- selective subject loss/attrition
- expt=r effects
- demand characteristics
- placebo effects
- Hawthorne effects
- testing intact groups/lack of random assignment
- regression towards the mean
ceiling effect - truncation of data at the top of a distribution due to limit on highest possible score
floor effect - truncation of data at the bottom of a distribution due to limit on lowest possible score
ex. of a 2 x 2:
two levels of one IV - stats prep: didn't take Psy 302/did take Psy 302
two levels of other IV - sex: male/female
interaction--when the effect of one independent variable differs depending on the level of a second independent variable
|
women |
men |
didn't take Psy 302 |
71 |
80 |
took Psy 302 |
83 |
84 |
Another example: sensitivity to various smells at different times of the day (morning and afternoon)
4 x 2 within subjects factorial design (each research participant smell 4 different vials, at two different times)
|
floral |
sweaty |
cinnamon |
perfumey |
A.M. |
|
|
|
|
P.M. |
|
|
|
|
When to Use Each Statistic:
1 continuous dep variable, 2 groups
(1 factor, 2 levels)
----> t-test
2 continuous variables, no groups
----> correlation (r)
1 continuous dep variable, 3 or more groups (1 factor/inde var, 3 or more levels)
----> ANOVA
(analysis of variance)
1 continuous dep variable, more than one set of 2 or more groups (more than one factor/inde var, each with 2 or more levels)
----> ANOVA
(analysis of variance)
categorical dep variable
-----> chi-square
F ratio =
explained variance (between groups) -------------------------------------------------- error variance (within groups)Statistical power -- ability to demonstrate an effect when there really is one OR ability to reject the null hypothesis when it is untrue
Power= 1- beta
More power with
-- big N (number of subjects or data points)
-- big effect size
-- bigger alpha
- exact replication
- partial replication
Advantages of replication:
-- Helps reduce possibility of alpha error
-- Increases generalizability