inferential statistics Flashcards
what are inferential statistics?
the use of mathematics probability to tell us whether we can accept or reject the null hypothesis
when we accept the null hypothesis…
are results are not statistically significant, the results could simply be due to chance alone (fluke result)
when we reject the null hypothesis…
our results are statistically significant and unlikely to be due to chance alone (not a fluke, due to something we have done e.g manipulating the IV)
how is a statistically significant results expressed?
in terms of probability (p)
p < 0.05
means …
the probability of the results of an experiment being due to chance alone is less than 5%
if p < 0.05, we can
reject the null hypothesis, as probability of it being down to chance is less than 5%
p > 0.05
means…
probability of the results of an experiment being due to chance alone is more than 5%
if p > 0.05, we can
accept the null hypothesis, as probability of it being down to chance is more than 5%
if it is significant we write
p < 0.05
if it is not significant, we write
p> 0.05
what is a type 1 error?
a false positive
what happens on a type 1 (false positive) error?
we would reject the null hypothesis and wrongly accept the alternative hypothesis when we should not have done so
when do you run the risk of a type 1 error?
if we were less strict and accepted the significance at the p < 0.1 level (90/10)
what is a type 2 error?
a false negative
what happens at a type 2 (false negative) error?
we would wrongly accept the null hypothesis, and reject the alternative hypothesis when we should not have done so
when does type 2 errors occur?
if we were stricter and insisted on p < 0.01 to demonstrate significance (99/1)
how many inferential statistical tests?
8
criteria for using inferential tests
- are we looking for a difference or a correlation?
- what level of data are we using (nominal, ordinal or interval) ?
- what experimental design are we using ( IMD, RMD, correlation) ?
mnemonic for inferential tests
carrots - c - chi squared
should - s - sign test
come - c - chi squared
mashed - m - mann whitney U test
with - w - wilcoxon T Test
swede - s - spearman’s rho
under - u - unrelated T Test
roast - r - related T Test
potatoes - p - pearson’s r
draw the decision table for inferential test
what goes across the top for the decision table for inferential tests
level of measurement/design
IMD RMD correlation
what goes down the column for the decision table for inferential tests
level of measurement/design
nominal
ordinal
interval
what are the 3 parametric tests?
interval level
- unrelated T Test
- related T Test
- pearson r’s
criteria for using a parametric test?
- the standard deviation is not significantly different
- data is normally distributed
- data needs to be interval or ratio
what are the non-parametric tests?
- chi squared
- binomial sign test
- mann-whitney U test
- wilcoxon T Test
- spearman’s rho
