QI Macros Lesson 6 - Anova and Other Analysis Tools | |||||||||||||||||||
We help people think! |
In this Issue
Using the Anova and Other Analysis ToolsMost Six Sigma black belts get into more detailed analysis of data to determine the variation. ANOVA (or ANalysis Of VAriance) seeks to understand how data is distributed around a mean or average. To use any of the statistical analysis tools of Excel through the QI Macros:
Anova and Analysis Tools in Excel 2003 and previous versions
Anova and Analysis Tools on Excel 2007 Ribbon
Analysis of VarianceAnalysis of variance can help you determine if two or more populations have the same "mean" or average. This is also called "hypothesis testing." The "null" hypothesis (Ho) is that Mean1 = Mean2. The goal is to disprove this (i.e., the populations have two different means) at a certain confidence level (95% or 99%). Excel and the QI Macros can perform single and two-factor analysis. Single Factor Analysis - Watch VideoFrom Intro to Statistical Quality Control by Montgomery, we want to compare how four different concentrations of hardwood affect paper tensile strength:
Using Excel and the QI Macros, select data in columns B2:E8 and run a single factor Anova at the 99% or alpha=0.01 level:
Since the P-value is less than alpha, the null hypothesis is not true (i.e., the means are different). To download test files anova1.xls and anova2.xls, right click on the link and choose SAVE TARGET AS to save the file. If you open these in Explorer with Excel as a subtask, you won't be able to run the analysis. Two-Factor AnalysisWhat if you have two populations of patients (male/female) and three different kinds of medications, and you want to evaluate their effectiveness? You might run a study with two or more "replications."
Then, using Excel and the QI Macros, run a two-factor analysis with replication (alpha=0.05 for a 95% confidence).
Here, the P-value for Male/Female is greater than alpha, so we can’t reject the hypothesis that the means are the same at the stated alpha level. The P-Value for Drugs is greater, so we can’t reject the null hypothesis at the stated alpha level. The P-value for the interaction of the drugs and patients is less than .05, so the effectiveness of three drugs is not the same for the two categories of patient. Analysis of Attribute Data (Counted)To analyze attribute data, Excel requires you to set the data up in a way that can be analyzed. The example below shows how to set up the data for two categories of patients treated with three different drugs.
Then, if you're just interested in the single factor DRUGS, select and run a single factor on the three drug columns.
If you're interested in a two-factor analysis (patients vs drugs), select and run a two-factor analysis with repetition (more than one patient in the category receives the same drug).
Determining Sample SizesIn manufacturing applications, you often need to figure out how many samples to take to ensure that you get a valid sample size of a larger lot.
For more detailed information Watch this Video or see our Selecting Sample Sizes article. In Lesson #7 we will cover Measurement System Analysis and Gage R&R. View previous lesson .......................View next lesson To sign up for this series tell your friends to send an email to qimacros@aweber.com
© 2007 KnowWare International Inc. (888) 468-1537 |
It contains over 70 fill in the blank templates such as the Ishikawa diagram, QFD, DOE, FMEA, PPAP, and Gage R&R for MSA. Performs ANOVA, t-test, F-test, and regression analysis.
|
|||||||||||||||||
