POPH192 - Lecture 23 Flashcards
what types of validity will a good study have?
both internal and external validity
what is external validity
- do the findings apply to other populations?
- can these results ne generalised to another population of interest?
- external validity depends on the population group of the study
what is internal validity?
is there a real association present in this study?
- are these results valid for the study population?
what does internal validity depend on?
chance, bias and confounding
- are an alternative explanaion for why we got the results we did
when is a study internally valid?
when the results of the study are not because of chance, confounding or bias
what is chance a component on?
internal validity
what is sampling?
- take a sample from a source population to do a study
- because it is impossible to do a study using the whole population
what does the study sample give?
- an ‘estimate’ value of the population study
- this is the unknown true value of the measure that the study is trying to estimate
what is sampling error?
sampling is unlikely to be perfect - due to chance
what happens each time you sample?
each sample you take from the same population would be slightly different - due to chance
what affect does chance have on the population?
because of chance, the study might not be an accurate representation of the population
what is chance?
a random sampling error
- most common reason for this is a small sample size
how do you reduce the likelihood of chance?
- increase sample size
what are the affects of increasing sample size?
1) reduces the sample variability
2) increases the likelihood of having a representative sample
3) increases the precision of the parameter estimate
how could you evaluate chance?
to decide if results are due to chance (or actually true), you can use confidence intervals and p-values
what is a 95% confidence interval?
we are 95% sure that our true population parameters lies between the bounds of the confidence interval
what is the interpretation of the confidence interval?
we are 95% sure that our true population parameter lies between upper bound** and **lower bound
what affects the precision of the 95% confidence interval?
-indicated by width of confidence interval
wide CI = smaller sample size = less precision
narrow CI = larger sample size = more precision
what is clinical importance?
it is a level set in RCT
- to decide if the intervention has a real, genuine and noticeable effect on daily life
what is considered ‘not clinically important’?
anything above the clincical importance threshold of 0.9 (reduction of 10%)
or
- the entire CI is above the level of clinical importance
what is considered ‘clinically important’?
- anything below 0.9 is clincally important
OR
- the entire confidence interval has to be below the level of clinical importance
what happens when the threshold is crossed?
it is considered ‘possibly clincally important’
- as we don’t know if the parameter is above or below threshold
is clincal importance the same as statistical significance?
no