types of tests Flashcards
Parametric testing
when data is normally distributed.
non-parametric testing
when data is not normally distributed.
Nominal scale
simplest. lowest level of measure. classification.
male/female… blood types… etc.
Ordinal Scales
rank ordering. intervales may not be constant.
what is pain on scale 1-5… course evals…MMT…etc.
interval scales
relative comparison
rank order of ordinal scale with equal distances between values.
(i.e., Temp in C or F)
Ratio scales
absolute comparison. interval scale with an absolute zero. Permits all statistical analysis. no negative values.
(i.e., ROM, height, weight, force.. etc.)
unpaired t-test
a parametric test. comparing 2 independent groups.
paired t-test
a parametric test. comparing 2 related scores
One way ANOVA (F test)
a parametric test. comparing 3 or more independent groups.
finding a difference.
One-way repeated ANOVA (F test)
a parametric test. comparing 3 or more related scores.
Mann-Whitney U test
a non-parametric (unpaired t-test equivalence) test. comparing 2 independent groups.
Sign test.
Wilcoxon signed-ranks test (T)
a non-parametric (paired t-test equivalence) test. comparing 2 related scores.
Kruskal-Wallis analysis of variance by ranks (H or X^2)
a non-parametric (one-way ANOVA equal) test. comparing 3 or more independent groups.
Friedman two-way analysis of variance by ranks (X^2 r)
a non-parametric (1-way repeated ANOVA equal) test. comparing 3 or more related scores.
2-way ANOVA
has 2 Independent variables.
Multivariate ANOVA (MANOVA)
has multiple independent variables and multiple dependent variables.
Post-hoc tests
A study that has already determined there is a significant difference between groups and now wants to determine which groups are different.
Pearson Product-Moment
is for interval or ratio scale correlation stats.
finding a relationship. Is x related to y.
Spearman Rank
RANK in name.. rank is ordinal!
for ordinal scale correlation stats. Example pain scales or MMT
Finding a relationship. is x related to y.
Chi square
finds association. If blank is associated with blank.
Goodness of fit (chi square)
examines if observed freq is different from expected.
if stat > crit then reject null
Test of independence (chi square)
examines association (or lack of) between 2 categorical variables (ordinal or nominal).
df= (R-1)(C-1)
*rows and columns
