Week 12 Flashcards
SE
the SD for the distribution of sample means
For larger samples, the SE gets smaller
A-level
Used to define the very unlikely outcomes if the Ho is true
Probability of rejecting the Ho when it is true, or making the wrong decision
AKA level of significance
a=0.05 (5%)
critical region
AKA region of rejection
Outcomes that are very unlikely to be obtained if the null hypothesis is true
‘very unlikely’ is determined by ‘a’, and the boundaries are determined by the alpha level
If sample data fall in the critical region, we reject the Ho
critical region boundaries
Use a-level and unit normal table and determine the precise z score location of critical region boundaries.
For α = .05, we split 5% into two –>2.5% in each tail. We then calculate z-score corresponding to p = .025 in the tail. Thus,z = ±1.96.
one sample t test
One sample t test: population mean known
independent t test
compare the means of two independent groups
paired t test
compare two means from the same sample
effect size
Aim to quantify the magnitude of a relationship or treatment effect
cohens d
tells the degree of separation between two distributions; how far apart are the means of two distributions
mean difference/ SD
type I error
incorrectly reject the null hypothesis (false positive)
Type II error
incorrectly accept the null hypothesis (false negative)
statistical power
probability of detecting an effect if there really is one; probability that the hypothesis test will reject the null hypothesis when there actually is a treatment effect
ways to increase the power of a study
use a higher a-level use a one tailed test increase n use a within-subjects design increase effect size
type II error identified by
beta (B)
power identified by
1-beta (B)
increase power- higher a level
bigger critical region= greater power
BUT increases type I error rate
increase power- use one tailed test
with a all in one tail, the critical value is smaller = greater power
increase power- increase sample size
more participants=greater power
increase power- use a within-subjects design
eliminating individual differences reduces SE
increase power- increase the effect size
10 weeks instead of 6 weeks of therapy may result in larger difference
parametric statistics
inferential procedures that require certain assumptions about the raw score population represented by the sample; used when we compute the mean
non parametric statistics
inferential procedures that do not require stringent assumptions about the raw score population represented by the sample; used with the median and mode
