Introduction to Minitab
These instructions are written for Minitab version 14.
When you open Minitab (Mtb), you find a session window in
the top half of the main window and the data window in the
bottom half. There is also a project manager window which
is usually hidden. To see the project manager window,
click on Window at the very top of the main window and
check Project Manager.
To activate one of these windows, click on it. The toolbars
at the top changes depending on the active window.
The main window menu bar has the usual menu items (File
for opening, closing, and saving; Edit for copying,
pasting, and deleting) and specific menu items for Data
manipulation, Calculation, Graphing, and Statistics.
The Editor menu item allows you to Enable Commands in
the session window and to make its Output Editable.
If you wish to clean up the session window by deleting e.g. error
messages and adding comments, you should uncheck Enable
Commands and check Output Editable. The Tools
item provides access to the Windows calculator, Windows Explorer,
and Notepad. It also allows you to customize all other menu
items, e.g. to add statistical routines or custom graph routines
to the menu.
Click on File in the main Mtb window to open Mtb files. You can open Mtb projects (file extension .mtj), Mtb worksheets (file extensions .mtw, .mtb, .mtp and others), and Mtb graphs (file extension .mgf). You can also open spreadsheets and text files. Usually you'll be interested in projects and worksheets.
Worksheets are essentially data files in spreadsheet format. Opening them creates a new data window. No other information is loaded into Mtb. Several worksheets can be open at the same time.
Projects contain also records of what happened in the session window, graphs, a list of all variables that are currently stored, possibly a report in MS Word format etc. You can open only one project at a time.
To save a worksheet, click on File > Save Current
Worksheet or File > Save Current Worksheet As ... and
follow the prompts. Have a USB drive or some other
media ready to save your worksheet.
To save a project, click on File > Save Project and follow the prompts. When you save the project, you save all the information about your work: the contents of all the windows, including the columns of data in each Data window, stored constants and matrices, the complete text in the Session window and History folder, and each Graph window. This allows you to interrupt your work and pick up later where you left off.
Activate the data window by clicking on it.
Check to insure that the arrow in the box in the upper left
corner of the data window page is pointing downward (click on it
to switch it).
Place the name of your variable in the top cell of the column
(directly under C1, or whatever column you put your data
in). Move the cursor to the first cell in your column
and enter your first data value. Press Enter. If the
arrow in the upper left corner is pointing down, the cursor will
automatically move to the next cell in the column. You can also
use the up and down arrows or the mouse to move to other cells.
Continue until you have all the values of that variable entered
into that column. Don't leave any empty cells.
Move to the next column and repeat the steps with your next
variable.
Minitab considers a data column as numerical as long as all entries in its cells are numbers. If one or more cells are non-numerical (text, symbols), the entire data column is considered categorical, and the column label is changed from e.g. C2 to C2-T. Integer entries with spaces are interpreted as dates, and the column label is changed from C2 to C2-D. It can be tedious to undo such a change in data type, so be careful when entering data.
Data types can be changed by going to Data > Change Data Type.
To copy data (cells, groups of cells or columns) within a Mtb
data window or between data windows, select the cells with the
mouse. You can also select a group of columns (highlight the
column names instead of the cells).
Go to Edit > Copy Cells.
Move the mouse to the location where you want to enter the data
and go to Edit > Paste Cells.
You can also copy and paste data to and from other applications
(spreadsheet columns, text files) in this manner.
Caution:
These boxes pop up when you choose submenus from the Data,
Calc, Stat, and Graph menus.
All dialog boxes list available variables in a big window on the
left and expect input in smaller fields on the right.
Click in a Variables input field to activate it.
Select one or more variables from the list on the left by typing
its name in the input field or by highlighting it and clicking on
the Select button. Double-clicking on the variable name also
works.
Other fields expect you to enter the location of output data or
to enter numerical values. Radio buttons or check boxes may
change the options for input. Click on OK when you are done.
Click in the Graph Variables input field to activate
it.
Select one or more variables from the list on the left by typing
in its name or by highlighting it and clicking on the Select
button.
Design your graph with the Data View window (e.g. symbols
or connected points or bar graphs). Available choices appear when
you click on the arrow symbols. Scale, Label,
and Data Options provide ways to add titles and
legends, control the axes, spacing of tick marks, grids etc..
To generate Multiple Graphs, click on the
corresponding button Frame > Multiple Graphs. By
clicking on the appropriate radio button, you can overlay several
graphs on the same page (e.g. side by side box plots).
Click on OK when you are done.
Go to Graph > Stem-and-Leaf.
Use the Graph Dialog window to select
the desired variable.
Within the dialog box you can choose to trim outliers (click in
box) or adjust the stem increments (i.e. split or combine stems).
Click OK to generate the plot.
The plot will appear in the Session window.
Go to Graph > Pie Chart.
Choose between charting raw data and values from a table.
Go to Graph > Bar Chart. Choose whether your data represent raw data ("counts of unique values") or values in a table. Then choose the type of bar chart you want to produce.
Your two-way table may be in a single column (say in the column DATA) , with classification variables for the table rows and columns in two other columns (say in columns ROW and COLUMN). Alternatively it may in the form of several columns, e.g. with labels DATA1 and DATA2, properly lined up, with an additional column, say ROWS, for the row labels.
This can be done either by using the Data View, Scale, Labels, Data Options buttons in the main dialog window before the chart has been made, or by editing portions of the chart after it has been made. See the section on changing the look of a histogram.
If the chart doesn't look the way you want it, close it and redo it. All previous settings are still in the dialog boxes. Change a few things at a time until the chart looks right.
Go to Graph > Histogram.
Choose the type of histogram and click OK..
In the Graph Dialog box, select
the variable(s) for which you want to make a histogram. Note that
you can choose several variables at once, so it is possible to
make histograms of several different variables simultaneously.
These will usually appear in separate graph windows, unless you
choose to have graphs overlaid on the same page in Multiple
Graphs.
Back to top
You can change the appearance of a histogram by clicking on the appropriate buttons in the graph dialog window.
You can also change the histogram after it has been made by clicking on the bars, the title, the axes, the axis labels etc. and editing them. For example,
Go to Graph > Boxplot.
Choose a graph type from the top or bottom row. "Simple
Y" here means that the data are in a single column,
"Multiple Y's" means several columns. Choose "With
Groups" if there is an additional column of categorical
variables for classification. This will allow you to make
side-by-side boxplots for comparison.
Enter the column name(s) for the data in the graph dialog window. Use the buttons in the dialig window to change the overall appearance and press OK.
Boxplots can also be edited after they have been made - double-click on the title, the box, axes and labels, or the entire window to bring up options for changing the appearance.
A graph can be moved around on the screen, its size may be
changed as any other window, and it may be closed. Once a
graph has been closed, it is gone and must be recreated. In
addition, the menu bar item Window will allow
you to manage the histograms. With Window > Close all
Graphs you can clean up a cluttered work area.
To copy a graph to a report (e.g. to MS Word), righ-click on it
and select Copy Graph, then paste the graph into
your report. In MS Word it is possible to resize the graph
easily.
Go to Stat > Basic Statistics > Diplay Descriptive
Statistics.
Enter your desired variable(s) in Variable box.
Click OK.
The descriptive statistics (number of observations, mean, median,
maximum, minimum, quartiles, standard deviation, and a few
others) will appear in the Session window.
By clicking on the Graphs button in the dialog window, you can produce histograms or boxplots.
Go to Calc > Calculator.
You can enter calculator operations for variables from the left
window in the Expression window and store the result in
another column. You can also use the functions from the
list (highlight and click Select).
Start by putting a title on the new column you want to form.
Go to Calc > Calculator.
In the Expression box, enter the formula for the
transformation.
Enter the name of the new column in the Store result in
variable box and click OK.
Example:
To transform centigrade temperatures (stored in C1) to
Fahrenheit and store the Fahrenheit data in C2, go to Calc
> Calculator. In the Store results
in variable box, enter C2. In the Expression
box, enter 32 + (9/5)*C1. Then click OK. Column C2
will now contain the Fahrenheit temperatures.
Go to Calc > Standardize.
Select the data columns you want to standardize in the Dialog box and enter the column where the
standardized columns go. Leave the radio button at "subtract
mean and divide by standard deviation". Click OK.
Go to Calc > Probability Distributions > Normal.
To find the probability that a variable with a normal
distribution is less than a certain value, say X, click the radio
button "Cumulative probability".
Enter the mean and standard deviation.
Click the button "Input constant" and enter the value X
in the field.
Click OK.
The answer appears in the session window.
To find a value x such that a variable with a
normal distribution is less than x with given
probability p, click the radio button "Inverse
cumulative probability".
Enter the mean and standard deviation.
Click the button "Input constant" and enter the
probability p in the field.
Click OK.
The answer x appears in the session window.
To produce a normal quantile plot, go to Graph > Probability Plot. Select whether your data are in a single column (for a single plot) or if there are several data columns and press OK.
Enter the name of the column(s) in the dialog window and press OK again.
The result is a normal quantile plot, flipped about the diagonal: The normal probabilities (probabilities for z-scores) are plotted against the data. The plot also has a straight line and error boundaries for comparison.
Alternatively, go to Stat > Basic Statistics > Normality Test.Enter the name of the column with the data, leave everything else at the default setting, and press OK.
The result is the same plot, with slightly changed scales and without error boundaries.
Go to Graph > Scatterplot.
Choose the type of scatterplot you wish to make.
In the Graph dialog box, enter the
column number or the name of the explanatory variable in the X
box and the column number of name of the response variable in the
Y box.
Leave all other options at their default setting and click OK.
To obtain the correlation coefficient between two quantitative
variables without using the Scatterplot instruction,
go to Stat > Basic Statistics > Correlation.
Under Variables, select the two columns of data and click
on OK.
Go to Stat > Regression > Regression.
In the Dialog Box, enter your response
variable (y-axis) in the box marked Response and your
explanatory variable (x-axis) in the box marked Predictors.
Click OK to read off the equation for the regression line in the
session window.
Note that in the Predictors: box there is room for more than one explanatory variable. We can enter more than one explanatory variables to do multiple regression.
While you are still in the Regression Dialog Box and before
you move on to obtain the regression analysis, you can store and
plot the residuals:
Click on Storage > Residuals and click on OK.
Back in the Regression Dialog Box, click on Graphs.
In the ensuing dialog box (Regression - Graphs), enter your
explanatory variable in the large box at the bottom: Residuals
versus the variables and click on OK
At this point Minitab will compute the regression analysis and draw the residual plot. The regression analysis will appear in the session window and the residual plot will appear immediately thereafter as a separate graph. The residuals will be stored in a new column called RESI1 in the worksheet.
This approach gives the least squares regression equation, the coefficients a and b, the coefficient of determination R2, and information on unusual observations (outliers) and influential observations, but not the regression line.
To obtain a graph with the least squares regression line
fitted to the scatterplot:
Go to Stat > Regression > Fitted Line Plot.
In the dialog box, select the response and explanatory
variables in the appropriate boxes.
If you haven't calculated the residuals yet, go to Storage
> Residuals and click OK.
Click OK to obtain the scatterplot with the regression line.
The regression analysis will appear in the session window. You will obtain the regression equation, the coefficients a and b, and the coefficient of determination. After this information appears in the session window, Minitab will draw the scatterplot with the least squares regression line, the regression equation, and R2.
If you want to see the Residual Plot at this time, activate
the data window by clicking on it and click on the Edit Last
Dialog icon, to bring back the Fitted Line Plot dialog
box.
Enter the residual column RESI1 as the Response Variable.
Leave the Predictor box unchanged.
Click on OK
You will get a graph of the residuals plotted against the
explanatory variable, i.e. a residual plot.
To conduct tests of the null hypotheses that the slope or the constant term are zero, go to Stat > Regression > Regression and carry out the regression as above. In the session window, you can read off the p-values for these tests.
Go to Graph > Time Series Plot.
Chose the type of graph you want to produce and click OK.
Select your variable(s) in the Series window. Mtb will
assume that the index of the data is the time variable. You can
change this in the graph dialog window by clicking on Time/Scale.
The Data View button allows you to change the appearance
of the time plot and to make a smoothed scatterplot.
If your two-way table is in a single column, with classification variables for the table rows and columns in two other columns:
Go to Stat > Tables > Cross Tabulation and
Chi-Square.
Select the two variables for rows and columns.in the Classification
variables window. Enter the column containing the
frequencies.
Check Count, Row Percent, etc., as you desire and click on
OK. The results will be displayed in the session window.
To do a c2 test for independence for data in this form, click on Chi -Square in the dialog window and check Chi-Square analysis as well as any other details you wish to display. Then click OK.
If your two-way table is is already stored in table form in the worksheet:
Go to Stat > Tables > Chi-Square test (Table in Worksheet).
Enter the columns that contain the table and click on OK.Select the two
variables for rows and columns.in the Classification variables window.
The results of the Chi-Square test for independence will be displayed in the
session window (degrees of freedom, expected cell counts, cell contributions to
c2, overall value of
c2, p-value).
To compute row percents and column percents, stack the columns of the
table with Data > Stack > Columns to turn the table into a single column. Make
sure there are row and column subscripts in two separate columns (Minitab will
make the column subscripts automatically). Then proceed as above.
Go to Data > Copy > Columns to Columns.
Enter the columns to be copied in and their destination.
Go to Data > Copy > Columns to Columns.
After entering the columns to be copied and the destination in
the dialog window, click on Subset and enter the rows to be used
or to be excluded.
Go to Data > Stack and choose Columns or Blocks
of Columns or Rows..
Select the columns you want to stack and their destination
columns in the dialog box.
You can keep track of the origin of the stacked columns by storing
subscripts in another column.
This is the same as "separating column data according to categories or characteristics".
Go to Data > Unstack Columns.
In the window Unstack the data in select the column you
want to separate.
In the Using subscripts in window, enter the column
containing the category or characteristic according to which you
want to separate the data.
Enter where you want to store the unstacked data, either in a new
worksheet or after the last column in use.
This can be done by first unstacking a block of columns by category and then deleting all columns that are not needed.
To draw a simple random sample from data stored in a column,
go to Calc > Random Data > Sample from Columns.
Select the column you want to sample from, enter the size of your
sample (Sample ... rows from column) and enter the column
where the samples are to be stored.
Choose whether you want to sample with or without replacement
(check box) and click OK.
To make a list of random digits, go to Calc >
Random Data > Integer.
Enter how many (groups of) random digits you need in
"Generate ... rows of digits".
Enter the column where the digits are to be stored.
To get single digits, choose Minimum value = 0 and Maximum value
= 9. To get two-digit groups, choose Minimum value =
0 and Maximum value = 99. Of course, other choices are also
possible.
Click OK to generate the random digits.
Work with the digits just like you would work with random digits
found in a table.
To simulate a series of n Bernoulli trials (outcomes are either 1 , "success", with probability p or 0, "failure", with probability 1-p), go to Calc > Random Data > Bernoulli. Enter the success probability p and the number n of trials ("Generate n rows of data"). Then enter the column where the data are to be stored. To simulate several series of n trials each, enter several column locations, one for each series.
To make a plot of a density curve, start a fresh worksheet (File > New > Worksheet). Let's say you want to make a density curve for a < x < b.
To compute an individual binomial probability of k
successes in n trials for a success probability
of p,
Click on Calc > Probability Distributions >
Binomial,
click the radio button for Probability,
enter the number of trials and the probability of success,
click the radio button for Input Constant and enter the
number of successes k,
and click Enter.
The probability of exactly k successes will appear in the session window.
To compute the binomial probability of k or fewer
successes in n trials for a success probability
of p,
Click on Calc > Probability Distributions >
Binomial,
click the radio button for Cumulative Probability,
enter the number of trials and the probability of success,
click the radio button for Input Constant and enter the
number of successes,
and click Enter.
The probability of k or fewer successes will appear in the session window.
To make a binomial probability table for n trials
with a success probability of p , first go to Calc >
Patterned Data > Simple Set of Numbers,
store the patterned data in C1 from first value of
0 to last value of n, in steps of 1.
Then go to Calc > Probability Distributions > Binomial,
click the radio button for Probability,
enter the number of trials and the probability of success,
click the radio button for Input column and enter C1,
enter C2 in Optional Storage, and click Enter.
Go to Stat > Basic Statistics > 1 Sample Z....
Select the variable for whose mean you want to find a confidence
interval in the dialog window.
Enter the standard deviation.
Click on Options, enter the Confidenc level (in
percent), and enter and leave the Alternative on
not equal.
Click OK twice and read off the result in the Session window. You
can choose Graphs in the dialog box to illustrate the
results.
Go to Stat > Basic Statistics > 1 Sample t.
Select the variable for whose mean you want to find a confidence
interval.
Click on Options, enter the Confidenc
level (in percent), and enter and leave the Alternative
on not equal.
Click OK twice and read off the result in the Session window. You
can choose Graphs in the dialog box to illustrate the
results.
Go to Stat > Basic Statistics > 1 Sample Z.
Select the variable for whose mean you want to conduct the test
and enter the standard deviation.
Enter the value for the null hypothesis in the Test mean field
and click on Options to choose the Alternative
Click OK twice. The results appear in the Session window. You can
choose Graphs in the dialog box to illustrate the results.
Go to Stat > Basic Statistics > 1 Sample t.
Select the variable for whose mean you want to conduct the test..
Enter the value for the null hypothesis in the Test mean field
and click on Options to choose the Alternative
Click OK twice. The results appear in the Session window. You can
choose Graphs in the dialog box to illustrate the results.
Go to Stat > Basic Statistics > Paired t.
Select the two variables for whose means you want to conduct the
test..
Click OK. Mtb will conduct a test of the null hypothesis of equal
means against the alternative that the means are not equal. It
will also produce a 95% confidence interval for the difference of
the means. To change any of these settings, go to Options.
The results (95% confidence interval and p-value for the
alternative of unequal means) appear in the Session window. You
can choose Graphs in the dialog box to illustrate the
results.
Go to Stat > Basic Statistics > 2 Sample t.
If all samples are in one column, click the top radio button and
select the variables that contain the data and the subscripts.
If the samples are in two different columns, click the middle radio button and select the columns that contain the samples.
Summarized data can also be entered (lower radio button).
Click on Options to select the alternative
and to choose the confidence level for a confidence
interval. Mtb assumes that the variances are unequal.
Click OK.
The results appear in the Session window, including a p-value and
an estimate for the number of degrees of freedom. You can choose Graphs
in the dialog box to illustrate the results.
Go to Stat > Nonparametrics > 1-Sample Sign Test.
Enter or choose the variable in the dialog window.
Click the Test median radio button and enter the
value for the null hypothesis.
Click OK. The result (p-value) appears in the session window.
Go to Stat > Nonparametrics > 1-Sample Wilcoxon.
Enter or choose the variable in the dialog window.
Click the Test median radio button and enter the
value for the null hypothesis.
Click OK. The result (p-value and Wilcoxon statistic) appears in
the session window.
Go to Stat > Basic Statistics > 1 Proportion.
If the data (1 for success, 0 for failure) are in a data column,
click the top radio button and enter the name of that column in
the top box of the dialog window.
If you have summarized data (number of trials and number of
successes), click the bottom radio button and enter the values.
Click Options to choose
Click OK twice to see the results in the session.
Go to Stat > Basic Statistics > 2 Proportions.
If the data (1 for success, 0 for failure) are in one data column
with subscripts in another data column, click the top radio
button and enter the name of the columns containing the samples
and the subscripts.
If the data (1 for success, 0 for failure) are in two data
columns, click the middle radio button and enter the name of the
columns containing the samples.
If you have summarized data (number of trials and number of
successes), click the bottom radio button and enter the values
for the two samples.
Click Options to choose
Click OK twice to see the results.