Why do we conduct experiments anyway

1
Why do we conduct experiments anyway?
I dunno!

• How do we conduct experiments? One answer

Independent Groups Designs
2
Other Questions???????

• When is manipulation a good thing?
• What makes a good experiment?
• What allows decisions re cause and effect?

3
Experimental Control

• Covariation of IV and DV
• Time-order relationship IV?DV
• Elimination of confounds
• Holding conditions constant
• Balancing

4
A First Independent Groups Design

• Random Groups Design
• Random selection versus
• Random assignmentcomparable groups
• A technique Block randomization

5
The Great Halloween Candy Caper

• How many participants in each of 4 conditions
• ( ) M Ms ( ) candy corn ( ) Kit Kats ( )
  Mary Janes ?????????
• Block 1
• Block 2
• Block 3
• Block 4
• Block ?

1-5-6-6-4-1-0-4-9-3-2-0-4-9-2-3-8-3-9-1-9-1-1-3-2-
2-1-9-9-9-5-9-5-1-6-8-1-6-5-2-2-7-1-9-5-4-8-2-2-3-
4-6-7-5-1-2-2-9-2-3-8-7-5-0-2-4-6-6-1
6
Steps Block Randomization

• Assign a number from 1 to 4 to the respective
  conditions, if there are 4 conditions
• 1M Ms, 2candy corn, 3Kit Kats, 4Mary Janes
• Use random numbers to select 4 sequences of the
  numbers from 1 to 4 to obtain 4 sequences for 4
  randomized blocks
• Skip numbers GT 4
• Skip numbers that repeat a number already in
  sequence
• Result is order of testing the conditions for the
  first 16 participants

7
Order of Testing
8
Issues of Validity.Optimizing vs. no-nos

• External?
• Replication
• Does random assignment produce a representative
  sample?

9
Issues of Validity.Optimizing vs. no-nos

• Internal?
• The problem of Intact Groups
• Subjective subject loss versus
• Mechanical subject loss
• Demand characteristics?
• Placebo controls
• Double-blind experiments
• Experimenter effects?
• Double-blind experiments

10
Another Design.Matched Groups

• Matching task (a pre-test)
• Split-litter technique

11
A Third DesignNatural Groups

• Correlation or causation?
• Problems with causal inferences
• Subject variables cant be manipulated
• Subject variables cant be randomly assigned
• Solution complex designs
• E.g., 2 x 2 Age x amount of dosage
• IV? IV?
• DV?

12
Summary Avoiding Problems Common to All
Independent Designs

• Eliminate confounding (internal validity)
• Select appropriate DV (construct validity)
• Replicate to increase external validity
  (convergent validity)

13
Analysis of Experiments

• Descriptive statistics to summarize results,
  only
• Inferential statistics to determine reliability,
  IV?DV

• Null-Hypothesis Testing

• Confidence Intervals

14
Confidence Intervals

• Sample mean, ? ?
• CIrange of values around ? , at ? confidence
• Question Do CIs for different study samples
  (conditions, groups) overlap?
• No overlap? difference between samples

• Yes, overlap? NO difference between samples

15
Constructing CI

• For a 95 CI
• Upper limit (t .05)( )
• Upper limit - (t .05)( )

16
Null-hypothesis Testing and Decision Errors

• Focus on mean differences
• Assume no effect for the null ( no difference)
• Use probability theory
• Decision errors
• Limitations
• Statistical significance vs. real significance
  (meaningfulness)
• Internal validity
• Truth of the null
• Reliability

17
Null-hypothesis Testing and Decision Errors

• Focus on mean differences
• Assume no effect for the null ( no difference)
• Use probability theory
• Decision errors
• Limitations
• Statistical significance vs. real significance
  (meaningfulness)
• Internal validity
• Truth of the null
• Reliability

18
Null-hypothesis Testing and Decision Errors

• Focus on mean differences
• Assume no effect for the null ( no difference)
• Use probability theory
• Decision errors
• Limitations
• Statistical significance vs. real significance
  (meaningfulness)
• Internal validity
• Truth of the null
• Reliability

19
Null-hypothesis Testing and Decision Errors

• Focus on mean differences
• Assume no effect for the null ( no difference)
• Use probability theory
• Decision errors
• Limitations
• Statistical significance vs. real significance
  (meaningfulness)
• Internal validity
• Truth of the null
• Reliability

20
Null-hypothesis Testing and Decision Errors

• Focus on mean differences
• Assume no effect for the null ( no difference)
• Use probability theory
• Decision errors
• Limitations
• Statistical significance vs. real significance
  (meaningfulness)
• Internal validity
• Truth of the null
• Reliability