HYPOTHESIS TESTING, COMPARING MEANS Flashcards
data
pieces of info collected in study
variable
measurement which varies btw subjects, e.g. height
2 types of data
categorical and numerical values
categorical values
can be sorted into categories/groups, represented by bar charts and pie graphs
numerical values
can be observed/measured, numbers placed in ascending/descending order, represented by line graphs and scatter plots
types of catagorical data
nominal and ordinal
nominal values
assigned a ‘label’ in the form of numbers, e.g. sex
can be counted but not ordered
ordinal values
can be counted and ordered/ have a rating scale attached, but not measured, e.g. house numbers, swimming level
samples
subset of larger data, used to draw inferences about larger set (population)
parameters
characteristics of population data
average used for normally distributed nominal data
mean
MoS= standard deviation
average used for skewed nominal data
median
MoS= interquartile range
average used for ordinal categorical data
median
MoS= interquartile range
average used for nominal categorical data
mode
MoS= none
Alternative hypothesis (Ha)
what we aim to gather evidence of
that there is a difference/relationship etc
Null hypothesis (H0)
what we assume is true to begin with
there is no difference/relationship etc
when can null be rejected
only if there is enough evidence to doubt it
type 1 error
Incorrect rejection of null hypothesis
e.g. study shows there is a difference but the difference does not exist in population
type 2 error
Failure to reject null hypothesis, when alternative hypothesis is true
e.g. study shows no difference but there is a difference in the population
t-tests
used to compare 2 population means
paired data
same individual studied at 2 different times/under 2 different conditions, use paired t-test
independent data
data collected from 2 separate groups, use independent samples test
one way analysis of variance btw groups
when you want to test the difference between 2 groups
two way analysis of variance without replication
double-test the same group, e.g. test a group before/after they take medication
two way analysis of variance with replication
test two groups on more than one thing, e.g. 2 groups of patients trying 2 diff therapies
one way analysis of variance
compares 2 means from 2 independent groups
looks at variation within and between groups
null hypothesis= 2 means are equal
significant result= 2 means are unequal
limitations of one way ANOVA
will tell you groups are different but wont tell you what groups
two way analysis of variance
two independents
results calculate main effect & interaction effect
assumptions for two way analysis of variance
population must be close to normal distribution
samples must be independent
population variances must be equal
groups must have equal sample size
homogeneity of variance
each group should have similar standard deviation
