Upcoming SlideShare
×

# Experimental designs

2,843 views
2,636 views

Published on

let share

1 Like
Statistics
Notes
• Full Name
Comment goes here.

Are you sure you want to Yes No
• Be the first to comment

Views
Total views
2,843
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
111
0
Likes
1
Embeds 0
No embeds

No notes for slide
• Difficult to establish cause-and-effect. Correlational research often done first to establish relationships that may be examined for cause-and-effect. Cause-and-effect are not established by statistics but rather by logical thinking and sound research design. You must establish that no other plausible explanation exists for the changes in the DV except the manipulation done to the IV.
• Random groups, controls for past history, maturation, testing, and sources of invalidity based on nonequivalent groups (statistical regression, selection biases, selection-maturation interaction Investigator must control present history, instrumentation, experimental mortality
• Makes it possible to ascertain that groups were equivalent at the beginning of the study. Not necessary if Randomization was used Large sample size
• Advantages: Fewer subjects needed (less costly) Sensitive to finding statistical differences because of control over participant differences Disadvantages: Order effect (practice, fatigue, carry-over)
• Counterbalancing - all possible orders of presentation are included in experiment Latin squares – a limited set of orders constructed to ensure that Each condition appears at each ordinal position Each condition precedes and follows each condition one time
• Time interval – may counteract the effects of treatment, but also increases time demands on subjects
• Obtain measure of matching variable from each subject Rank from highest to lowest based on score Form matched pairs Randomly assign members of pairs to conditions
• Use Figures 10.1 and 10.2 to demonstrate graphically at end of slide.
• Significant interaction means that the effect that one factor has on the dependent variable depends on which level of the other factor is being administered
• Advantages of factorial designs: Greater protection against Type I error More efficient (1 analysis vs. multiple one-ways) Can examine the interaction (not possible with one-way ANOVAs) Disadvantages: Only for fixed models (levels of IV chosen by researcher) and when subjects are assigned randomly
• Makes it possible to ascertain that groups were equivalent at the beginning of the study. Not necessary if Randomization was used Large sample size
• Makes it possible to ascertain that groups were equivalent at the beginning of the study. Not necessary if Randomization was used Large sample size
• Makes it possible to ascertain that groups were equivalent at the beginning of the study. Not necessary if Randomization was used Large sample size
• Developmental research – studies the ways that individuals change as a function of age; age is the independent variable Longitudinal (similar to repeated measures) Powerful (within subject) but several problems Time consuming Attrition due to move, death, school rezoning may change sample characteristics (e.g., more obese subjects die, leaving non-obese subjects in sample – knowledge about obesity is not changing but rather sample is changing) Subjects become familiar with test items (learning effect or items may cause change in behavior) Cross-sectional (similar to independent groups) Less time consuming, but problems Cohorts – a group of people born at about the same time, exposed to same events in society, and influenced by same demographic trends such as divorce rates and family size. Are all age-groups really from same population? Are environmental circumstances that affect jumping performance the same today for 6 yr olds as they were when the 10 yr olds were 6?
• ### Experimental designs

1. 1. ExperimentalResearch Designs
2. 2. Experimental Design Advantages  Best establishes cause-and-effect relationships Disadvantages  Artificiality of experiments  Feasibility  Unethical
3. 3. Causality Temporal precedence Covariation between IV and DV Eliminate alternative explanations
4. 4. Types of Experimental Designs Simple True Experimental Complex True Experimental Quasi-Experimental
5. 5. Types of Experimental Designs Simple True Experimental Complex True Experimental Quasi-Experimental
6. 6. Simple True Experimental Characteristics Types Variations
7. 7. Characteristics of True Designs Manipulation (treatment) Randomization Control groupCharacteristics of simple true designs One IV with 2 levels (T, C) One DV
8. 8. Types Randomized posttest control group design Randomized pretest-posttest control group design
9. 9. Randomized posttest control group design R T Post R C Post
10. 10. Randomized pretest-posttest control groupdesign R Pre T Post R Pre C Post
11. 11. Advantages & Disadvantages Advantages of pretest design  Equivalency of groups  Can measure extent of change  Determine inclusion  Assess reasons for and effects of mortality Disadvantages of pretest design  Time-consuming  Sensitization to pre-test
12. 12. Solomon four-group design R Pre T Post R Pre C Post R T Post R C Post
13. 13. Variations Independent groups (between groups) Repeated measures (within groups)
14. 14. Repeated Measures Design Advantages:  Fewer subjects needed (less costly)  Sensitive to finding statistical differences Disadvantages:  Order effect (practice, fatigue, carry-over)
15. 15. Dealing with Order Effects Counterbalancing  n!  Latin squares
16. 16. Latin Squares 1 2 3 4 A B D CRow 1 (60) (0) (120) (180) B C A DRow 2 (0) (180) (60) (120) C D B ARow 3 (180) (120) (0) (60) D A C BRow 4 (120) (60) (180) (0)
17. 17. Dealing with Order Effects Counterbalancing  n!  Latin squares  Randomized blocks Time interval between treatments
18. 18. Variations Independent groups (between) vs. repeated measures (within) designs Consider external validity when deciding which design to use.
19. 19. Types of Experimental Designs Simple True Experimental Complex True Experimental Quasi-Experimental Pre-Experimental
20. 20. Characteristics of True Designs Manipulation (treatment) Randomization Control groupCharacteristics of simple true designs One IV with 2 levels (T, C) One DV
21. 21. Complex True Experimental Randomized matched control group design Increased levels of IV Factorial design Multiple DVs
22. 22. Complex True Experimental Randomized matched control group design Increased levels of IV Factorial design Multiple DVs
23. 23. Randomized matched control groupdesign M R T Post M R C Post • Used in small samples ∀ ↑ cost in time & money
24. 24. Complex True Experimental Randomized matched control group design Increased levels of IV Factorial design Multiple DVs
25. 25. Increased Levels of IV Provides more complete information about the relationship between the IV & DV Detects curvilinear relationships Examines effects of multiple treatments
26. 26. Reward Amount \$0 \$1 \$2 \$3DV Performance level (% complete) Amount of reward promised (\$) IV
27. 27. Increased Levels of IVDV Performance level (% complete) Amount of reward promised (\$) IV
28. 28. Complex True Experimental Randomized matched control group design Increased levels of IV Factorial design Multiple DVs
29. 29. Factorial Design >1 IV (factor) Simultaneously determine effects of 2 or more factors on the DV (real world) Between Factor vs. Within Factor ID’d by # of factors and levels of factors 2X2
30. 30. Do differing exercise regimens (hi,med, lo intensity) have the same effecton men as they do on women? 3 X 2 (Exercise Regimen X Gender) 2 factors  Exercise Regimen – 3 levels  Gender – 2 levels  Between factors  DV?  Experimental IVs or Participant IVs?
31. 31. Gender Male Female HighExerciseIntensity Medium Low
32. 32. Do strength gains occur at the same rate in menas they do in women over a 6 mo. training period?Measurements are taken at 0, 2, 4, 6 mo. 2 X 4 (Gender X Time) ? factors  Time – 4 levels  Gender – 2 levels  Between or within factors?  DV?  Experimental IVs or Participant IVs?
33. 33. Time 0 mo. 2 mo. 4 mo. 6 mo.Gender Male Female
34. 34. Cell means, Margin meansMain Effects, Interactions Time 0 mo. 2 mo. 4 mo. 6 mo. Gender Male 50 70 90 130 85 Female 30 60 75 90 64 40 65 83 110 74 Cell means Margin means Grand mean
35. 35. Time 0 mo. 2 mo. 4 mo. 6 mo.Gender Male 50 70 90 130 85 Female 30 60 75 90 64 40 65 83 110 74
36. 36. Parallel lines indicate no Is there a main interaction. effect? Interaction of Exercise Intensity and Gender 70 65VO2 Max (ml/kg/min) 60 55 Male 50 Female 45 40 35 High Medium Low Exercise Intensity
37. 37. Is there a main Interaction of Exercise Intensity and Gender effect? 70 65VO2 Max (ml/kg/min) 60 55 High Medium 50 Low 45 40 35 Male Female Gender
38. 38. Non-parallel lines indicate an there Is interaction. a main Interaction of Gender and Time effect? 140 120Weight Lifted (lbs.) 100 80 Male 60 Female 40 20 0 0 mo. 2 mo. 4 mo. 6 mo. Time
39. 39. Is there a main Interaction Between Gender and Time effect? 140 120Weight Lifted (lbs.) 100 0 mo. 80 2 mo. 60 4 mo. 6 mo. 40 20 0 Male Female Gender
40. 40. Interpretation Always interpret the interaction first (graphical) If no significant interaction, interpret main effects
41. 41.  Advantages of factorial designs:  Greater protection against Type I error  More efficient  Can examine the interaction Disadvantages: ↑ subject # for between factor designs Consider external validity when deciding which design to use.
42. 42. IV A: Exposure to Violence – violent vs.nonviolent videoIV B: Gender – male vs. female B1DV: # ads recalled (0-8) B2 9 B 1 2 1 1 5 3A 5 2 9 5 7 5 5 1 1 2 A
43. 43. IV A: Exposure to Violence – violent vs.nonviolent videoIV B: Gender – male vs. female B1DV: # ads recalled (0-8) B2 9 A: Yes B: No 5 AxB: Yes 1 1 2 A
44. 44. Complex True Experimental Randomized matched control group design Increased levels of IV Factorial design Multiple DVs
45. 45. Do strength gains occur at the same rate in menas they do in women over a 6 mo. training period?Measurements are taken at 0, 2, 4, 6 mo. Time 0 mo. 2 mo. 4 mo. 6 mo. Gender Male 50 70 90 130 85 Female 30 60 75 90 64 40 65 83 110 74
46. 46. Types of Experimental Designs Simple True Experimental Complex True Experimental Quasi-Experimental
47. 47. Characteristics of True Designs Manipulation (treatment) Randomization Control group Less control More real-world Program evaluation
48. 48. Randomized posttest control group design R T Post R C Post
49. 49. Randomized pretest-posttest control groupdesign R Pre T Post R Pre C Post
50. 50. Quasi-experimental Designs One group posttest-only design One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
51. 51. Quasi-experimental Designs One group posttest-only design One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
52. 52. Randomized posttest control group design R T Post R C Post
53. 53. One group posttest-only design(One shot study) T Post No control of IV threats Use?
54. 54. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
55. 55. Randomized pretest-posttest control groupdesign R Pre T Post R Pre C Post
56. 56. One group pretest-posttest design Pre T Post •History •Maturation •Testing Use control group •Instrument decay •Regression
57. 57. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
58. 58. Randomized posttest control group design R T Post R C Post
59. 59. Non-equivalent control group design(Static group comparison design) T Post C Post •Selection bias
60. 60. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
61. 61. Randomized pretest-posttest control groupdesign R Pre T Post R Pre C Post
62. 62. Non-equivalent control grouppretest-posttest design Pre T Post Pre C Post •Can check selection bias
63. 63. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
64. 64. Time series Pre Pre Pre Pre T Post Post Post Post
65. 65. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
66. 66. Quasi-experimental Designs One shot study One group pretest-posttest design Non-equivalent control group design Non-equivalent control group pretest-posttest design Time series Single subject designs (Case study) Developmental designs
67. 67. Developmental Research Designs Longitudinal Cross Sectional Powerful (within  Less time consuming subject)  Cohorts problem Time consuming Attrition Testing effect
68. 68. Choosing a Research Design Best addresses the problem Ethics Cost in time and money Validity (internal & external)