1. Slides to accompany Weathington, Cunningham & Pittenger (2010), Chapter 9: A Model for Research Design

2. Objectives Importance of good design Independent variable
Confounding variables
Generating and evaluating hypotheses

3. Importance of Good Design
Good science requires good research design
Important to understand and try to prevent common sources of bias at each phase of the research process
Figure 9.1
Especially beware of demand characteristics and self-fulfilling prophecies
You may want to consider discussing/working through each phase of Figure 9.1 with the class (next two slides can help)

4. Figure 9.1, part 1

5. Figure 9.1, part 2

6. Independent Variable Issues
Subject vs. manipulated
Age vs. amount of caffeine
Between- vs. within-subjects
Testing participant in one condition vs. across multiple conditions (or over time)
Fig. 9.2

7. Between- vs. Within-Subjects IV
Figure 9.2

8. Dependent Variable Issues
Seek reliable and valid DVs
Reduce bias when possible
Single- and double-blind procedures
Consider use of placebo or control group
Cover stories and deception may help strengthen your manipulation
Should follow-up with a manipulation check
Single-blind = participant does not know the hypothesis (reduces demand characteristics)
Double-blind = data collector and participant both do not know the actual hypotheses
Placebo/control group option helps to clearly identify the effects of the manipulated IV vs. ordinary effects/processes
A manipulation check = test of whether manipulation is working as desired

9. Confounding Variables
We need to carefully consider possible sources of confusion
Confounding variables
Common example is a carryover effect
Previous experiences influencing future behavior/responses to stimuli

10. Hypothesis Details
Hypotheses state relationships between/ among variables (Table 9.1)
Directional vs. nondirectional
As mathematical statements
Null and alternative/research versions are necessary
H0 true until evidence suggests false (innocent until shown guilty)

11. Errors in Hypothesis Testing
Hypothesis testing relies on probabilities and is conditional
Trying to achieve a certain degree of confidence that your data do not support Ho
Type I error: Ho is true, but we reject it
False alarm
Type II error: Ho is false, but we retain it
Miss
Consider recreating Table 9.2 on board – discuss alpha and beta and their relationships
Type I error like convicting an innocent person
Type II like setting a criminal free (fail to be sufficiently convinced that Ho is wrong and the criminal is really guilty as described by H1)

12. Evaluating Hypotheses
Is the evidence strong enough to reject Ho?
Can consider effect size (d) as relative difference between two M
Larger ES = stronger IVDV relationship
Influenced by between- and within-groups variance (Fig. 9.4)
d formula is equation 9.1
Between-groups variance = difference among group means (true diffs between mu plus effect of random sampling or measurement errors).
Can increase this by strengthening your IV manipulation and in ensuring the groups to be compared are as different as possible
Within-groups variance = variability among scores within a group (natural variation and measurement error).
Can be dealt with by using homogeneous samples and reliable tests

13. Figure 9.4
Highlight the influence of both between- and within-groups variance on effect size

14. What is Next?
**instructor to provide details