making decisions Flashcards
z score
measure of distance between the sample statistic and population parameter
how to decide exact probability that sub group and pop have same mean
use normal tables
what z score puts the sample in the rarest 5%
equal or more than 1.96
what z score puts the sample in the rarest 1%
equal to or exceeds 2.58
alpha level
risk we are willing to take that we reject null hypothesis when its really true
alpha level of 5% =
a score 1.96 (5% risk we are wrong)
alpha score og 1% =
z score 2.58 (1% risk we are wrong)
why not increase level of significance
would increase number of times we accept null when we should reject it
type 1 (alpha) error
reject null but should be accepting it
type 2 (beta) error
fail to reject null but its not actually true
z test
comparing z scores to a critical value
2 tailed test / non directional
reject null if sample statistic reaches critical value in either tail
1 tailed test / directional test
when association can only be one direction we allocate 5% risk of error to one side of distribution
when are parametric assumptions used
only for ratio / interval data
what to do when the population size is less than 30 or don’t have the pop SD
t test or sign test
which test do you use if data are normal
t test (parametric)
which test do you use if data are skewed
sign test (nonparametric)
how to know if distribution is normal
look at histogram
check SPSS output
kurtosis
how peaked the distribution is
platykurtic
flat distribution
leptokurtic
very peaked distribution
why is a 2 tailed test always used
as we use Alpha level 5%
what’s the t test based on
sample and pop mean and sample SD
what is sign test based on
median