lecture 4 Flashcards
characteristics of hypothesis
states relation between 2
testable
can be refuted
null and alternate
null hypothesis
no difference between varaibles of interest
statistical
alternate hypothesis
directional or non directional
principle of hypothesis testing
calculate the probability of results
if we test null can we reject this meaning can we see a difference
alpha value
critical value of a statistical test
that the chance something is unlikely due to chance
e.g 0.05- 5 percent chance the differnece is chance and not iv
type 1 error
reject null hypothesis
get a significant finding when nothing really happened
protect by having smaller p value
type 2 error
fail to reject the null hypothesis
get a non signifcant finding when there really is a significant effect
power
sensitivity of a test to find a significant effect
steps in sampling
define population by specifying inclusion and exculsion criteria
determine sample size needed
implement sampling procedures
compare the critical values of the sample
inclusion and exclusion criteria
ensure patient safety during study
only appropriate subjects and included
maintain ethical
help control confounding variable
participant selection
must possess all characteristics idenitified
inclusion
must not possess any charactersitics in exclusion
sampling types
probability or non-probability
probability sampling
simple random
stratified random
systemic random
cluster sampling
multi-stage sample
non-probability
purposive sampling
convenient sample
qouta sample
snowball
choose on some basis
advantages and weakness of non probability
adv- low cost, easier to recruit, easier planning
weakness- difficult to generalise to pop/ potential biasis
normal distribution of population
Determines the proportion of the values that fall
within a specified number of standard deviations from
the mean.
z score
no. of SD’s from the population mean a
data point is
non-normal distribution
if towards the left positively skewed
if towards right negatively skewed
parametric test
Assume parameters of population are set
* i.e., tests that assume the underlying source
population(s) to be normally distributed
* Measurement must be interval or ratio
non-parametric test
No set parameters around population
* i.e., population is not normally distributed, skewed
how to analyse descriptive stats
summarise and describe
inferential statistics
generalise/ make inferences about the larger set
POPULATION VS SAMPLE mean and SD
we can estimate the degree to which sample mean varies from population mean by standard error of mean
SEM=
standard deviation / square root of number in sample
confidence interval 95 percent
mean + or - (1.96 x SEM )
range of the population
mean + or - 1.96 ( 1SD)
95 percent confidence= mean difference + or -
3x SE average
