Minitab Commands

Minitab Menu Commands

Display Data Numerically and Visually

Categorical

Counts/Per Cents

Stat>Tables>Tally specify a categorical variable and check counts and/or per cents

Two-Way Table

Stat>Tables>Cross Tablulation specify the two categorical variables and check desired counts and per cents (row, column, totals)

Pie Chart

Graph > Pie Chart

Bar Chart

Graph > Bar Chart

Quantitative

Summary Stats

Stat > Basic Statistics > Display Descriptive Statistics

Stem and Leaf Plot

Graph > Stemplot

Histogram

Graph > Histogram

Time Series

Graph > Time Series Plot

Box Plot

Graph > Box Plot

Scatter Plot

Graph > Plot (y is response variable, x is explanatory variable)

Scatter Plot
Grouped

Graph > Scatter Plot (y is response variable, x is explanatory variable)
If you want separate symbols for different categories (e.g. M & F),

Display	For Each	Group Variable
Symbol	Group	Your category variable

Scatter Plot
Smoothed

Graph > Plot (y is response variable, x is explanatory variable)
Grouped or not; Set Display to Lowess

Histogram with superimposed normal curve

Stat > Basic Statistics > Display Descriptive Statistics > Graph > Histogram of Data with Normal Curve

Graphical display of summary statistics

Stat > Basic Statistics > Display Descriptive Statistics > Graph > Graphical Summary

Convert column to standardized variables

Calc > Standardize

Given a value x, find proportion less than x

Calc > Probability Distributions > Normal > Cumulative Probability

(set mean, s.d., x value)

Given a proportion, find value that determines that proportion

Calc > Probability Distributions > Normal > Inverse Cumulative Probability

(set mean, s.d., p value)

Scatter Plot

Graph > Plot >

Calculate, Manipulate, and Misc

Calculate products of entries in two columns

Calc > Calculator, put in, e.g. C1 * C2

Find sum of column entries

Calc > Column Statistics > Sum

Load and run a macro

File > Other Files > Run an Exec

Put number of times in Number of Times to Execute

Select Fileload the .mtb file you want.

Find mean of each row

Calc -> Row Statistics.Select "Mean";

Generate random data from a normal distribution

Calc -> Random Data -> Normal

To generate 20 columns and 50 rows, generate 50 rows of data, store results in C1-C20

Set m and s

Stack Columns

Manip -> Stack/UnStack

Save several graphs to print in a report

Right click on graph, click on Append to Report
Repeat this step for each graph
When done, use the Window menu to go to Project Manager
In Project Manager, click on Report Pad Your graphs are all there. You can type right in the Report Pad or cut and paste from the Session window.
While Report Pad is highlighted, use the File Menu to Save Report Pad. This saves the Report as an RTF file.
In Word, open the file you saved by setting the file type to Rich Text Format or RTF. Then you can edit and save as a Word file.

Z-Tests

Construct Z confidence intervals:

Stat -> Basic Statistics -> 1-Sample Z

Use, e.g.C1-C20 as Variables, enter 100 for Sigma

Hit "Options" to specify a confidence level

Perform Z-test for mean

Stat -> Basic Statistics -> 1-Sample Z

Enter variable, sigma (s) and test mean (=m_n)

Options: specify greater than, not equal to, or less than for H_a

Output:Z = test statistic and p = p-value

Calculate Power at a specific value or sample size or difference to detect value for Z-test

Stat ->Power and Sample Size -> 1-Sample Z

Enter two of: Sample size, Difference from mean you want to detect, Power level (in decimal)

Enter Sigmavalue

Options: specify greater than, not equal to, or less than for H_a, significance level

(For trial and error investigation of different sample sizes or differences, store Sample Sizes, Differences, and Power Values in (for example) C1, C2, C3)

Calculate Probability of Type II Error at a specific value for Z-test

Stat ->Power and Sample Size -> 1-Sample Z

Enter Sample size, Difference from population mean, Sigma

Options: specify greater than, not equal to, or less than for H_a, significance level

Probability of Type II Error is 1 - Power

T-tests

Given a value t, find probability of a reading less than x in t-distribution (i.e. find p-value, given t)

Calc > Probability Distributions > t > Cumulative Probability

(set d.f., put t into input constant box)

Given a probability, find t-value that determines that probability

Calc > Probability Distributions > t > Inverse Cumulative Probability

(set d.f., put p value into input constant box (using decimals, so 90% would be entered as .9)
(This value is T such that the Pr(x < =T) = p-value)

Construct t confidence intervals:

Stat -> Basic Statistics -> 1-Sample t

Enter variable(s)

Options: specify a confidence level

Perform t-test for mean

Stat -> Basic Statistics -> 1-Sample t

Enter variable, test mean (=m_n)

Options: specify greater than, not equal to, or less than for H_a

Output:t = test statistic and p = p-value

Construct matched pairs t confidence intervals:

Stat -> Basic Statistics -> Paired t

Enter first and second variables

Options: specify a confidence level

Perform matched pairs t-test for differences of means

Stat -> Basic Statistics -> Paired t

Enter first and second variables

Options: specify a confidence level, difference mean, direction forH_a

Output:t = test statistic and p = p-value

Construct 2-sample t confidence intervals:

Stat -> Basic Statistics -> 2-Sample t

If both samples are in one column, enter sample and subscript names

If samples in different columns, enter first and second variables

Options: Specify alternative and confidence level

Perform 2-sample t-test for difference of means

Stat -> Basic Statistics -> 2-Sample t

If both samples are in one column, enter sample and subscript names

If samples in different columns, enter first and second variables

Options: specify a confidence level, test mean, alternative

Output:t = test statistic and p = p-value

Proportion Tests

Construct confidence interval, perform z-test for proportion

Stat -> Basic Statistics -> 1 Proportion
If you have the data in a column, check Samples in columns: enter column
If not, check Summarized data, enter total number of trials and number of successes

Options: Check Use test and interval based on normal distribution

Specify a confidence level, test proportion (p₀,) correct alternative

Construct confidence interval, perform z-test for difference of proportions

Stat -> Basic Statistics -> 2 Proportions

If both samples are in one column, enter sample and subscript names

If samples in different columns, enter first and second variables

Options: Use test and interval based on normal distribution

Specify a confidence level, test difference, correct alternative

Chi Square Tests

Chi-Square Test the raw data in columns

Stat > Tables > Cross Tabulation. Select the column names you will compare. Check Chi-Square analysis, and select Above and expected count.

Chi-Square Test given a table of observed totals

Enter the table of observed values in columns of a new worksheet.

Stat>Tables>Chi-Square Test

Enter the columns that your observed values are in

Anova tests

Perform ANOVA test on 2 or more means, where samples are in different columns

Stat -> ANOVA -> 1-way unstacked
Use column names as response variables

Perform ANOVA test on 2 or more means, where samples are in same column with another column used as sorting variable

Stat -> ANOVA -> 1-way stacked

Use column name as response variable, Sorting column name as Factor variable

If p-value is significant, check individual comparisons

Stat -> ANOVA -> 1-way stacked-> Comparisons -> Tukey

Put in family error rate (often 5%).Look for confidence intervals that include 0.

Bar chart of means

Graph -> Chart -> Function: Mean

Y=data variable, X=sorting variable

Right click -> Edit to type letters above bar charts to show relationships

Correlation and Regression

Find Correlation

Stat > Basic Statistics > Correlation

Find Regression information

Stat > Regression > Regression
Use Storage > Residuals if you want to save the residuals into a new column

Plot a regression line

Stat > Regression > Fitted Line Plot >Linear

Use Storage > Residuals if you want to save the residuals into a new column

Make prediction and inferences for prediction

Stat -> Regression -> Regression

Hit Options, and put value in prediction intervals for new observations. If, for example, you enter 10000, this produces a confidence interval for the mean response for all individuals with x=10,000 and a prediction interval for the response of a single individual when x=10,000).

Disclaimer