# INTRODUCTION TO MINITAB VERSION 13

Minitab Training Agenda
•Worksheet Conventions and Menu Structures •Minitab Interoperability •Graphic Capabilities •Pareto •Histogram •Box Plot •Scatter Plot •Statistical Capabilities •Capability Analysis •Hypothesis Test •Contingency Tables •ANOVA •Design of Experiments (DOE)

Worksheet Format and Structure

Menu Bar Session Window Tool Bar

Worksheet Data Window

8 • 4 .Data Window Column Conventions Text Column C1-T (Designated by -T) Date Column C2-D (Designated by -D) Numeric Column C3 (No Additional Designation) ©The National Graduate School of Quality Management v.

8 • 5 Data Rows . Count & Amount Entered Data for Data Rows 1 through 4 ©The National Graduate School of Quality Management v. Date.Other Data Window Conventions Data Entry Arrow Column Names (Type.

8 • 7 .Menu Bar .File Menu Key Functions •Worksheet File Management Save Print Data Import ©The National Graduate School of Quality Management v.

Menu Bar .8 • 9 .Manip Menu Key Functions •Data Manipulation Subset/Split Sort Rank Row Data Manipulation Column Data Manipulation ©The National Graduate School of Quality Management v.

Menu Bar .Calc Menu Key Functions •Calculation Capabilities Column Calculations Column/Row Statistics Data Standardization Data Extraction Data Generation ©The National Graduate School of Quality Management v.8 • 10 .

Stat Menu Key Functions •Advanced Statistical Tools and Graphs Hypothesis Tests Regression Design of Experiments Control Charts Reliability Testing ©The National Graduate School of Quality Management v.Menu Bar .8 • 11 .

Key Functions
•Data Plotting Capabilities Scatter Plot Trend Plot Box Plot Contour/3 D plotting Dot Plots Probability Plots Stem & Leaf Plots

Key Functions
•Advanced Edit and Display Options Data Brushing Column Settings Column Insertion/Moves Cell Insertion Worksheet Settings
Note: The Editor Selection is Context Sensitive. Menu selections will vary for: •Data Window •Graph •Session Window Depending on which is selected.

Key Functions
•Advanced Edit and Display Options Font Connectivity Settings

8 • 15 .Menu Bar .Graph Window Editor Menu Key Functions •Advanced Edit and Display Options Brushing Graph Manipulation Colors Orientation Font ©The National Graduate School of Quality Management v.

Menu Bar .8 • 17 .Help Menu Key Functions •Help and Tutorials Subject Searches Statguide Multiple Tutorials Minitab on the Web ©The National Graduate School of Quality Management v.

MINITAB INTEROPERABILITY ©The National Graduate School of Quality Management v.8 • 18 .

Minitab Interoperability Minitab Excel PowerPoint ©The National Graduate School of Quality Management v.8 • 19 .

.Starting with Excel. Load file “Sample 1” in Excel…. ©The National Graduate School of Quality Management v..8 • 20 .

8 • 21 ..Starting with Excel.. The data is now loaded into Excel…. ©The National Graduate School of Quality Management v.

Starting with Excel. ©The National Graduate School of Quality Management v.8 • 22 ... Highlight and Copy the Data….

Open Minitab and select the column you want to paste the data into…... ©The National Graduate School of Quality Management v.8 • 23 .Move to Minitab.

Select Paste from the menu and the data will be inserted into the Minitab Worksheet…. ©The National Graduate School of Quality Management v..8 • 24 .Move to Minitab..

Lets say that we would like to test correlation between the Predicted Workload and the actual workload….Use Minitab to do the Analysis..8 • 25 .. ©The National Graduate School of Quality Management v. Fitted Line Plot…. •Select Stat… Regression…..

©The National Graduate School of Quality Management v. •Select “Actual Workload” and hit the select button…... Minitab is now asking for us to identify the columns with the appropriate date….. •This will enter the “Actual Workload” data in the Response (Y) data field. •Click in the box for “Response (Y): Note that our options now appear in this box.Use Minitab to do the Analysis.8 • 26 ...

..Use Minitab to do the Analysis. •Now click in the Predictor (X): box….8 • 27 . •Both data fields should now be filled…. Then click on “Predicted Workload” and hit the select button… This will fill in the “Predictor (X):” data field. ©The National Graduate School of Quality Management v. •Select OK.....

. ©The National Graduate School of Quality Management v... •Note that in this case there is a graph and an analysis summary in the Session Window… •Let’s say we want to use both in our PowerPoint presentation….Use Minitab to do the Analysis. •Minitab now does the analysis and presents the results.8 • 28 ..

..Transferring the Analysis. ©The National Graduate School of Quality Management v. Copy Graph. •Let’s take care of the graph first….8 • 29 ... •Go to Edit….

..Transferring the Analysis.. •Go to Edit…. •Open PowerPoint and select a blank slide….8 • 30 . ©The National Graduate School of Quality Management v.. Paste Special.

•Select “Picture (Enhanced Metafile)… This will give you the best graphics with the least amount of trouble.. ©The National Graduate School of Quality Management v..Transferring the Analysis.8 • 31 .

. We can now size and position it accordingly…. ©The National Graduate School of Quality Management v.Transferring the Analysis.. •Our Minitab graph is now pasted into the powerpoint presentation….8 • 32 .

8 • 33 . ©The National Graduate School of Quality Management v.Transferring the Analysis.. •Select Edit…... •Highlight the text you want to copy….. Copy…. •Now we can copy the analysis from the Session window…..

. •Now go back to your powerpoint presentation….8 • 34 .Transferring the Analysis. ©The National Graduate School of Quality Management v. •Select Edit…. Paste….....

•Well we got our data... ©The National Graduate School of Quality Management v.8 • 35 . but it is a bit large….. •Reduce the font to 12 and we should be ok…..Transferring the Analysis.

.. •Now all we need to do is tune the presentation….. ©The National Graduate School of Quality Management v.Presenting the results. •Then we are ready to present…...8 • 36 .. •Here we position the graph and summary and put in the appropriate takeaway.

Graphic Capabilities

Pareto Chart....
•Let’s generate a Pareto Chart from a set of data…. •Go to File… Open Project…. Load the file Pareto.mpj…. •Now let’s generate the Pareto Chart...

Pareto Chart....
•Go to: •Stat… •Quality Tools… •Pareto Chart….

8 • 40 . ©The National Graduate School of Quality Management v...... •Add title and hit OK.Pareto Chart. •Labels in “Category”… •Frequencies in “Quantity”…. Fill out the screen as follows: •Our data is already summarized so we will use the Chart Defects table..

.Pareto Chart..8 • 41 . Minitab now completes our pareto for us ready to be copied and pasted into your PowerPoint presentation….. ©The National Graduate School of Quality Management v.

•Now let’s generate the Histogram of the GPA results.mpj…. •Let’s generate a Histogram from a set of data….. ©The National Graduate School of Quality Management v.8 • 42 .Histogram... •Go to File… Open Project…. Load the file 2_Correlation...

. •Go to: •Graph… •Histogram… ©The National Graduate School of Quality Management v.8 • 43 ..Histogram..

..8 • 44 .. ©The National Graduate School of Quality Management v..Histogram. Fill out the screen as follows: •Select GPA for our X value Graph Variable •Hit OK….

This data does not look like it is very normal….Histogram... Let’s use Minitab to test this distribution for normality….. ©The National Graduate School of Quality Management v. Minitab now completes our histogram for us ready to be copied and pasted into your PowerPoint presentation…...8 • 45 .

©The National Graduate School of Quality Management v.. •Go to: •Stat… •Basic Statistics… •Display Descriptive Statistics…..8 • 46 ..Histogram.

8 • 47 .Histogram.... Fill out the screen as follows: •Select GPA for our Variable….. ©The National Graduate School of Quality Management v. •Select Graphs….

.8 • 48 .Histogram.. ©The National Graduate School of Quality Management v... •Select OK again on the next screen... •Select OK…. •Select Graphical Summary….

. note that our P value of . As for the normality question. So.. This is a great summary slide.038 rejects the null hypothesis (P<...8 • 49 .Histogram.. Note that now we not only have our Histogram but a number of other descriptive statistics as well….05). we conclude with 95% confidence that the data is not normal…. ©The National Graduate School of Quality Management v..

mpj…. Load the file overfill.Histogram....8 • 50 . •Let’s look at another “Histogram” tool we can use to evaluate and present data…. ©The National Graduate School of Quality Management v. •Go to File… Open Project….

. •Go to: •Graph… •Marginal Plot… ©The National Graduate School of Quality Management v.Histogram..8 • 51 ..

8 • 52 . ©The National Graduate School of Quality Management v.Histogram.... •Select head for the X Variable •Select OK….. Fill out the screen as follows: •Select filler 1 for the Y Variable….

.. Note that now we not only have our Histogram but a dot plot of each head data as well. Note that head number 6 seems to be the source of the high readings…..Histogram. This type of Histogram is called a “Marginal Plot”. ©The National Graduate School of Quality Management v..8 • 53 ....

Boxplot..8 • 54 . •Let’s look at the same data using a Boxplot….. ©The National Graduate School of Quality Management v..

... ©The National Graduate School of Quality Management v.Boxplot.8 • 55 ... •Go to: •Stat… •Basic Statistics… •Display Descriptive Statistics.

.8 • 56 ...Boxplot. •Select Graphs….. ©The National Graduate School of Quality Management v. Fill out the screen as follows: •Select “filler 1” for our Variable….

.8 • 57 ...Boxplot. •Select OK…. •Select Boxplot of data….. •Select OK again on the next screen... ©The National Graduate School of Quality Management v.

. We now have our Boxplot of the data..Boxplot..8 • 58 .. ©The National Graduate School of Quality Management v..

.8 • 59 . •There is another way we can use Boxplots to view the data... •Go to: •Graph… •Boxplot.Boxplot. ©The National Graduate School of Quality Management v.....

Boxplot....
Fill out the screen as follows: •Select “filler 1” for our Y Variable….

•Select “head” for our X Variable…. •Select OK…..

Boxplot....

Note that now we now have a box plot broken out by each of the various heads.. Note that head number 6 again seems to be the source of the high readings…..

Scatter plot....
•Let’s look at data using a Scatterplot…. •Go to File… Open Project…. Load the file 2_Correlation.mpj…. •Now let’s generate the Scatterplot of the GPA results against our Math and Verbal scores...

Scatter plot....8 • 63 . •Go to: •Graph… •Plot. ©The National Graduate School of Quality Management v...

8 • 64 .... Fill out the screen as follows: •Select GPA for our Y Variable…..Scatter Plot.. •Select OK when done.. ©The National Graduate School of Quality Management v. •Select Math and Verbal for our X Variables….

©The National Graduate School of Quality Management v.Scatter plot... We now have two Scatter plots of the data stacked on top of each other… We can display this better by tiling the graphs….8 • 65 ..

To do this: •Go to Window… •Tile....8 • 66 . ©The National Graduate School of Quality Management v...Scatter plot.

Scatter plot.. Now we can see both Scatter plots of the data… ©The National Graduate School of Quality Management v..8 • 67 ..

. •Go to: •Graph… •Matrix Plot... ©The National Graduate School of Quality Management v...8 • 68 . •There is another way we can generate these scatter plots….Scatter plot.

Scatter Plot.8 • 69 .. ©The National Graduate School of Quality Management v..... Fill out the screen as follows: •Click in the “Graph variables” block •Highlight all three available data sets… •Click on the “Select” button. •Select OK when done...

.Scatter plot...8 • 70 . We now have a series of Scatter plots. ©The National Graduate School of Quality Management v. each one corresponding to a combination of the data sets available… Note that there appears to be a strong correlation between Verbal and both Math and GPA data….

8 • 71 .Minitab Statistical Tools ©The National Graduate School of Quality Management v.

PROCESS CAPABILITY ANALYSIS ©The National Graduate School of Quality Management v.8 • 72 .

8 • 73 . ©The National Graduate School of Quality Management v.Let’s do a process capability study…. Open Minitab and load the file Capability.mpj….

SETTING UP THE TEST…. Capability Analysis (Weibull)…. ©The National Graduate School of Quality Management v.8 • 74 . Go to Stat… Quality Tools….

Enter a lower spec of 10 and an upper spec of 30.SETTING UP THE TEST….. Select “Torque” for our single data column. ©The National Graduate School of Quality Management v..8 • 75 . Then select “OK”….

. This equates to a Z value of 3*0.29 standard deviations or sigma values.INTERPRETING THE DATA….8 • 76 . Note that the Long Term capability (Ppk) is 0. Note that the data does not fit the normal curve very well.43=1. This equates to an expected defect rate PPM of 147.43..055. ©The National Graduate School of Quality Management v.

HYPOTHESIS TESTING ©The National Graduate School of Quality Management v.8 • 77 .

Setting up the test in Minitab •Load the file normality.8 • 78 .mpj….. ©The National Graduate School of Quality Management v.

. •Go to STAT….8 • 79 . ©The National Graduate School of Quality Management v. Normality Test…. •It’s important that we check for normality of data samples. •Let’s see how this works….Checking the Data for Normality…. Basic Statistics..

•Check AndersonDarling •Click OK ©The National Graduate School of Quality Management v.Set up the Test •We will test the “Before” column of data….8 • 80 .

05 we can assume the “Before” data is normal •Now repeat the test for the “After” Data (this is left to the student as a learning exercise.Analyzing the Results •Since the P value is greater than ..) ©The National Graduate School of Quality Management v.8 • 81 .

. •To accomplish this go to Manip… Stack… Stack Columns…. ©The National Graduate School of Quality Management v. •We now want to see if we have equal variances in our samples. our data must be “stacked”. •To perform this test.8 • 82 .Checking for equal variance.

. •Select OK…. •Select both of the available columns (Before and After) to stack. •Type in the location where you want the stacked data….. ©The National Graduate School of Quality Management v.Checking for equal variance...8 • 83 . •Type in the location where you want the subscripts stored… In this example we will use C3…. In this example we will use C4….

©The National Graduate School of Quality Management v..8 • 84 . •Now that we have our data stacked.Checking for equal variance..… •Go to Stat… ANOVA…. Test for equal Variances. we are ready to test for equal variances..

Setting up the test…. •Then select “OK”. •We set our Confidence Level for the test (95%). •Our response will be the actual receipt performance for the two weeks we are comparing.. In this case we had put the stacked data in column C4….8 • 85 . •Our factors is the label column we created when we stacked the data (C3). ©The National Graduate School of Quality Management v.

the PValue of . We can see from the P-Value of . ©The National Graduate School of Quality Management v.229 indicates that we would fail to reject the null hypothesis. Again. Since they overlap.Analyzing the data….8 • 86 •Here we have box plot representations of both populations. •Here. Note that the F test assumes normality •Levene’s test also compares the variance of the two samples and is robust to nonnormal data. we see the 95% confidence intervals for the two populations.263 that again we would fail to reject the null hypothesis. . we know that we will fail to reject the null hypothesis. •Note that we get a graphical summary of both sets of data as well as the relevant statistics…. •The F test results are shown here.

©The National Graduate School of Quality Management v.Statistics…... Basic Statistics.Lets test the data with a 2 Sample t Test •Under Stat… Basic . •Go to Stat…. In this example we will be using a 2 Sample t Test…. 2 Sample t….8 • 87 . We see several of the hypothesis tests which we discussed in class.

we will load C4 for our samples and C3 for our subscripts. •Since we have already tested for equal variances.8 • 88 .Setting up the test…. ©The National Graduate School of Quality Management v. •Since we already have our data stacked. we can check off this box… •Now select Graphs….

•We see that we have two options for our graphical output. For this small a sample.Setting up the test….8 • 89 . ©The National Graduate School of Quality Management v. Hit OK again on the next screen…. Boxplots will not be of much value so we select “Dotplots of data” and hit “OK”.

.922. We therefore find that the data does not support the conclusion that there is a significant difference between the means of the two populations.Interpreting the results…... •Note that here we have a P value of .8 • 90 . ©The National Graduate School of Quality Management v. •In the session window we have each population’s statistics calculated for us.

•The dotplot shows how close the datapoints in the two populations fall to each other.8 • 91 . The close values of the two population means (indicated by the red bar) also shows little chance that this hypothesis could be rejected by a larger sample ©The National Graduate School of Quality Management v.Interpreting the results….

 It can also be used in measurement system studies to determine if operators are getting the same mean value across the same set of samples.8 • 92 .Paired Comparisons  In paired comparisons we are trying to “pair” observations or treatments.mpj ©The National Graduate School of Quality Management v. An example would be to test automatic blood pressure cuffs and a nurse measuring the blood pressure on the same patient using a manual instrument.  Let’s look at an example: 2_Hypothesis_Testing_Shoe_wear.

2_Hypothesis_Testing_Shoe_wear.8 • 93 . Ho: Material “A” wear rate = Material “B” wear rate Ha: Material “A” wear rate  Material “B” wear rate ©The National Graduate School of Quality Management v.  Our data has been collected using ten boys.mpj  In this example we are trying to determine if shoe material “A” wear rate is different from shoe material “B”. whom were asked to wear one shoe made from each material.

•Basic Statistics… • Paired t….8 • 94 . ©The National Graduate School of Quality Management v..Paired Comparison •Go to Stat….

©The National Graduate School of Quality Management v.8 • 95 .Paired Comparison •Select the samples… •Go to Graphs….

. •Then select OK.8 • 96 . ©The National Graduate School of Quality Management v..Paired Comparison •Select the Boxplot for our graphical output.

Paired Comparison We see how the 95% confidence interval of the mean relates to the value we are testing. In this case. This gives us confirmation that the shoe materials are significantly different. ©The National Graduate School of Quality Management v. the value falls outside the 95% confidence interval of the data mean.8 • 97 .

8 • 98 .CONTINGENCY TABLES (CHI SQUARE) ©The National Graduate School of Quality Management v.

load the file Contingency Table.mpj. ©The National Graduate School of Quality Management v.8 • 99 ... •Enter the data in a table format. For this example.Entering the data….

©The National Graduate School of Quality Management v.8 • 100 .Let’s set up a contingency table…. Tables… Chi Square Test…. •Contingency tables are found under Stat….

•Select the columns which contain the table. Then select “OK” ©The National Graduate School of Quality Management v.Setting up the test….8 • 101 .

©The National Graduate School of Quality Management v. Minitab calculates the Chi Square statistic for each data point and totals the result.8 • 102 .846. •Here.Performing the Analysis…. •Minitab builds the table for you. The calculated Chi Square statistic for this problem is 30. Note that you will have the critical population and test statistics displayed in the session window. Note that our original data is presented and directly below. Minitab calculates the expected values.

8 • 103 .ANalysis Of VAriance ANOVA ©The National Graduate School of Quality Management v.

Let’s set up the analysis •Load the file Anova example.mpj… •Stack the data in C4 and place the subscripts in C5 ©The National Graduate School of Quality Management v.8 • 104 .

•Select Stat… •ANOVA… •One way… ©The National Graduate School of Quality Management v.8 • 105 .Set up the analysis….

©The National Graduate School of Quality Management v.8 • 106 . •Select • C4 Responses • C5 Factors •Then select Graphs….Set up the analysis….

. •Choose boxplots of data.Set up the analysis…. •Then OK ©The National Graduate School of Quality Management v.8 • 107 ..

Note that the P value is less than .05 that means that we reject the null hypothesis ©The National Graduate School of Quality Management v.8 • 108 .Analyzing the results….

•Choose Stat •ANOVA •Main Effects Plot…. ©The National Graduate School of Quality Management v.Let’s Look At Main Effects….8 • 109 .

Main Effects Select •C4 Response •C5 Factors •OK ©The National Graduate School of Quality Management v.8 • 110 .

Data Means for Liters Per H 22 21 Liters Per H 20 19 18 Liters/Hr 1 Liters/Hr 2 Liters/Hr 3 Formulation Formulation 1 Has Lowest Fuel Consumption ©The National Graduate School of Quality Management v. Main Effects Plot ..Analyzing Main Effects.8 • 111 .

DESIGN OF EXPERIMENTS (DOE) FUNDAMENTALS ©The National Graduate School of Quality Management v.8 • 112 .

. Go to •Stat… •DOE… •Factorial..First Create an Experimental Design..8 • 113 ... ©The National Graduate School of Quality Management v.. •Create Factorial Design.

.8 • 114 .. Select 2 Level Factorial design with 3 factors Then go to Display Available Designs….First Create an Experimental Design. ©The National Graduate School of Quality Management v.

We will be using the Full (Factorial) for 3 factors and we can see that it will require 8 runs… Now.Bowling Example (continued) We can now see the available experimental designs….8 • 115 . Once at the main screen select Designs. ©The National Graduate School of Quality Management v.. select OK and go back to the main screen..

select OK and go back to the main screen.Bowling Example (continued) Select your design…..8 • 116 . Once at the main screen select Factors. We will be using the Full (Factorial) and again we can see that it will require 8 runs… Now. ©The National Graduate School of Quality Management v..

select OK and go back to the main screen. ©The National Graduate School of Quality Management v.Bowling Example (continued) Fill in the names for your factors…. Once at the main screen select Options. Then fill in the actual conditions for low (-) or high (+) Now...8 • 117 .

Now. ©The National Graduate School of Quality Management v.. select OK and go back to the main screen.8 • 118 . Once at the main screen select OK.Bowling Example (continued) Remove the option to Randomize Runs…..

We are now ready to analyze the results… ©The National Graduate School of Quality Management v. type your Data from each of your experimental treatments into C8.8 • 119 .Bowling Example (continued) Minitab has now designed our experiment for us…. Now.

.8 • 120 .. ©The National Graduate School of Quality Management v.Bowling Example (continued) Go to •Stat….. •Analyze Factorial Design. •DOE… •Factorial..

Then.Bowling Example (continued) Highlight your Data column and use Select to place it in the Responses box. ©The National Graduate School of Quality Management v.8 • 121 . select the Terms Option.

We need do nothing further.8 • 122 . Select OK. Then.Bowling Example (continued) Note that Selected Terms has all of the available choices already selected. at the main screen select Graphs ©The National Graduate School of Quality Management v.

05. Select OK to return to the main screen and then select OK again.8 • 123 . ©The National Graduate School of Quality Management v.Bowling Example (continued) Select your Effects Plots and reset your Alpha to .

All the remaining factors and interactions are not statistically significant. ©The National Graduate School of Quality Management v.Bowling Example (continued) Note that only one effect has a significance greater than 95%.8 • 124 .

8 • 125 .Bowling Example (continued) •Another way we can look at the data is to look at the Factorial Plots of the resulting data. •Factorial… •Factorial Plots…. ©The National Graduate School of Quality Management v. •Go to •DOE….

8 • 126 .Bowling Example (continued) •Select Main Effects Plot and then Setup… ©The National Graduate School of Quality Management v.

•Then select “OK” and OK again on the main screen.Bowling Example (continued) •Select C8 as your response •Select “Wristband”. ©The National Graduate School of Quality Management v.8 • 127 . “Ball” and “Lane” as your factors.

Bowling Example (continued) •The magnitude of the vertical displacement indicates the strength of the main effect for that factor.8 • 128 . Here we see that the wristband has dramatically more effect than any other factor. •This plot also shows you the direction of the main effects. We know from our earlier plots that the wristband is the only statistically significant effect @ 95% confidence. ©The National Graduate School of Quality Management v. We clearly see that the “with” condition is related to the higher level of performance.

Bowling Example (continued) •Now lets look at the interactions.. •Go to •DOE…. •Factorial… •Factorial Plots… ©The National Graduate School of Quality Management v...8 • 129 .

©The National Graduate School of Quality Management v.8 • 130 .Bowling Example (continued) •Select InteractionPlot and then Setup…..

•Select “Wristband”. ©The National Graduate School of Quality Management v. “Ball” and “Lane” as your factors. •Then select “OK” and OK again on the next screen….Bowling Example (continued) •Select C8 as your response variable.8 • 131 .

©The National Graduate School of Quality Management v.Bowling Example (continued) •We know from our earlier analysis that none of these interactions were statistically significant for this experiment…. •The more the lines diverge from being parallel. •We see that the strongest interaction (still not significant) is between the lane and the ball..8 • 132 . the more the interaction.

.8 • 133 . •Note that Wristband has the strongest effect followed by the interaction between the Wristband and the Lane...Bowling Example (Session Window) •This is where Minitab shows us the Main Effects and Interaction Effects. •You can also see that there is zero error •This is because only 1 run was performed with no replications ©The National Graduate School of Quality Management v.