Study Guides (248,539)
Psychology (776)
PSYC 3000 (24)
all (3)
Quiz

# Hypothesis Testing.docx

12 Pages
60 Views

School
Department
Psychology
Course
PSYC 3000
Professor
All Professors
Semester
Winter

Description
Hypothesis Testing 11/12/2012 5:47:00 AM With hypothesis testing: we can answer questions about our model - Null: = original value: no change - Alternative: the equation; if we reject the null. > < or does not equal (two tailed) - before doing tests: we check independence condition, random condition, success/failure condition - once we solve: we then ask how likely the data we observed would be were the null hypothesis a true model of the world - we look at probability: (p) - p high: fail to reject/accept null hypothesis - p low: reject; what we have observed is unlikely if our null was true The breakdown: 1) Hypotheses. One tailed/one sided alternative. Or Is it two tailed: increase/decrease/has it changed 2) Model; are the conditions satisfied (page 379. independence, random, success/failure: p(n) q(n) > 10 each) 3) Mechanics 4) Conclusion - sometimes we follow up with confidence interval: (page 383)  ^p +- ME  SE(^p) = Sqrt (p^q^/n) - we can not find the standard deviation from the null proportion (* if we rejected it, we can not use) - More about Tests. Chapter 21 11/12/2012 5:47:00 AM Example from class: 1) Descriptive 2) Analyze 3) Bring to DV 4)Statistics 5) Make 99 - made 99 percent confident interval; 99 percent confident between 17 and 23 - we can reject if do not see value in our interval - 0.136 not a part - Example two: - generate statistic: 0.38 - interval: 0.35 – 0.42 - 55 not in it therefore reject null - P value: conditional probability - probability of getting results unusual as seen given the null is true p(0.03): given the null there’s a 3% chance of observing the statistic value observed - big P values: (taken from z) results are inline with null assumption How Small is small? - we set a point: Alpha level: if p falls below; we reject - a: 0.1, 0.05, 0.01 - use alpha dependent on situation; safety of airbags; 0.01 - alpha level: significance level - if p > alpha level: fail to reject the null A 1 sided 2 sided 0.05 1.645 1.96 0.01 2.33 2.576 0.001 3.09 3.29 page 406: - compare the observed z score with the critical value. If z is less than z critical. Then we do not reject - you can also approximate hypothesis test by examining confidence interval - ask if null (Po) is contained in the interval confidence interval: two sided: correspond to two tailed - if falls outside; reject Errors in hypothesis testing: - null hypothesis is true; but we reject it (Type 1) - null hypotheses is false, but we accept (type II) - type I: make innocent person guilty type II: Make guilty person innocent - when you choose alpha level; your setting the probability of a type I error to a - - A tests ability to detect false null hypothesis: power of the test - Power: probability that test correctly rejects false null hypothesis (program -> course software -> spss -> spss sample power -> sample power -> 3) type 1 error: alpha type II error: B 1-B: power: probability that the test does reject Effect size: difference between null value and alternative value To increase the power: - increase the significance level - increase effect size - increase sample size Chapter 21: Notes From Pass and Textbook - Strong link between confidence interval and hypothesis testing; Same set of assumptions and conditions  independence  randomization  Success/failure >10  Less than 10% condition - we can reject or accept null hypothesis; if our confidence interval does not show our null; reject - to arrive at same conclusion (accept our reject both methods need to have same condition)  each value in CI is a plausible value for unknown population parameter. Values located outside are considered implausible.  Confidence interval: two sides; correspond to two sided tests Error Decision Matrix: 1) H0 True -> Reject (Type I error) - reject null when you should not - the lower the alpha; the lower this type of Error can occur alpha of 0.01 includes bigger range than alpha of 0.1 there we increase significance level, 90 % vs. 99% - widen the net; better able to catch - chance of this error, 1%, 5%, 10% : it’s the Alpha Ho True: Accept - correct decision - the chances of this happening are 99%, 95% 0r 90% Ho is False - if we reject the ho; this is correct; rejecting null when you should Power; the ability to correctly reject the null when the time’s right
More Less

Related notes for PSYC 3000
Me

OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Join to view

OR

By registering, I agree to the Terms and Privacy Policies
Just a few more details

So we can recommend you notes for your school.