Ch 9 Significance Tests Flashcards
Null hypothesis
-definition? -called? -symbol?
claim being tested
-sometimes called “status quo” hypothesis
represented by Ho
Alternative hypothesis
-definition? -symbol?
other claim
represented by Ha
define “rejecting the null hypothesis” or “evidence for the alternative hypothesis”
if you get an outcome that’s very unlikely to happen if Ho is true, then this is good evidence that the null is NOT TRUE
step procedure for a significance test?
step -> hypothesis, significance level a, define parameter
plan-> same as ch. 8
do-> test statistic
conclude-> because..reject/fail to reject null, context
what is the DO step really assessing?
assess how far the estimate is from the parameter, standardize the estimate
estimate-hypothesized value /
test statistic= standard deviation of estimate
p value
probability computed assuming Ho is true, that the observed outcome would take as extreme or more extreme than that actually observed
(probability that measures the strength of evidence against Ho in favor of Ha) (probability that results happen given null is true)
small p values are evidence against Ho because?
because they say that the observed result is UNLIKELY TO OCCUR when Ho is true
(large p values fail to give evidence against Ho)
statistically significant
if p value is smaller than alpha -> reject null & say there’s convincing evidence in favor of alternative.
“significant” means
means not likely to happen just by chance
rare, unusual, low probability
under ____% is rare
5%
alpha is usually what value?
.05 if unstated
type 1 error
if we reject Ho when Ho is true
type 2 error
if we fail to reject Ho when Ha is true
significance level a is…
the probability of a type 1 error
conditions for plan part, dealing w/ a population proportion
random
normal- npo & n(1-po)
independence- n * 10
Do: what test is used from a population proportion?
1-prop z test
name p-value & z
chance of type 1 error?
a
chance of type 2 error?
b
power
1-b
3 ways to increase power
increase sample size
increase significance level (a) or decrease b
make Ha further away from Ho
a
probability of type 1 error (rejecting Ho when you shouldnt bc Ho is true)
b
probability of type 2 error (failing to reject Ho when you should because Ha is false)
1-b
power, probability of rejecting Ho when you should bc Ho is true
-higher power -> more sensitive to the test, want good chance of correctly rejecting Ho
a, b, 1-b, relationship?
a increase, b decreases -> inversely related
b decrease, power increase-> inversely related
a increase, power increase -> directly related (+ up to 1)
knowing z score vs t score
z score- normcdf
t score- tcdf
what is a significance test doing?
making a guess about a population w/ a sample
do: what test for population mean?
t-test
name t, df, & p-value
paired data
study designs that involve making 2 observations on the same individual, or 1 observation on each of 2 similar individuals yields this
when do you use 1 sample t procedures?
after paired data, use to perform inference about the mean difference mD
random selection
random selection from a population allows us to make an inference about the population
random assignment
random assignment of subjects to treatments allows us to make cause & effect conclusion
State step for paired data, what is the null usually?
null is usually set equal to 0
