Research tests

Question 1

nonparametric vs parametric tests

Answer 1

non-used for samples that do not share distributions or are not normally distributed

para-more powerful. They also are most dangerous. Equal distribution. More focused.

Question 2

Nominal

Answer 2

Frequency

ex. sex, political views

Question 3

Ordinal

Answer 3

Ranks

ex. order/organization, level of pain

Question 4

Scale/Continuous

Answer 4

Interval/Ratio

ex. level of pain can be continuous too. Age. ROM.

Question 5

Independent t-test

Answer 5

used when there are two conditions (one factor with two levels) and different subjects have been used in each condition (between subjects factor)

Independent means

Indep samples are not related in any way.

ex. comparing males to females

Question 6

Dependent (paired) t test

Answer 6

Correlated means

Samples NOT indep are for example, repeated samples or paired samples

ex. comparing pre and post scores

Question 7

Analysis of variance (ANOVA)

Answer 7

Tests for differences in two or more means.

Compares variances in specific ratio (signal=between group variance, noise=within group variance)

Generates an F statistic

F=between variation (mean square)/ within variation (mean square) –Larger this ratio, more likely there is at least one group difference

Multiple means-don’t want to do multiple t tests. Compounds the probability of type I error ( 20 tests then at least 1 will be falsely significant)

Question 8

ANOVA assumptions

Answer 8

indep data

random sample

normal distribution

homogeneity of variance

ALSO important: equal size in groups (balanced)

Question 9

Post Hoc testing

Answer 9

ANOVA tells me at least two of groups are different-need to test which groups differ from each other

Question 10

Confidence interval

Answer 10

wider confidence interval= less precise

Narrower confidence interval= more precise