Please enable JavaScript.
Coggle requires JavaScript to display documents.
Chapter 21 - Hypothesis Testing - Coggle Diagram
Chapter 21 - Hypothesis Testing
statistical tools used to draw conclusions based on data.
Extremely accurate statistical conclusions if team can ensure the measurements are good, have the right sample size and know how to set up the right type of hypothesis test
covers three broad categories
Comparing statistics to hypothesis about the data or population
Answering the question whether something has changed within the data.
Testing whether the data you have fits the data model
Guidlines
create null hypothesis and alternative hypothesis
Statistic compared against reference criteria or distribution
Statistic computed from sample data
How the calculated statistic compares to the reference criteria determines whether you accept the null hypothesis or reject in favor for alternative
Hypothesis tests have two main parts
Null hypothesis - Ho - a statement about the data that reflects no effect or difference
Alternative Hypothesis - Ha - there IS a statistical difference...
Risk of error
Type I - null is rejected when it's actually true; producer risk
Type II - null is accepted when it's actually false; consumer risk
Hypothesis Tests for Discrete Data
1 Proportion Test - when there is only one factor for X and one level of interest for the input and
2 Proportion Test - when there is one factor of X and 2 levels of interest for the input
Hypothesis Tests for Continuous Normal Data
1 Sample Test or Paired T Test - used when comparing means - compares the mean of a sample to the target mean
Chi Square Test (1 variance test) - compares standard deviation or variance
2 Sample Test - compares means - between two samples of the different x factor
Hypothesis Tests for Continuous Non-Normal Data
One Sample Wilcox - compares medians between sample and hypothesized sample or a new sample to a previous sample before changes were made
Mann-Whitney Test - compares medians between samples of two factors of X
Chi Square Test
Hypothesis Tests help Six Sigma must answer the question of "is this number so statistically different we must take action on this information?