quiz 4 Flashcards
what is categorical fashion?
- no variation
- almost all experiments fit here
what is a grouping variable?
separates data into samples
what is ordinal
- categories that have a logical ranking to it
- nonparametric
basic question
does your grouping variable categorize your data points into distinct groups?
experimental question
does your independent variable have an effect on your dependent?
statistical question
are your samples drawn from the same population or not?
how do we distinguish samples?
- compare measures of central tendencies
- how much overlap is there between the two samples?
what does a high variability graph look like and what does it mean?
they overlap closer to the top of the curve. this means it could almost be one sample and there may just be sample error.
what does a medium variability curve look like?
overlaps in the middle of the curve
what does a low variability curve look like?
overlaps at the bottom of the curve
what are the two types of tests that compare samples?
parametric and non- parametric
parametric
- will compare means of samples
- requires your data set to meet a set of assumptions
what is the null hypothesis for every assumptions test?
the null hypothesis is that your data will meet every assumption.
on a independent samples test, what does “F” mean?
testing assumption
on a independent samples test, what does “Sig.” mean?
p-value
on a independent samples test, what what must be interpreted first?
Levene’s test
non-parametrics test
- compares medians
- does not meet assumptions (null is rejected)
- uses different approach like ranking data points
what is wrong with the non-parametrics test?
- greatly increases the risk of a type 2 error
why is using individuals as their own control group preferred?
- reduces the needed sample size
- eliminates the influence of variation between subjects
matched control design
matching two people w/ similar demographic where one gets drug and one gets placebo
if data is repeated and meets assumptions, what test should be used?
paired samples T-test
if data is repeated and does not meet assumptions, what test should be used?
Wilcoxon Signed Rank Test
if data is independent and doesn’t meet assumptions, what test should be used?
Man Whitney U test
If data is independent and does meet assumptions, what test should be used?
Independent samples T-test
Glioblastoma
a tumor arising from glioblast cells
- tough cancer
- grows quickly
angiogenesis
recruits new blood vessels
- when small, they start metastasizing
angiogenic drugs
- important in preventing recurrence in patients that survive initial tumors
two independent samples
- has a categorical independent with only 2 states
- has a continuous or ordinal dependent variable
- NOT using repeated measures
is the independent samples t-test parametric or non-parametric?
parametric
- has more power
is the Man Whitney U Test parametric or non-parametric?
non-parametric
Independent samples T-test description
- has two parametric assumptions
- samples must be normally distributed (mean, median, and mode must be the same)
when is KS test used?
when data sets have above 100 samples
when is the shapiro wilks test used?
when sample size is below 100
what test is used to check for equal variance?
levenes test
z-transformation
take each individual data point and do calculation on it. This will be the new number.
Arcsin
may not understand as well
what do you do if you fail assumption on normality only?
Use correction or use non-parametric test
what do you do if you fail assumption on equal variance only?
- use non-parametric test
what do you do if you fail assumption on normality & equal variance?
non-parametric test
HA 1 tailed
the mean of one sample A is larger than sample B
- which one we think is bigger (has direction)
HA 2 tailed
the means of the two samples are different
- “it has an affect” without direction (+/-)
in independent Samples Test, what does “t” mean?
how different 2 samples are (we dont care about sign)
in independent Samples Test, what does “df” mean?
it interprets the test statistic
what does statistically significant mean?
likely to be real
equation for Cohens D
D=2t/sqr root of df
effect size correlation equation
r= sqrt(t^2/t^2 + df)
Mann Whitney U Test description
- non-parametric
- works by ranking data in order
- cant calculate mean
inattention meaning
cannot focus on readily tasks
hyperactivity
higher than normal motor activity
impulsivity
low self inhibition
- seeks immediate rewards (gambling)
ADHD
- most often diagnosed in children
- strongly heritable
- often treated with stimulants, high rate self treatment by adults with caffeine
Paired samples Test
- has a categorical independent variable with only two states
- has a continuous or ordinal dependent variable
- uses repeated measures
repeated measures
- data points in one sample are not affected by points in the other sample
- most common cause of non-independence is repeated measures of the same individual
2 important benefits of repeated measures:
- each individual serves as their own control as comparisons are made within individuals
- gives more confidence
Wilcoxon Signed Ranked tests
- ranks the absolute values of the differences
