A) Parametric statistics Flashcards
independent groups
sample split into two groups. each group does one of the conditions. aka between-subject design
matched pairs
same as independent groups however, each ppt is matched on important characteristics with someone in the other group
repeated measures
the same group of ppt takes part in both conditions
what statistical test for independent groups (parametric and non parametric)
parametric: independent samples t-test
non parametric Mann-Whitney
what statistical test for matched pairs (parametric and non parametric)
parametric: paired samples t test
non parametric: Wilcoxon
what statistical test for repeated measures (parametric and non parametric)
parametric: paired samples t test
non parametric: wilcoxon
Difference between parametric tests and non parametric tests
- parametric tests eg t tests are calculated from the data (using mean and standard error)
- non parametric tests eg Mann Whitney are computed from ranked scores, not using the actual data helps to prevent outlier scores impacting the analysis
what do parametric statistics assume
assume the data you have collected come from a population that can be modelled on a normal distribution
what do non-parametric statistics assume
are sometimes referred to as ‘distribution-free’ because they do not make that assumption
what are parametric tests preferred
they are more sensitive/powerful
if there is a true difference between conditions, a parametric test is more likely to find that difference
what assumptions need to be met to use a parametric statistics (3)
- data needs to be at least interval level (continuous)
- assume the data collected comes from a population that can be modelled on ‘normal distribution’
- if comparing two groups, it assumed that they have similar variance
what data is appropriate for t-test (4)
- nominal- info is put into categories or named
-ordinal - info or scores are put in order or ranked
-interval- there are equal intervals on a measurement scale
-ratio- same as interval but there is a true zero point
what is a normal distribution
bell shaped curve
symmetrical distribution around the centre of all scores
skew distribution explanation
more developed on one side or in one direction than another, not symmetrical
kurtosis distribution explanation
the sharpness of the peak of a frequency-distribution curve - pointless/heaviness of tails
negative skewness values
pile up of scores of the right, tail to the left
positive skewness values
pile up of scores on the left, tail to the right
two tests of assessing normality
- numerical methods (eg skew/kurtosis values and statistical tests)
- graphical methods (eg visual inspection of graphs)
outliers meaning
they impact mean and standard deviation. mean and strd deviation are used to calculate t-test. therefore the presents of outliers biases both descriptives and inferential stats.
homogeneous variance
both groups have similar variance
heterogeneous variance
the groups have different variance
para vs non-parametric (3)
- parametric analyses can analyse non normal distributions for many datasets
- nonparametric analyses have other firm assumptions that can be harder to meet
- the answer is often contingent upon whether the mean or median is a better measure of central tendency for the distribution of your data
