state the hypothesis
- assuming sth to be true
two sided - when we have a equal sign
one sided - when we have >
statistic test
- we identify test a hypothesis with many types of test (ex. t test)
specify the level of significance
- depends of seriousness of making a misteake
type I error (alfa) = if we reject the H0 and it happend to be positive (flase positive)
type II error (beta)= if we fail to reject the Ho and it happend to be false (false negative)
as alfa increases beta deacreses. decrease both is to be done while decrease n
state the decision h0
- we reject if:
2 tail - |test stat|>|crit value|
1 tail
right = test stat > crit value
left = test stat < crit value
t-stat
when alfa
xśr-H0/(sd/n^1/2)
or
when 1-alfa
xśr=+/- crit value*(sd/n^1/2)
test statistic
odpowiedni wzór w zalezności co chcemy policzyć (t)
critical value
excel function:
T.INV(x:n-1)
x depends what tail we have but its alfa or 1-alfa
right tail (5%)
95% of the area is on the left
left tail
95% is on the right
which test
depend on a data
how to do it
- state the hypothesis (h0;ha)
- identify apropriate test
- specify the level of significance (alfa; beta)
- state the decision rule (critical values)
- calculate test stat
- make the decision (reject or not)
parametric testing
for mean and st dev;
distributional assumptions
non parametric testing
- with not distributional assumtpions
- when outliers
h0 (nuul)
- we hope to reject the null
- null is always equal
wartość tak jakby nie było testu.
ha (alternative)
- we hope to accept the alternative
- alternative is always not equal
wartość którą testujemy
level of significance
alfa
level of confidence
1-alfa
magnitude
