This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: TwoSample Designs Lecture 9 Yesterday Recap Null hypothesis testing Type I (α) and Type II (β) error 1 vs. 2 tailed tests 1 sample t test t test of r vs. 0 (correlations) Today More on t tests Experimental design Independent samples t test Assumptions of the t test Effect size Confidence intervals Power Basic NHST Procedure 1. Identify the null and alternative hypotheses. 1. Determine what test you will perform (1 sample t test, t test of r vs. 0, two sample t test, ANOVA, etc.) 1. Determine your alpha level, whether your test in 1 tailed or two tailed, degrees of freedom, and find the critical t value in Table D. Draw a picture! 1. Conduct a statistical test of the hypothesis. 1. Based on your test, make a decision on whether you will reject or retain the null hypothesis? 1. Interpret your results in clear English (write out your conclusion in symbols and words). Determining alpha Often, this is listed in the problem – “conduct a test at the .05 level of significance” – “test at the .10 level” – “conduct a test at α =.01” If not listed, use .05 Determining alpha In your own studies: – If it’s important to not jump to conclusions unless there is really an effect (reduce Type 1 error), use a lower alpha level (.01, .001). – If you want to make sure and find an effect that exists (reduce Type 2 error), use a higher alpha level (.05, .10). Looking up critical t values in Table D What if my degrees of freedom are not listed? – Look at the numbers it falls between. If the degrees of freedom is closer to one or the other, use that as your critical value. If it’s about even, use the more conservative value (with the lower df) Example: df = 100 – Df = 60 and df = 120 are listed, so I can use t values for either. I would probably pick 120, since 100 is closer to 120, but either is acceptable. Interpreting t tests Use a picture to decide whether to reject or retain H . Using your decision (retain or reject H ), go back to what this means (based on the question of the study). Associated symbols: if reject H , t(df) = t obtained , p < α level if retain H , t(df) = t obtained , p > α level Beyond one sample If we want to compare our sample to a population value, we can use a 1 sample t test Often, we want to compare 2 samples to see if they are similar or different. In this case, use a 2 sample (or independent samples) t test Experimental Design Does hot weather increase aggression in...
View
Full Document
 Spring '10
 Ryne
 Statistics, Null hypothesis, Hypothesis testing, Statistical hypothesis testing, Statistical significance

Click to edit the document details