Introduction to MINITAB in the Saint Mary's Microcomputer Lab

V. Some descriptive statistics

A: One-variable description

Most complete description (pretty much all the one-variable description you need)

For the most commonly used descriptive measures, select Stat>Basic Statistics>Descriptive statistics. With the cursor in the "Variables" box, select the columns you want described (highlight them and click on " Select " - or simply double-click on them) and click on OK (or press the Return/Enter key).

The printout gives, for each variable:
Variable the name of the variable (unless you didn't name it - then you get the column number)
N the number of values in the column
median the "middle" value (half the values are at or above, half are at or below)
mean the arithmetic mean (x-bar) of the values
TrMean the 5% trimmed mean - the mean with the largest 5% and smallest 5% of the values deleted
StDev the standard deviation (s) of the values
SEMean the standard error of the mean - the standard deviation divided by the square root of N
Min the smallest value in the list
Max the largest value in the list
Q1 the first quartile - value separating the smallest quarter of the values from the rest
Q3 the third quartile - value separating the smallest three-quarters of the values from the rest

Other descriptions

Grouped frequency distribution : Use Graph>Character Graphs>Histogram. Select the variable(s) by double-clicking in the list. The Interval is the class width (class interval, range for the class), the First midpoint is the class midpoint of the first class. The resulting printout shows class midpoints (no class limits) and frequencies. [For an actual histogram, which will appear in a separate graph window, use the Graph>Histogram command - see IV. Graphs for details.]

Stemplot: Use Graph>Stem-and-Leaf . Select the variable(s) by double-clicking in the list at the left of the window. You can set the increment at 5 to get a split stemplot. In the display, an extra column at the left of the display shows the "depth" of the class - the distance (number of items) from the nearer end (large or small) of the data. The row with the median is marked in parentheses.This is really a character graph - it shows up in the Session window - it's on the main graph menu because people use it a lot.


Listing the data in the session window is accomplished with the File>Display data command (select the column or columns you want displayed) - useful if you want a complete list of the values of one variable without printing the whole data window.


B: Two-variable description

Correlation coefficient (Pearson's r): Use Stat>Basic Statistics>Correlation.... Select the variables for which you want the correlation. If you select more than two variables, you will geat a table giving the correlations between all pairs of variables.

Regression line and coefficient of determination (r-square): Use Stat>Regression>Regression. For the "response" select the variable to be the "y" variable (predicted by the eauation) and for the "predictor" use your "x" variable (used to predict the response). The result gives the regression equation and the coefficient of determination (r-squared - labeled "R-sq") and further information - notably pairs with large residuals (potential outliers) and values whose x-values are ver different (influential points) - used for analysis of the equation.
The command Stat>Regression>Fitted line plot will produce the same output in the session window plus a scatterplot with the rgression line graphed on it.

For graphs, see IV. Graphs


Last update 8/10/00

Maintained by