You do not need to print out any of the graphs that you generate for this activity. Just look at them on the screen and answer questions based off of them in the activity. There is one graph (the histogram) that you need to copy onto your paper.
This step can be skipped if you go to File / Open Worksheet and open the walk.mtw file the instructor created for you. This section is still good reading for you to learn about creating worksheets, though.
There is a little bit of setup that you need to do
before entering the data. Minitab can also help by creating some of the
values for you.
Once you have your data entered into Minitab, you may work with it. One of the most common things we will do is display the descriptive statistics.
Display Descriptive Statistics is the option in Minitab that we will use more than any other. As such, it's at the top of the menu choices. Whenever you see "describe" or "summarize" in the instructions, there's a good chance you should head for Stats / Basic Statistics / Display Descriptive Statistics.
This screen will give you the following statistics by default.
You can change the statistics that are given by clicking on the statistics button. In particular, the N* and SE Mean won't be used right now. The SE Mean will be used in later chapters, but the number of missing cases is rarely used. Other options in the statistics menu that we will use occasionally are the variance, range, interquartile range, and sum of squares.
You may describe more than one variable at a time. However, in this problem, we only have one variable, time, that we want to describe. The other two variables are categorical variables used for classification purposes only, it would make no sense to describe them. Sample output from the descriptive statistics command is shown in the figure.
This is the way to describe the time for all of the teams and all of the heats.
You should get some output that looks something like this.
Descriptive Statistics: time Variable N Mean StDev Variance Minimum Q1 Median Q3 Maximum
time 42 5.949 0.987 0.975 4.040 5.130 6.025 6.493 8.600
This is the way to describe the time for
each of the heats. We use the "By Variable"
option to do this. The column used for the By Variable should be a
categorical variable such as the gender, race, age group (but not age as a
number), or heat number. There should be few categories for
this variable, do not use variables that have large numbers of unique values
for the By Variable. Do not use measurement variables (height, weight, age,
time) as the by variable.
A common mistake people make after doing this activity is that they always want to put something in the "by variable" box. Most of the time, you don't want to break you data down by categorical variable and the "by variable" should be left blank.
A histogram is a good way to look
the data and see where it lies. We can also use it to let Minitab count the
number in each group for us, rather than us having to do it manually.
Normally, we would let Minitab just automatically assign groups for us, but in this case, we're specifically looking for bars that are one standard deviation wide. That means that we're going to have to do some extra work that we wouldn't normally have to do.
For this example, let's assume that we're working with the same data we had earlier and that the mean is 5.949 and the standard deviation is 0.987.
Find the mean minus three times the standard deviation and the mean plus three times the standard deviation: 5.949 - 3(0.987) = 2.988 and 5.949 + 3(0.987) = 8.910. These numbers correspond to our lowest and highest class boundaries and will be used later. Notice that our data falls within this range since our minimum is 4.04 and our maximum is 8.60. If our data extended beyond ±3 standard deviations, then we would need to extend those ranges to fit all of the data.
The old version of Minitab (version 13) would allow you to set all kinds of options before you generated the graph. The new version (version 14) allows you to look at the graph and then play with the settings. There's arguments in favor of both directions, but for most people, the new way is probably better. The rest of this will involve changing the graph to give us what we want to have.
The graph to the right is the modified version that we're
interested in.
Since the file "walk.mtw" was already provided for you, you will probably be okay without saving a project file for this activity. There's not much work that wouldn't be easy to recreate if you messed up and had to go back.
For other projects, be sure to save your work! This allows you to go through and work on the activity incrementally (you can do part of it one day and finish it another day). All open windows are saved when you save the project, but if you close a graph, it won't be saved. When you are completely done with the project, you may wish to close the graphs. This will make the files smaller and keep us from running out of room on the drive.
Once you have this histogram created, this chart is easy to finish.
| % of values | w/in 1 st dev | w/in 2 st devs | w/in 3 st devs |
|---|---|---|---|
| Our Sample Data | 76.2% | 95.2% | 100% |
| Empirical Rule | approx 68% | approx 95% | approx 99.7% |
| Chebyshev's Rule | not applicable | at least 75% | at least 88.9% |
Hmmm, good question. The amount within 2 and 3
standard deviations is pretty close, but the amount within one standard
deviation is off by about 8%.
The whole 68-95-99.7 thing is approximate anyway, and it's meant to help you determine if the data is normally distributed. The normal model is talked about in chapter 5 and so we won't dwell on it too much right now, but there are other things we can look at, of course, we just don't know about them yet. So, if you take the time to read this section, you'll be ahead of the game for later.
To the right is a graphical summary (Stats / Basic Statistics / Graphical Summary) for the time variable.
Here are some things to help you decide if the data is normally distributed.
The Anderson Darling Normality Test is designed to test the claim that the data comes from a population that is normally distributed (unimodal, symmetric, bell-shaped, no outliers). The p-value is the likelihood of getting the results we got if the data has a normal distribution. Typically, as long as the p-value is more than 5% (0.05), there's no reason to question the normal distribution. In this case, the p-value is 0.246, so we'll retain the claim that the dat is normally distributed.
The skewness is a measure of the symmetry of a graph. A symmetric graph has a skewness of 0, here the skewness is 0.528277, so it's a little skewed to the right of the mean.
The kurtosis is a measure of the peakedness of a curve. A normal curve has a kurtosis of 0, a negative value means it is flatter than a normal model and a value greater than 0 means it is sharper than a normal model. The kurtosis for our data is 0.904122, which indicates that our data may be more peaked than what we would like.
Notice I didn't really answer the question. That's because in statistics, there isn't always a definite answer. We deal a lot with levels of gray. Later, we'll make an assumption that our data comes from a normal population and base our decisions on that assumption. All in all, our data is close enough to not cause problems with that assumption.
The Analysis of Variance is used to test whether or not three or more means are the same. The response variable is the variable whose value you're testing and the factor variable is a categorical variable used to define the groups.
The sample data that I used to create the graphs above is from the Spring 2005 semester. There were two sections of Math 113 and there ended up being 14 teams of 3 people each. The data is available if you would like to look at it or use it.
You can copy and paste the data into Minitab. If it asks you whether to use spaces as delimiters or paste it as a single column, use the spaces as delimiters. That works since there are no spaces in the data.
If you have spaces in the data, then you can copy from Internet Explorer into Microsoft Excel, then copy from Excel and paste it into Minitab.
If you are using Firefox instead of Internet Explorer, then you can copy directly from the web page to Minitab without any problems or additional questions.