research lecture 6 Flashcards
if it’s a two-tail test….you interpret the p
value in SPSS using ___. SPSS is adjusting it for
you (dividing it in half)
.05
if you need to adjust the alpha
further…ex/ Bonferroni adjustment for 2 t-
tests….then compare the SPSS p-value to ___
to determine significance.
.025
what is the difference between T-test and ANOVA
T-test: compares 2 means and only has one IV
ANOVA: compares 3+ means and only used when u have more then 1 IV
can you use a T- Test from a one way repeated measures
yes bc only 1 IV
can you use a T-test from a posttest only randomized group and posttest only non randomized
yes
if we want to compare several means why don’t we
compare multiple pairs of means with t-tests?
inflates the type 1 error rate
like a t-test, ANOVA tests the ____ hypothesis that the means are approximately the same.
null
what does a experimental hypothesis mean for an ANOVA
the means differ between each other
ANOVA is an ____ test
Omnibus
ANOVA is an Omnibus test .. what does it test, tell us and doesn’t tell us
-It tests for an overall difference.
– It tells us that the means are different.
– It doesn’t tell us exactly which means differ.
what is An “overall test“ - tests a number of different comparisons at once
omnibus
what does an 2x3x4 ANOVA tell us
3 IV
on has 2 variables
one has 3 variables
one has 4 variables
what ANOVA is it called if there is 2 or more IV ar least one is independent and one is repeated
2+x2+ mixed or split plot ANOVA
how do u find the mean squares
sum of squares / degree of freedoms
how many DF is there for an ANOVA
these are always 2 .. one has to do with the amount of groups minus 1 and other one has to do with the number of ppl
can the F result from the omnibus ANOVA tell us if it is significant
no not by itself
a significant omnibus (global) test tells us that there is a difference among the ___ or the group ___ are different somewhere
means 2x
what does a significant omnibus does not tell us
specifically which group means different from which
do you use follow up test with a significant omnibus test
yes always
how do you do follow up test for an omnibus test?
planned contrasts or post HOC comparisons
do not use multiple t test
is a planned contrasts or Post HOC comparisons get significance easier ?
planned
what is a planned contracts
planned prior
ex: u only want to compare
the placebo group to each tx.
group.
what is a post HOC comparisons
not planned (no hypothesis)
compares all paris of means
when do u use a post hoc
any stat test after global omnibus ANOVA (usually has 2 means)
what do u run after global (omnibus test) and have significant interaction
simple effects
when do u do a pairwise
usually after global test (only 2 means)
what are the post hoc statistical test designed to control
the alpha level ( basically control the family wise (overall) error rate)
what are the 3 most common post hoc test
- bonferroni
-turkey
-games-howell
what post hoc test do.u use when there are only small deviation form normality , sample sizes in each cell are equal and the variances are homogeneous
-bonferroni
-turkey
what post hoc test do u use if variances are not homogenous
games-howell
bonferroni and turkey are conservative and controls type __ error when is each more powerful
1
bonferroni is more power if the number of comparisons are small and turkey is more power if the number of comparisons are large
what part in the research do u include the significance difference in
the results
what is the non-parametric equivalent to an independent one-way ANOVA.
kruskal - wallis test
when do u select the kruskal walllis test
when data doesn’t meet assumptions
what does kruskal wallis test compare
compares mean ranks among the groups
what are the follow up test for the kruskal wallis test
mann - whitney u test ( non-parametric equivalents to independent t-tests)
for a repeated measures one way anova what do u test first and then what do u follow up with
perform a global or omnibus test first (is there a difference somewhere) and then follow up with bonferroni post hoc comparisons
what is the assumptions that the variances of the differences between all possible pairs of groups (ex/ levels of the repeated IV) are equal.
sphericity
what is the test for the assumption of sphericity called
the manuchleys test
do you want the Manuchley’s test to be significant or not
no
what values are used in a RM one way ANOVA if the assumptions (sphericity) is violated
EPsilon
- greenhouse
-hunyhn feldt
- lower bound
if your sig is .188 on a RM one way anova is sphericity violated or not
no it has not been violated bc p > .05 it is only violated if the sig is < .05
what does the 1St Degree of freedom for an anova represent
related to the # of groups - one
what does the 2nd Degree of freedom for an anova represent
the # of participants per group
what is the non parametric equivalent to a RM – one-way ANOVA
freiedmans ANOVA
when runngin a friedman ANOVA, In SPSS this will give you the global result (is
there a difference somewhere)…and all pair-wise
comparisons are analyzed using ____ ___ __
Wilcoxon Signed Ranks.
what is a two way independent anova
two + IV
in a 2 way independent anova there are different participants in what condtions
all
if there is more than one IV what is it called
a factorial design or a factorial anova analysis
what is a benefit of a factorial deigns
we can look at how variables interact
what do the interactions of the factorial design tell us
the effect of one IV depends on the vaule of another IV
the first analysis of the ANOVA consists of ___ test
3
what is included in the omnibus or global ANOVA
interaction
main effect for one IV test
main effect for second IV
if the ___ is significant u don’t have to worry about the main effect
interaction
on a graph when will the interaction not be significant
when they are parallel to each other
when do u run simple effects
run after global and significant interaction
if you run a global anova (interaction and main effects) and the main effect(s) and the interaction is not significant what do u do
run follow up main effect pairwis comparisons (post HOC comparison and adjust the alpha levle
if there is only 2 levels for the main effect do u do a follow up test
no
if you run a global anova (interaction and main effects) and the main effects are significant and the interaction is significant what do u run after
run simple effects and “forget: about the main effects and adjust the alpha level
when do u run simple effects
only run if u have a significant interaction
when u run simple effects can u adjust ur alpha ?
yes ofc
you can only look for difference inside the ANOVA cells if what
u have a significant interaction
•You can ONLY look for differences inside the ANOVA cells IF you have a significant interaction.
Otherwise you are forced to ONLY report the what
outcome of the main effects .. the main effect post hoc test if more then 2 levels
if u have to many levels from ur anova what happens to the alpha
it get smaller
to test for differences between group means when
we know that an extraneous/confounding variable
can affect the ____ variable.
outcome
if u can identify the covariate and measure it what does that do to your analysis
strengthens it and more likely to find significance
First, we use SPSS to determine if compliance really did affect the pain scores…if yes….we use SPSS to adjust the means to take this into account….so the difference isprimarily based on the intervention and not compliance.. any time an ANOVA is done this way what is it called
ANCOVA ,,, an analysis of covariance
when we want to test for difference between groups when we have several DV’s what do we run
MANOVA
why is running a MANOVA when testing for difference between groups when we have several DV’s better then running multiple ANOVAs
-controls family wise error rate (type 1)
- takes into account the relationships between the DVs , so more powerful then seperate ANOVAs
what is the next step after a significant global MANOVA
DFA (discriminating function analysis )
what is the discriminant analysis used to determine
which of the groups 4 DV’s are best at discriminating (separating) the groups
