Significance Tests Flashcards
(15 cards)
Ho
claim that we want to disprove. Default to this unless we can disprove it.
Ho = x
“any difference is due to random chance”
Ha
competing claim
Ha > x Ha < x Ha =/= x
“difference is not random chance”
p-value
probability that outcome would occur if Ho is true/probability that difference is due to random chance.
lower p-value = reject Ho
p < alpha
reject Ho
Type 1 error
reject correct null hypothesis
Type 2 error
fail to reject incorrect null hypothesistep
steps of sig test
- state paramter, Ha, Ho, alpha,
- checks
- calculate test stats and p-value
- interpret
test statistic
z score of sample prop if Ho is true
z = statistic - parameter/st. dev of statistic
statistic= sample parameter=population
finding p-value
normcdf(LB=z,UB=9999, mew=0, st.dev=1)
tcdf(LB=t,UB=9999, mew=0, st.dev=1)
probability of type 1 error
significance level/alpha (0.05)
power
probability of not making a type 2 error
decreasing chance of type 2
- increase alpha
- bigger sample size
- make Ha and Ho distinct
1-Prop-ZTest
find p and test statistic for proportion
1-Mean-TTest
find p and test statistic for mean
2-Prop-ZInt
comparing two proportions