1. Receives no IV, or a
traditional level of it 2. Standard of
Comparison 3. Control for Rival Hypothesis If
changes occur in both groups extraneous
variable..it allows for the detection of
extraneous variables
11
True Research Design Pretests

• C9-4 Know why you would want to pretest
  participants

• Why Pretest?
• Increased Sensitivity compare pre and post test
  scores (small changes can be noticed)
• Tests for sensitivity of DV ceiling effect and
  floor effect
• Tests for initial attitude of participant (can
  affect scores)
• Initial Comparability shows if P's are randomly
  assigned on relevant variables (groups are
  initially equal)
• Evidence of change
• Disadvantages
• Expensive Demand Characteristics

12
Between and Within Designs

• C9-5a Know the difference between Within
  participants designs and Between participants
  designs

Between Participants (groups) Designs different
participants used in each condition Within
Participants (groups) Designs The same
participants are used in all experimental
conditions Mixed Design A combination of within
and between (more later)
13
Between and Within Designs

• C9-5a Know the difference between Within
  participants designs and Between participants
  designs

Mixed Design
14
Between and Within Designs Posttest only

• C9-5b Know the three post test only designs . Be
  able to briefly describe them.

• Between Participants (groups), Posttest Only
• Participants randomly assigned to a condition
• Groups get treatment
• DV measured
• Simple Randomized Participants (groups) design
• a between participants, post test only design
• used to measure multiple levels of the IV
• Within Participants (groups), Posttest Only
• all Ps participate in all conditions, RA, must
  counterbalance administration of IV (which
  treatment a P gets first)

15
Factorial Design

• C9-6a Know what a factorial design is, know
  what they are used to determine, and be able to
  identify the total number of groups, IVs, and
  levels, given a factorial design

• What is a factorial design?
• 2 or more IVs studied (often with multiple
  levels)
• Participants randomly assigned to a condition
• A cell is a specific treatment condition
• Factorial Designs are used to determine
• A. Main effects The effect of one IV alone
• B. Interaction Effects the effects one IV may
  have on different levels of another IV
• Examples A 3x3 factorial design has 9 groups
  total
• A 3x3 factorial design has 2 IVs
• A 3x3 f.d. has 2 IVs with 3 levels each.

16
More Factorial Designs

• C9-6a Know what a factorial design is, know
  what they are used to determine, and be able to
  identify the total number of groups, IVs, and
  levels, given a factorial design

• A 2x3x6 has how many groups? IVs? Levels per IV?
• 36 groups, 3 IVs, IV1 2, IV2 3, IV3 6

• A 3x2x2x2 has how many groups? IVs? Levels per
  IV?
• 24 groups, 4 IVs, IV1 3, IV2 2, IV3 2, IV4
  2

• A 2x2 has how many groups? IVs? Levels per IV?
• 4 groups, 2 IVs, IV1 2, IV2 2

17
More Factorial Designs

What kind of a factorial design is this?
Hours spent practicing with nunchucks (per wk)
2 4 8
Hours spent watching Walker Texas Ranger (per
wk)
2
4
DV Acts of Justice (how would you operationally
define this?)
18
More Factorial Designs

What kind of a factorial design is this?
2 x 3 factorial
Hours spent practicing with num-chucks (per wk)
2 IVs
1 3 9
Hours watching Walker Texas Ranger (per wk)
1 IV has 2 levels The other has 3
1
8
DV Acts of Justice Delivered (how would you
operationally define this)
19
More Factorial Designs

What kind of a factorial design is this?
Hours spent watching Mathew McConaughay
2 4
Hours spent watching Johnny Depp
2
4
DV Level of excitement (how would you
operationally define this?)
20
Interpreting Charts and Graphs

• 9-6b given a graph and a chart with raw data, be
  able to interpret if there is an interaction.
  Also be able to calculate row and cell means to
  analyze any potential main effects

Hrs a day practicing with nunchucks
2 4 8
Low
Fondness for Chuck Norris
1, 7, 4
2, 1, 3
3, 9, 6
High
0, 1,2
6.5, 4, 3
10,7, 7
DV Acts of Justice
21
2 4 8
Low
3, 9, 6
1, 7, 4
2, 1, 3
4
High
4.5
0, 1,2
6.5, 4, 3
10,7, 7
3.5
4.25
5

• To investigate main effects, calc. MEANS of each
  cell, then calc. the means of each row and
  column and compare. Since there are 2 IVs
  there are 2 possible main effects.
• Graph CELL means, to determine if there is an
  interaction Interactions complicate main
  effects.

22
2 4
Low
2,6,4
4,4,4
4
High
6, 6, 12
8
10, 6, 8
6
6
Graph it on the board
23
Rules for interactions on graphs
9-6b given a graph and a chart with raw data, be
able to interpret if there is an interaction.
Also be able to calculate row and cell means to
analyze any potential main effects

• Parallel lines No Interaction

• Intersecting lines Yes Interaction

• Lines that look like they will intersect if
  continued yes interaction

24
Rules for interactions on graphs
NO Interaction
Interaction
NO Interaction
Interaction
Interaction
Interaction
25
Interpreting Charts and Graphs

• 9-6b given a graph/chart, be able to interpret if
  there is an interaction

Hrs a day practicing with nunchucks
2 4 8
Low
6
4
2
4
High
1
4.5
8
4.5
3.5
4.25
5
Interaction? Yes
DV Acts of Justice
26
Advantages/Disadvantages of Factorial Design

• C9-7. Be able to list the advantages and
  disadvantages of the factorial design

• Advantages
• 1. Test more IVs, with various levels
• 2. Can include extraneous variables into
  experiment (Gender, Age, etc.)
• 3. More like real life experience

• Disadvantages
• 1. More participants are needed
• (more factors more P's)
• 2. Hard to consistently manipulate more than 2
  variables at the same time
• 3. Hard to interpret multiple interactions

27
Between Participants Pretest Posttest Design

• C9-8. Describe the between participants pre-test
  posttest design and explain how it controls for
  extraneous variables

• Pretest-Post test design
• 1. Participants randomly assigned to 2 or more
  groups and pretested (control, G1, G2)
• 2. treatment (IV) is delivered, and groups are
  posttested on DV.

• Experimental Control
• 1. Control group allows us to assess history and
  maturation. Differences will show in both groups
• 2. RA controls for uniform maturation and
  selection variables by equally distributing them
  across groups.