Stats vocabulary Flashcards
50% data below 50% above
median
treatments, can have control, can show cause and effect
data collection
experimental study
a collection of stuff
data
look right now, ask
observation
cross section
Subject does not know treatment to prevent bias (placebo)
Improve experiment
blinding
repeat over and over
Improve experiment
Replicate
fake treatment, need to fail to show real treatment works
Improve experiment
Placebo
mode is best if?
a outcome is very common
average distance from mean
data
standard deviation
relationship between the problems
correlation
in order spacing (no true zero)
level of measure
interval
to the left low value
shape of data
left skewed
order but no spacing
Level of measure
ordinal
look in future, wait and see
observation
prospective
less than 30
small sample
data collector cares about outcome
intro bias
bad source
what to expect
measures of center
wrong # of people
intro bias
under coverage
really far away from the rest of the data
outlier
how many/ how much
quantitive
high value data
shape of value
right skewed
no rational reason behind
Level of measure
imposed
many ask but few respond
intro bias
non-response
choose every nth
non srs
systematic
all of something
population
standard dev squared- s^2, σ^2
variance
break into groups and choose all of some
non srs
cluster
average of the data
mean
split into smaller groups and add some control, smaller experiments, some bias
Improve experiment
block
ask but do not measure
reported results
a group is left out (using tcu students as rep of Texas)
intro bias
missing data
lots of things
confounding
just looking, always having lurking variables, always have confounding things (lots of things)
data collection
observational study
asked wrong group
intro bias
bad frame
no space (4, 4.1, 4.2)
continuos
most common outcome
mode
no true order
Level of measure
nominal
break into groups and choose some of all
non srs
stratified
look in the past, look it up
observation
retrospective
25 % above, 75 % below, or 75% above 25% below
data
quartiles
numbers tell us what it is
qualitive
a lot with values in the middle
shape of data
uniform
Q3-Q1
IQR
some
sample
even uncontrolled variables
Improve experiment
randomize
subject and data recorder do not know treatments
Improve experiment
Double blinding
do you want a high paying job or a fun job? (likely choose first) Do you wanna be punched or kicked? (likely choose 2nd)
order of questions
easy to get
non-srs
convenience
mean is best if?
there are no outliers
have breaks (1, 2, 3)
discrete
lurking variables made the same to avoid confounding
Improve experiment
Control
Maximum- minimum
data
Range
median is best if?
there are outliers
data is given without being asked
intro bias
voluntary response
ask but only one right answer
loaded question
big deal
practically
in order with spacing
level of measure
ratio
every individual is likely to be selected
Simple random sample (SRS)
same on the left as is on the right
shape of data
symmetric
rare
statistically
