introduction to quantitative data anlaysis Flashcards
- IQR
datapoints split up into 25% chunks. IQR is the upper quartile (UQ) minus lower quartile (LQ)
how do we get the SD?
amount points differ form the mean. Square each measured deviation and then calculate the average. This gives us the variance and its squared toot is the standard deviation.
why is SD better for interpretation than variance?
tells us the amount of variation in align with the scale of the variable e.g., if measured in cm’s an SD of 2.5. is 2.5 cm.
Variance not aligned with the scale of the variable so not directly interpretable.
if a variable is normally distributed, where does its estimate lie?
CI gives us a window of potential values. 95% of the units lie 2 SD on either side of the mean
o Hetro vs homoscedasticity.
If the variability between age and gender on weight is the same for both groups, then the variability is homoscedastic. If there was massive variability in one and not the other, then variability is hetroscedastic.
what do we infer with inferential statistics?
Infer the population parameter
Sample variability?
fi we do some tests on a sample of the population, the results will be slightly different to those of a different sample of the same population.
The difference between these two = sampling error
sample statistic?
any summary measure calculated by the data. Includes means, SD and regression coefficients
when would a sample statistic turn into an estimate
the second its used to make an inference about the population
how can we measure the sample variability
Standard error. Measures how much the observed values differ from the estimate
what is precision?
The inverse of the standard error. If SE is high then precision is low.
what two things affect the standard error
what is a confidence interval and what does it tell us?
indicator of the vairability
in responses that we might see (sampling variabilty).
To give an estimate while accounting for this variation we use a confidence interval.
gives plausible range of values that our true (but unknown) treatment effect could take.
tells us how precise a trial has estimated a treatment effect.
what does the 95 represent in the 95% confidence interval
just means if we repeated the study 100 times, in 95% of the time the CI would contain the true population effect
however. a single CI might not contain this!! 5% of the time the CI will miss the true population effect
when interpreting the trial results what 2 things do we bare in mind?
need to fully consider the clinical relevance and statistical significance of the estimate .
