Exam 3 Flashcards
extraneous variable
a variable that is not systematically manipulated in an experiment but that still may affect the behavior being observed
repeated measures design
same participants participate in different treatment conditions
independent measures design
two different groups participants participate in different treatments
null hypothesis (Ho)
the iv has NO EFFECT on the dv
Ho: mu1 = mu2
alternative hypothesis (Ha)
the iv HAS AN EFFECT on the dv
H1: mu1 > (< or = with slash) mu2
directional
increase or decreases
ONE-TAIL TEST
nondirectional
has an effect or doesn’t have an effect
TWO-TAIL TEST
alpha level (a)
defines the maximum probability that research result was obtained
p-value
indicates how likely it is that a result occurred by chance
type 1 error
you reject the Ho when you should’ve retain the Ho
type 2 error
you retain the Ho when you should’ve reject the Ho
effect size vs practical significance
significance: whether or not there was a difference and how likely it would occur by chance alone
effect size: how large the difference was
decision rule for obtained probability
obtained probability </= a -> reject Ho
obtained probability > a -> retain Ho
why do we evaluate Ho first?
easier to disprove our hypothesis than prove it because we can never 100% prove something
power
the ability to detect an effect when one is present
- value can vary from 0 to 1
power (a priori use)
determine sample size necessary to detect an effect
power (a posteriori use)
determining whether sample size and research were adequate to detect an effect
the effect N has on power
N increases = power increase
the size of real effect on power
effect size increases = power increases
the effect of alpha level on power
alpha level closer to 1 = stronger power
alpha level closer to 0 = weaker power
explain the relationship between power and beta
the power of a test is the probability of rejecting the Ho, given it is false
- power= 1-Beta
- power + beta = 1
why do we never accept Ho and instead reject Ho?
we cannot assert that Ho is true thats why we reject it
distribution of sample means
the collection of sample means for all the possible random samples of a particular size that can be obtained from a population
sampling distribution of a statistic
a distribution of statistics obtained by selecting all the possible samples of a specific size (n) from a population
characteristics of distributions of sample means
- sample mean pile up around the population mean
- sample mean is approximately normal in shape
- larger the sample size the closer the sample mean should be to the population mean
central limit theorem
when n is large the distribution of the sample means will approach a normal distribution
critical region
area under the curve that contains all the values of the statistic that allow rejection of the null hypothesis
critical value
the value that bounds the critical region
conditions for use of the one-sample z test
- N >/= 30 (normal distribution)
- Xbar is known
- mu and population standard deviation is known
t test vs z test (single sample)
z test: needs population mean and standard deviation
t test: needs population mean and sample standard deviation
degrees of freedom (single sample t test)
the number of values that are free to vary when calculating the statistics
df: N- 1
T distribution characteristics
- flatter than the normal distribution
- more spread out than the normal distribution
- more variability in t distribution
conditions to use t-test for single sample
- only 1 sample
- population mean is known but not population SD
- sample SD is known
- sampling dist. is normal
sampling distribution of t
sample means to the population mean when the population standard deviation is not known
cohen’s d effect size
small effect: d </= 0.2
medium effect: d = 0.21-0.79
large effect: d >/= 0.8
confidence intervals
range of values that probably contains the population value
- are always TWO-TAILED AND NON DIRECTIONAL
confidence limits
the values that bound the confidence interval
correlated group designs
two sets of data with no population parameter
1. repeated measures design
2. matched subject design
repeated measures design
same participants receive every level of IV
matched subject design
participants are matched on a specific variable to hold constant
assumptions for a t test for correlated groups
- N >= 30 (normally distributed)
- participants are the same in each condition or matched
- IV is nominal and DV is interval or ratio
t test for independent samples
use to analyze the mean difference between the two groups
assumptions for t test for independent samples
- two scores of sampling
- IV is nominal with 2 levels; DV is interval or ratio
- N >= 30
- homogenity of variance
- each condition contains groups of separate people
- participants receive only one level of the IV
estimated standard error of the mean difference (independent groups)
tells us how far away, on average, two sample mean would be from each other if the null is true
are t-test or z-test more powerful?
t-test are less powerful because the critical t’s are larger than the z
test decision rule
|tobt| > |tcrit| -> reject Ho
z-test formula
z = Xbar - mu / (PopSD/sqrtN)
single sample t-test formular
t= Xbar - mu / (sampleSD/ sqrtN)
correlated sample t-test formula
t= Dbar / sqrt(SSd/N(N-1))
SSd = sigma D^2 - (sigma D)^2/N
independent sample t-test (same size) formula
t= Xbar1 - Xbar2/ sqrt (SSxbar1 + SSxbar2/ n(n-1))
independent sample t-test (different size)
t= Xbar1 - Xbar2/ sqrt (SSxbar1 + SSxbar2 / n1 + n2 - 2) (1/n1 + 1/n2)
degrees of freedom t independent sample test
n1 + n2 - 2
t-test correlated groups with raw date
t= Dbar - mu / sqrt (SSd/N(N-1))
where SSd = sigma D^2 - (sigma D)^2/N
MU IS ZERO