null hypothesis significance testing (NHST)

null hypothesis significance testing (NHST)

logical framework used to make decisions with respect to the likelihood of obtaining a particular sample statistic value from a known or assumed population

- whether it is probable or improbable, or if it fits with a hypothetical alternative population

null and alternative hypotheses

mathematical statements that reference some population parameter (often the population mean); directional or non-directional

- null describes a distribution located in a particular place or within a particular range

- alternative describes a distribution located somewhere other than null distribution

reject the null hypothesis

too improbable that the sample mean observed would be obtained from the stated population

- sample mean is not "in keeping" with the scenario

fail to reject the null hypothesis

it is probable, or not improbable enough, that the sample mean observed would be obtained from the stated population

- sample mean is "in keeping" with the scenario

assumption

a priori criteria of the data or feature of the test that must be true, otherwise the result is misleading to some degree

observed test statistic

sample statistic that we use to make inference about the population (e.g., Zobs, Tobs, Fobs)

if-then link

if the null hypothesis is true, and the test assumptions are true, then our test statistic will have a particular theoretical distribution, which is the null sampling distribution

alpha

amount of probability we allot to extremity

- the larger it is, the easier to reject the null (& vice-versa)

critical values

values (e.g., z-scores, t-scores) that correspond to the slices of the null distribution defined by alpha

- threshold for extremity (rejecting H0)

p-value (Pobs)

probability of observing a test statistic as or more extreme than the one obtained in your sample, given the null hypothesis is true