What are the steps to hypothesis testing?

1. State the research hypothesis
   1. Null hypothesis: H₀:μ₁ = μ₂
   2. Alternative hypothesis: H₁: μ₁ ≠ μ₂
2. Decide on the appropriate statistical test.
   1. Depending on the type of measurement (nominal, ordinal, or continuous), will determine the type of test (chi-squared, Mann-Whitney or Wilcoxon, t-test, ANOVA)
3. Select the level of significance for the statistical test.
   1. Type I error (false positive) - Incorrectly reject H₀ when indeed it is true.
      1. How to avoid: (1) conduct appropriate test, and (2) select p value.
   2. Type II error (false negative) - Failure to reject H₀ when indeed it is not true.
      1. How to avoid: increase power, by increasing sample size.
4. Perform the calculation
5. Draw and (re)state the conclusion.

3 Search Strategies for Pubmed

1. Single best search term
   1. Therapy (Randomized Controlled Trial) - Clinical trial [publication type]
   2. Diagnosis (Cross-Sectional Study) - Sensitivity [text word]
   3. Prognosis (Cohort Study) - Cohort Studies
   4. Etiology (Cohort Study or Cross-sectional) - Risk [text word]
2. Using "clinical queries" and search with "Clinical Study Categories"
3. Using limits and going to "Type of Article" and clicking on "Clinical Trial"

What is a Type I error and how to avoid?

1. Type I error (false positive) - Incorrectly reject H₀ when indeed it is true. (Positive because you concluded p < .05, when indeed no difference exists)
   
   1. How to avoid: (1) conduct appropriate test, and (2) select p value. (3) Conduct a Intetntion-to-treat analysis, which answers: who should I give therapy? (as opp. to efficacy of the treatment in PPA).

What is a Type II error and how to avoid?

1. Type II error (false negative) - Failure to reject H₀ when indeed it is not true. (Negative because you concluded p > .05, but indeed a difference exists).
   
   1. How to avoid: increase power, by increasing sample size.
      1. Increase power by conducting a per protocol analysis, which includes only patients who adhered to the protocol, e.g. the treatment. Excludes those who did not adhere, or died. Answers: how effective was the treatment? (once given, as opp. to who should I give in ITT).