L2 Flashcards
population
is the fun set of units to which the study results will be generalised
sample
the subset of the population that has been selected/ sampled to participate in the research study
data
are information collected from sample units e.g. blood pressure
the population is usually
infinite in size- all units that exit and ever will exist are relevant to study questions
uncertainty about the true answer is due to
- variability between people in which you are trying to measure
- the sample is only a subset of the pop. and is not perfectly representative of it
stats ar eisend to
summarise data in the sample and quantify the uncertainty in the sample results
descriptive stats methods
describe and summarise data samples e.g. how common are certain characteristics and how diff characteristics are associated with eachother
inferential stats methods
use sample data to make inferences about characteristic and relationships in the pop.
SE and CI
estimation
P values
hypothesis testing
Standard eror
how close is the estimate to the truth
- indicates how far on average the same estimate i expected to be from he true population parameter value, if you did the study many times with diff samples of the same size
what does SE indicated
If we did the study a large number of times with different samples of the same size then, on average, our sample estimates would be 0.6 units away from the true mean difference. It doesn’t mean that the estimate we have in our study is 0.6 units from the truth
SE quantities the
precision of the estimate
the smaller the SE
the closer the estimate to the true value in the population
larger the sample size the
smaller the SE
diff between SE and SD
SE summarises the precision of the estimate from a research study but standard deviation summarises the variability of observations within the sample for a quantitative variables.
hypothesis testing can only be used
to provide evidence about what the true answer is
why are confidence intervals more useful
since they tell you something about what the true answer in the population is
relationship between 95% CI’s and p-value
can be used to carry out a test of the null hypothesis at the 5% level of significance
if the 95% CI includes the null hypothesis then p value is
larger than 0.05
if the 95% CI excludes the null hypothesis then p value is
smaller than 0.05
if the lower limit of the 95% CI is the same as the null hypothesis value then
p= 0.05
CI represents the
true parameter value
we cannot draw any conclusion about the population based on just the mean diff between the intervention and control groups in the sample depressions core so…
we use inferential stats methods to tell us something about the truth in the population
if the mean depression score in intervention group is 1.3 lower than the control group, what would a 95% CI of 0.1 to 2.4 tell us
we are 95% certain that in the population of infant with sleep problems at 7 months the mean maternal depression score at 12 months is somewhere between 0.1 and 2.4 units lower for the intervention group than the control group
CI is the
range of value within which we can be certain with some degree of confidence that the true population parameter lies- 95% confident that true answer lies within this parameter
With Confidenc intervals it is better to
- have narrow CI
- better to have 95% CI than 90%
a larger sample size will provide a
narrower CI
CI are more
informative than SD
Hypothesis testing
using p values
-estimate is our estimate consistent with the null hypothesis we construct a statement about the ‘true’ parameter value (null hypothesis)
a p-value of 0.03 indicates
moderate evidence that there is a significant difference between the groups
the null hypothesis
the most boring truth imaginable
- not necessarily what you think the truth is
example of null hypothesis
mean systolic blood pressure in diabetics is the same as in the general, healthy population
the smaller the p value
the more evidence that the null hypothesis is false
traditionally 0.05 has been used as the threshold to reject or not reject the null hypothesis.
if the p<0.05 then reject null hypothesis- significant evidence strict adherence is not recommended (not much diff between 0.049 and 0.051
summary SE
how close might our sample estimate be to the true answer
summary CI
what IS the answer
hypothesis testing summary
what isn’t the answer