Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
19 Cards in this Set
- Front
- Back
background for hypothesis testing
|
1.expose participants to 2 or more levels of IV
2.determine size of f-ratio by comparing between and within group variance 3.set the cut-off for acceptable prob. that results are due to chance (alpha level) 4.calculate actual prob(p-value) |
|
P-value less than Alpha level
|
prob that difference is due to chance is acceptable
Null hypothesis rejected statistically significant difference b/w the groups |
|
P-value more than alpha level
|
prob. that difference is due to chance is not acceptable
Null hypothesis is accepted NOT statistically significatn |
|
Null hypothesis
|
no difference between experimental and control groups in terms of DV
hypothesis of "no difference" observed difference due to change, not IV |
|
testing Null hypothesis
|
reject null when prob is small <.05
accept null when prob is large <.05 |
|
alternative hypothesis
|
2 groups are different in terms of DV
hypothesis of "differences exist" observed difference due to IV (hopefully) alt.rejected if null accepted alt.accepted if null rejected |
|
type I error
|
when reject Null, but shouldnt have
results say there is difference between groups..but difference is due to chance prob of type I error=Alpha |
|
setting smaller alpha
|
..and get significant results:
conservative more confidence that difference not due to chance |
|
if set larger alpha value
|
..and get significant results:
liberal less confidence difference is not due to chance |
|
Power
|
=ability of a statistical test to detect a relationship between variables (corr) or differences between groups (exp) when they exist
|
|
power + alpha level
|
smaller alpha reduces power, bigger alpha increases power
requires more between-group variance to show a difference usually experimenter will not set higher than 0.5 |
|
power + effect size
|
(overlap between populations)
larger(less overlap)=increases power manipulate IV using extreme differences b/w levels to show a difference potential solution: pilot study |
|
power + within group variance
|
less within group variance increases power
reduce it by reducing reliabilty of DV, consistent treatment of participants |
|
power + sample size
|
increasing sample size increases power
larger the sample,the more confident we can be that the result was not due to chance |
|
determining power
|
can determine # of participants required to get statistically significant findings
use large samples |
|
ANOVA
|
can test any number of groups simultaneously to look for a difference
Tests the Null that no difference among all possible pairings of groups A thru D simultaneously |
|
T-test
|
test for differences between 1 pair of groups t a time
(ex. AandB, AandC, AandD..) |
|
probability pyramiding
|
with 4 levels (6 possible pairings), prob of Type I error increases from 0.5 to 0.26
(26% differences due to chance) with 3 levels, probability of Type I error increases from 0.5 to 0.14 |
|
Post-Hoc
|
"after the fact" test
tells which pairs of groups are statistically different |