ATHK1001 Lecture Notes - Lecture 12: Null Hypothesis, Type I And Type Ii Errors, Regression Analysis
16 Jun 2018
School
Department
Course
Professor
Lecture - Inferences
Significance Level
• The p-value is the probability that the data would be at least as extreme
as that observed, if the null hypothesis was true
• Decided before study what the critical p-level is usually .05
• Using lower p-level means we are less likely to reject the null hypotheses
when we shouldn’t type 1 error
• Less likely to accept alternative when it is correct (type 2 error)
Sample Size Matters
• The larger the numbers of samples, the more closely the sample
distribution will approximate the normal distribution
• Easier to reject null hypotheses with bigger sample
Using p <. 05
• Significance level of .05 means we can reject the null hypothesis at the .05
level if there is less than a 5% chance of getting the outcome observed if
the full null hypothesis was true
p-value
• The specific probability of the sample being drawn from the population
• Tells you how likely there is an effect, not how big the effect is
Logic of the Inference
• If 2 samples are drawn from the same population, our best guess is that
the difference between their means is zero (null hypothesis)
• The CLT tells us that for repeated samples form the same population the
difference between two means will be roughly normally distributed
• Therefore if 2 samples did come from the same population then 95% of
the time they are within two SEs
Statistical Tests
• Formal way of calculating the probability of the null hypothesis
• Produce a score that can be compared against an appropriate statistical
distribution with know properties to see how likely that score is
Polls
• Are categorical
Null Hypothesis
• The pattern of frequencies in one variable is the same across all levels of
the other variable, there is no association between the two categories
find more resources at oneclass.com
find more resources at oneclass.com
Document Summary
Sample size matters: the larger the numbers of samples, the more closely the sample distribution will approximate the normal distribution, easier to reject null hypotheses with bigger sample. Statistical tests: formal way of calculating the probability of the null hypothesis, produce a score that can be compared against an appropriate statistical distribution with know properties to see how likely that score is. Null hypothesis: the pattern of frequencies in one variable is the same across all levels of the other variable, there is no association between the two categories. Alternative hypothesis: the pattern of frequencies is different across levels of the other variables, there is an association between the two categories, reject if p value is less than 0. 005. Is on that varies with both variables being considered and thus obscures their true relationship: cant set up a randomized control study.
