Final Paper and Presentation

Final Paper and Presentation

Hypothesis testing

lIn addition to descriptive statistics, most researchers compute inferential statistics

Mathematical tests to answer RQs

•Define hypotheses

•Choose appropriate statistical test

•Choose significance level and criterion for accepting/rejecting hypothesis

lbinary decision-making process

Hypotheses

lPredictions: expected relationships between variables

lH_o: no relationship between variables or groups

lH_A: expected result of experimental research

  This is what I ask of you in your project

  directional vs. non-directional

Briefly…

lCan you state the hypothesis / hypotheses for your proposed study?

•It is expected that…

•It is predicted that…

•Think: is your hypothesis directional or non-directional?

Hypothesis testing

lIn reality, error is always possible

Hypothesis testing

lWe try to minimize error

lSampling and power

sample size (n) determines the power of the statistical tests

•larger sample à more powerful test

•samples with n < 10 are subject to Type II error

power is the probability of rejecting H_o when it is false (a correct decision)

Hypothesis testing

lParametric statistical tests rely upon the normal distribution

model for evaluating data

frequency pattern of scores

•unimodal

•mean and median are centered

Hypothesis testing

Hypothesis testing

lCentral Limit Theorem (CLT)

•a distribution will be normal if the sample is large enough

  (n > 30)

•may not always hold for clinical populations, which are often

•highly variable

•multimodal

Statistical tests of significance

lParametric vs. non-parametric tests

Assumptions underlying parametric statistics

•Normal population distribution

•Homogenous population variances