Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
67 Cards in this Set
- Front
- Back
Association
|
when some values of one variable tend to occur more often with some values of the second variable than with other variable of the same (second) variable
|
|
Confidence statement
|
statistic plus or minus the MOE
|
|
Confounding
|
when the effects of two variables on the response cannot be distinguished from one another.. often caused by lurking variables
|
|
Correlation
|
measure of straight line association (think scatter plots); numerical value that determines that direction and strength of liner association
|
|
Correlation coefficient
|
common numerical measure of a straight line association; always between -1 and 1; identified as "r"
|
|
Explanatory variable
|
a variable that might cause changes in the response variable
|
|
Ha
|
alternative hypothesis
|
|
Ho
|
null hypothesis
|
|
inference
|
act of drawing conclusions based on information we have that we assume to be accurate
|
|
Lurking variable
|
variable not directly studied that can compromise the ability to attribute any changes in response to a treatment
|
|
Margin of error (MOE)
|
numerical way of acknowledging that you know the sample statistic is not going to give perfect knowledge about the population parameter; attached to some notion of confidence interval
|
|
negative association
|
scatterplot downwards to the right
|
|
Placebo
|
dummie pull/treatment
|
|
Placebo effect
|
tendency for patients to show a real response to a fake treatment
|
|
positive association
|
a scatterplot that is upwards to the right
|
|
randomization
|
action that produces groups that are similar in all aspects; action that eliminates potential biases
|
|
response variable
|
variable that measures the outcome of a study
|
|
sampling distribution
|
plot of repeated sampling of a statistic
|
|
scatterplot
|
x-y plot taken on different subjects
|
|
sensitivity
|
ability of the test to identity positive outcomes as positive outcomes correctly
|
|
simple random sample (SRS)
|
sample of "n" individuals chosen from the population in such a way that every set of "n" individuals had the same chance of being chosen
|
|
specificity
|
ability of the test to correctly identify negative outcomes as negative
|
|
statistical significance
|
when differences in treatments are sufficiently large that they are unlikely to be due only to chance
|
|
Simpson's Paradox
|
when trends are one way for certain groups, but when the groups are combined, the trend reverses
|
|
2 sources of confounding
|
1) inadequate or improper comparison
2) lack of randomization |
|
US population
|
about 300 million
|
|
babies born in US each year
|
4 million
|
|
Americans that die each year
|
2.4 million
|
|
1 in 4 of who die,
|
die of heart disease
|
|
1 in 4 of who die,
|
die from cancer
|
|
deaths in traffic accidents
|
43,000
|
|
deaths that are homicides
|
17,000
|
|
deaths from AIDS
|
16,000
|
|
black Americans
|
40 million
|
|
Americans that identify as Latino
|
14%
|
|
reasoning process
|
process of reasoning from premise to a conclusion
|
|
examples of poor inference
|
decimal point error
poor graphs implausible number percentages or absolutes (both can be deceptive) incorrect comparisons |
|
experiments vs. surveys
|
experiments: make a concerted effort to control conditions under which data is collected
|
|
correlation doesn't mean...
|
causation
|
|
causation
|
a relationship between two variables in which a change in the level of one causes a change in the other
|
|
in regards to "r" the correlation coefficient, the closer to 1/-1 the. . .
|
better or stronger the correlation
|
|
inference can be defined as a conclusion drawn from evidence?
|
true
|
|
goal of sampling
|
to make inferences about a population from what we know about sample data
|
|
qualities of Push Polls
|
goal: to influence opinions
often negative often short sample size is very large biased |
|
population
|
large collection of subjects/items that you are interested in understanding something about
|
|
sample
|
subjects/items that you are able to measure.interviews; chosen from population
|
|
parameter
|
number that describes population
|
|
statistic
|
number that describes the sample
|
|
sampling variability
|
variability seen in a statistic from sample to sample; same as sampling distribution
|
|
shape of sampling distribution
|
bell shapes, bell curve
peaks at parameter |
|
confidence interval
|
helps to quantify the sampling error
|
|
types of non-sampling errors
|
data entry
nonresponse biased questions question order false information voluntary responses |
|
MOE doesn't apply when...
|
when there are non sampling errors
|
|
in hypothesis testing we want high ________ and low ______
|
high sensitivity
low FPR |
|
false positive rate
|
when the patient is really negative, but the test comes back positive
|
|
false negative rate
|
when the patient is really positive, but the test comes back negative
|
|
1-sensitivity
|
FNR
|
|
1-specificity
|
FPR
|
|
in hypothesis testing treatments are either. .
|
effective (Ha) p > p0
ineffective (Ho) p< p0 |
|
steps for when you see "statistically significant" or "not statistically significant"
|
1 establish comparison
2 express comparison in forms of Ha and Ho 3 determine who comparison turned out 4 articulate risk involved with decision/result |
|
practical significance
|
whether the observed difference is big enough that it is practically worth caring about
|
|
when accepting Ho, FPR is
|
greater than 0.05
|
|
when rejecting Ho, FPR is
|
less than 0.05
|
|
p-value is another word for...
|
FPR
|
|
you have _________ _______ if FPR is less than 0.05
|
statistical significance
|
|
in the context of testing Ho versus Ha, what did it mean to have a "false positive"
|
you deiced Ha was true, when in fact Ho was true
|
|
what do you hope to see when working with specificity and sensitivity?
|
high sensitivity and high specificity
high sensitivity and low FPR BOTH ARE SAME THING |