1Introduction to Hypothesis TestingI. What is Hypothesis Testing?- Is the pattern of data in the sample likely to be found in the population?- Or, is this pattern likely to be Sampling Error?- If 95% sure (5% doubt) the pattern is not just Sampling Error, then pattern is Statistically Significant.II. What is Sampling Error- Every sample randomly drawn from a Population is different from the population…- Differences can be small or large2- Sample Size influences the likelihood of differences- Small Samples: Large differences are more likely- Large Samples: Large differences are less likely, though possible.III. Hypothesis Testing is Decision Making- Decide if pattern of sample results is different from a Chance Pattern (Sampling Error). - State Decision in terms of the Null & Research HypothesisNull Hypothesis – No Significant Pattern(Ho): Sample Pattern = Chance PatternAlternate Hypothesis – Significant Pattern Exists(Ha): Sample Pattern ≠ Chance Pattern- Decide: Reject Hoor Reject HaIV. Making the Decision1. Quantify the pattern of data in sample- Inferential Statistics – t, r, X2, F- Type of Stat depends on data (e.g. continuous vs. discrete)32. Quantify Sampling Error- The Critical Value – weakest pattern of data still considered significant.- Depends on- Size of the sample: n- How sure you need to be: α or p- 95% sure = 5% doubt (α or p = .05)- α or p = .05 means that 5 time out of hundred the pattern would occur by chance- .05 is the largest acceptable level in most sciences.- Every Inferential Statistic has its own Table of Critical Values3. Compare Sample Stat to Critical Value- If Sample Stat > Critical Value @ p = .05, Then Reject Ho – Pattern is Statistically Significant- If Sample Stat < Critical Value @ p = .05, Then Reject Ha – Pattern is Not Statistically SignificantNote: Decisions are based on likelihood- We don’t know if Hois true or false- Never Accept Hoor Ha- Only Reject or Fail to Reject- Hypotheses are never Proven – We only find statistical support for
View Full Document