Final ANOVA study guide Flashcards
type one error
rejecting a true null hypothesis
alpha
type 1 error
type 2 error
accepting a false null
beta
type 2 error
decreased alpha leads to what
increased beta & vise versa
increased type 2 error leads to
decreased level of significance, decreased type 1 error, less power, lacking ability to detect significance when it does exist
increased type 1 error leads to
increased level of significance, decreased type 2 error, greater power, increase likelihood of telling significant difference (power)
power
probability of rejecting a false null hypothesis and obtaining a statistically significant result
power equation
1-beta
if power =.8
find significance 80% of the time
ways to increase power
increase sample size, increase type 1 error rate, homogenous groups, intensify tx’s to observe greater differences, one tail over two tail, stronger design in pair t-test rather than pooled
a decreased type 1 error
decreased power and less likely to tell significant differences
p < alpha
rejecting null
p > alpha
fail to reject null (retain it)
Beta
probability that x has no effect when it does __ % of the time
alpha
probability of concluding x has an effect when it does NOT __ % of the time
power = .99
b=.01, effects of x truely have an effect on y, it will be detected 99% of the time
CR Assumptions
variances are equall (homogeneity of variance), normally distributed across groups of DV , no outliers, indendence observed in design
making F more robust
increase sample size
post hoc liberal tests
greater power, increased type 1 error, greater ability to detect a significant difference when it does exist
RM & RMF assumptions
independence observe in design, normally distributed DV, homogeneity of variance, sphercity (homo across p levels)
If sphericity is violated and ANOVA is sig
researcher is confident in result, type 1 error < alpha
increased type two error rate with violated sphericity
insig anova and –> less confidence in results
CRF-pq assumptions
independence, normality across DV groups, homogenity of variances; tx combos=SS, interaction between Tx effects sig?, fix or random tx, no interatiion BTWN groups
SPF assumptions
independence, Tx level a normally distrubuted, tx a observations have equal variances, sphericity, no interaction between tx groups
ANCOVA assumptions
independence, normal distribution of tx a, equal variances of tx a, x & y relate but don’t interact, w/in group reg coeff are equal, cov is w/o error & has no effect on IV
Regression & correlation assumptions
x has linear relationship with y, all important x’s in model, all unnecessary x’s (unrelated to y) not in model, data pair independence (x1 does not relate to x2)
specific regression assumptions
homoskedasity of residuals (check scatter plot & residual plots), residuals are normal (mean = 0), model is properly specified with above assumptions (R2 and eliminated no sig x’s)
correlation assumptions
bivariate normality of x and y, even random dispersion of residuals
CR
one between, one indepent factor. no single person in multiple groups; one level of observation
RM
one within, subjects have commonality across all levels & recieve all tx’s. one group with p levels of obvs.
CRF-pq
2 groups with one level, participants are not repeated in groups. looking at interaction. Main effect a is looking at significance of group 1 and then of group 2. SME looking at effect of gender specific to paper or electronic; looking at format specific to male & female
SPF
1 btwn 1 within. people take one group but complete all levels within group. take paper or electronic with all subjects (math, science, and english)
ANCOVA
look at effects of IV on DV when controlling for a continuous factor
are april ACT scores sig different for males or females?
CR, one between, one way
are dec and april act scores different across individuals?
RM, one within
is there an interaction between gender and course taken on April ACT scores?
CRF-pq (2-3)
are april ACT scores sig different for males and females in different courses?
CRF-pq (2-3)
does the tennis ball bounce differently depending upon court and swing?
crf-pq 22
are dec and april act scores sig dif for students in different courses?
spf
are april ACT scores different for students in different courses when taking into account practice ACT?
ANCOVA
