*Spend a lot of time on this chapter (18)
Experimental Design
HK 396 Quantitative Research Methods
For the following problems, draw and label the experimental design (independent
variables and levels, dependent variables, name of the design), suggest the possible
major threats to internal and external validity. Suggest why each design and the levels of
the independent variables were selected. Each student will hand in their own copy
of responses in point form at the end of class.
1. To determine the influence of training on change in strength of associated muscle, an
experimenter randomly assigns 48 untrained female participants to four equal groups.
All groups (except controls) train for 16 weeks, three times per week, one hour per
training session, and all do the same strength-training routine. The difference among the
groups is their strength-training load, which is based on 1RM and adjusted every four
weeks: 80% of 1RM, 60% of 1RM, 40% of 1RM, and controls (who do not train). All groups
are pretested; tested after 4, 8, 12, and 16 weeks; and tested for detraining effects after
20 weeks. The groups’ 1RM strength is measured at each time period, and muscle tissue
is taken for fiber typing.
- Independent
o Time (6- pre, 4, 8, 12, post, and detraining)
o Training Load (4 – 80%, 60%, 40%, control)
- Dependent
o 1RM strength
o Fiber type
- Pretest-Posttest randomized groups design (of course, testing is extended to 6
time periods)
*Chapter 18 – most common in kines.
*Randomized does not refer to random selection
- Threats to internal and external validity – ch.18
- 4x6 MANOVA with repeated measures on the 2nd factor.
*we didn’t go over MANOVA: M = more than one dependent variable
R
R
R
R
Observa
tion1
7
O13
O19
Test2
O2
T2
O3
T3
O4
T4
o5
O6
T5
T9
T13
O8
O14
O20
T6
T10
T14
O9
O15
O21
T7
T11
T15
O10
O16
O22
T8
T12
T16
O11
O17
O23
O12
O18
O24
2. A researcher wants to investigate the influence of relaxation techniques on expert male
bowlers. Expert bowlers (n = 45) are selected and randomly assigned to three equally
sized groups. For 5 minutes prior to bowling a game, one group goes through a
Dr. Sasho MacKenzie – HK 396
relaxation training protocol using a videotape, the second group watches a videotape of
a professional bowling match, and the third group does mental arithmetic problems. All
groups then bowl a regulation three-game series.
- Independent
o Type of pregame training for bowlers (3 levels)
- Dependent
o Bowling performance – sum of the 3 game scores
- Randomized-groups design
- Threats to internal and external validity:
o Threats to external – not likely that bowler’s will be in these situations
before a game. Solution could be to simulate other situations, make one
bowler feel like he’s really late, for example.
o Threats to internal –
R
T1
O1
R
T2
O2
R
T3
O3
3. The purpose of a study is to evaluate the influence of carbohydrate loading on 10K
running performance. Forty 10K runners are randomly assigned to four equal groups. All
agree to run a series of five 10K races during a 16-week period (one race every 4 weeks)
and to maintain the same level of training during the 16 weeks. All runners’ 10K times for
each race are recorded. Before the second race, one group of 10 runners follows a specific
protocol of carbohydrate loading. Before each subsequent race (third, fourth, and fifth),
a different group follows the same carbohydrate loading protocol, while the other four
groups do not.
- Independent
o Time (5 races )
o Carbohdrate loading (1)
- Dependent
o 10K running time
- Switched-replication design (4 groups x 5 times)
- Regression techniques (not ANOVA)
- Threats to internal and external validity:
R
O1
T1
O2
R
O6
O7
R
O11
O12
O13
R
O16
O17
O18
T2
O3
O4
O5
O8
O9
O10
O14
O15
T3
O19
T4
O20
Dr. Sasho MacKenzie – HK 396
4. This study seeks to investigate the influence of behavioral games and rewards on
playground disciplinary problems. A group of 60 third-grade children with misbehavior
problems are randomly assigned to a group that receives a behavioral game treatment
(groups compete for fewest misbehaviors on the playground with some training by
teachers on what misbehaviors are and how to handle problems) or a group that does
not. Within each group, children are randomly assigned three types of rewards (tied to
the number of misbehaviors): no rewards, free time, or special food choice. The children
are videotaped on the playground, and misbehaviors are determined by use of a coding
instrument before and after the treatment period. Each child’s misbehavior score is the
difference in number of misbehaviors before and after the treatment.
- Independent
o Behavioural games (2 – receives training or not)
o Reward type (3 – none, free time, special food choice)
- Dependent
o Misbehavior score = difference between misbehaviours before and after
treatment
- Design: factorial randomized groups (not a pretest-posttest design because a
difference score is used) ***Sasho disagrees with this – he considers it a pretestposttest
- 2x3 ANOVA
- Threats to validity:
o Internal o External -
Dr. Sasho MacKenzie – HK 396