Please enable JavaScript.
Coggle requires JavaScript to display documents.
Inferential statistics - Coggle Diagram
Inferential statistics
-
Confidence intervals= gives lower and upper limits bracketing the sample statistic, this is done for a specified confidence level. Absolute certainty is not informative
Can be plotted on a graph using a point symbol, shown by a capped interval
It is a probability so that people using the same methods would get the same results, NOT a measure of anyone's beliefs or impressions
calculating= probability of getting results as extreme (larger/smaller) as those observed, called P values.
-
-
population: may exist tangibly, less tangible (measurements repeated)
-
-
Sampling distribution= we can imagine having other samples of the same size, giving us many means and correlations. use mathematical deduction from a model or computer simulation. use a normal distribution
The normal distribution= symmetric bell-shaped histogram. centred on the mean, median and mode. +1SD are the inflection points where curve changes from convex to concave. also known as Gaussian distribution
The normal is unbounded and the limits and infinite, 95% of values lie within 1.96SD of the mean. works on human heights. SAMPLE MEANS CAN BE ROUGHLY NORMALLY DISTRIBUTED EVEN IF THE ORIGINAL DATA IS NOT.
-
Null hypothesis= statement of no difference, no effect. Reject it if sample evidence is overwhelmingly against it, e.g. legal analogy
-
Scientific or practical interpretation= bigger sample for firmer results,
Pitfalls= 95% confidence level is 5% significance level, data snooping (reporting on significance tests with the strongest results)