<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=windows-1252">
<META NAME="Generator" CONTENT="Microsoft Word 97">
<TITLE>Welcome to the Practical Statistics Add-in</TITLE>
<META NAME="Template" CONTENT="C:\PROGRAM FILES\MICROSOFT OFFICE\OFFICE\html.dot">
</HEAD>
<BODY LINK="#0000ff" VLINK="#800080">
<H1>Welcome to the StatPro Add-in!</H1>
<P>The purpose of the StatPro add-in is to provide you with some powerful statistical capabilities in an environment you're used to-the environment of Excel. StatPro contains a number of procedures, such as creating a scatterplot, testing a variable for normality, and running a regression. Each of these procedures can be run from the StatPro menu that is added to the Excel menubar when you load the add-in. Most of the procedures provide some explanations as you run them. In addition, each gives you the option of reading help screens specific to that procedure. Finally, the help screen you're reading now provides an overall guide to the procedures.</P>
<H2>Contents of this Help File</H2>
<UL>
<LI><A HREF="#Goals">Goals of StatPro</A> </LI>
<LI><A HREF="#DataSetup">StatPro dataset requirements: extremely important!!!</A> </LI>
<LI><A HREF="#Conventions">StatPro conventions</A> </LI>
<UL>
<LI><A HREF="#ChartPlacement">Placement of charts</A> </LI>
<LI><A HREF="#NumericalPlacement">Placement of numerical output</A> </LI>
<LI><A HREF="#VariablePlacement">Placement of new variables</A> </LI>
<LI><A HREF="#Missing">Treatment of missing values</A> </LI>
<LI><A HREF="#Character">Treatment of character variables</A> </LI>
<LI><A HREF="#Formulas">Use of formulas versus values</A> </LI>
<LI><A HREF="#NamedRanges">Named ranges and hidden sheets</A> </LI>
<LI><A HREF="#Comments">Use of comments in cells</A></LI></UL>
<LI><A HREF="#Summary">Summary of procedures</A> </LI>
<UL>
<LI><A HREF="#SummStats">Summary Stats</A> </LI>
<UL>
<LI><A HREF="#OneVar">One-variable summary stats</A> </LI>
<LI><A HREF="#Corr">Correlation, covariances</A> </LI>
<LI><A HREF="#Auto">Autocorrelations</A></LI></UL>
<LI><A HREF="#Infer">Statistical inference</A> </LI>
<UL>
<LI><A HREF="#OneSamp">One-sample analysis</A> </LI>
<LI><A HREF="#TwoSamp">Two-sample analysis</A> </LI>
<LI><A HREF="#PairSamp">Paired-sample analysis</A> </LI>
<LI><A HREF="#SampSize">Sample size selection</A> </LI>
<LI><A HREF="#OneWay">One-way ANOVA</A> </LI>
<LI><A HREF="#TwoWay">Two-way ANOVA</A> </LI>
<LI><A HREF="#ChiSq">Chi-square test for independence</A> </LI>
<LI><A HREF="#Runs">Runs test for randomness</A> </LI>
<LI><A HREF="#RandSamp">Generating random samples</A></LI></UL>
<LI><A HREF="#Normality">Tests for normality</A> </LI>
<UL>
<LI><A HREF="#ChiNorm">Chi-square test for normality</A> </LI>
<LI><A HREF="#Lill">Lilliefors test for normality</A> </LI>
<LI><A HREF="#QQ">Q-Q normal plots</A></LI></UL>
<LI><A HREF="#Regr">Regression analysis</A> </LI>
<UL>
<LI><A HREF="#Simp">Simple</A> </LI>
<LI><A HREF="#Mult">Multiple</A> </LI>
<LI><A HREF="#Step">Stepwise</A> </LI>
<LI><A HREF="#Forw">Forward</A> </LI>
<LI><A HREF="#Back">Backward</A> </LI>
<LI><A HREF="#Block">Block</A></LI></UL>
<LI><A HREF="#Forecast">Forecasting</A> </LI>
<LI><A HREF="#Quality">Quality control charts</A> </LI>
<UL>
<LI><A HREF="#XBR">X-Bar, R charts</A> </LI>
<LI><A HREF="#PCht">P charts</A> </LI>
<LI><A HREF="#CCht">C charts</A> </LI>
<LI><A HREF="#UCht">U charts</A></LI></UL>
<LI><A HREF="#Utilities">Data utilities</A> </LI>
<UL>
<LI><A HREF="#Stack">Stack variables</A> </LI>
<LI><A HREF="#Unstack">Unstack variables</A> </LI>
<LI><A HREF="#Dummy">Create dummy variables</A> </LI>
<LI><A HREF="#Interact">Create interaction variables</A> </LI>
<LI><A HREF="#Transform">Transform variables</A> </LI>
<LI><A HREF="#Lag">Create lagged variables</A> </LI>
<LI><A HREF="#Diff">Create difference variables</A></LI></UL>
<LI><A HREF="#Charts">Charts</A> </LI>
<UL>
<LI><A HREF="#Hist">Histograms</A> </LI>
<LI><A HREF="#Scat">Scatterplots</A> </LI>
<LI><A HREF="#Box">Boxplots</A> </LI>
<LI><A HREF="#TSPlot">Time series plots</A></LI></UL>
</UL>
</UL>
<H2><A NAME="Goals"></A>Goals of StatPro</H2>
<P>We have kept several goals in mind while developing this add-in. First, we want it to be easy to use. Second, we want it to "look" like Excel-we want you to be comfortable with the interface. Third, we want it to be "natural" for statistical use. Finally, we want it to be a reasonably powerful tool, especially for learning statistics in a course environment.</P>
<H2><A NAME="DataSetup"></A>StatPro Dataset Requirements: Extremely Important !!!</H2>
<P>We now expand on the third goal. There are a number of other statistical Excel add-ins available, including the Data Analysis ToolPak that ships with Excel. (If you installed it, you'll find it under the Tools menu.) All of the add-ins we've seen are structured around <I>ranges</I>, where you are asked to select one or more ranges. You typically select these by pointing and dragging with your mouse, and it is usually not very convenient. We have taken our cue, however, from non-Excel statistical packages, which are almost all structured around <I>variables</I>. In these packages you typically select a procedure and then select one or more variables to use in that procedure. This is not only an easier interface to work with, but it makes more sense from a statistical point of view.</P>
<P>Therefore, we impose some structure on users, much like the structure that is automatically imposed on all users of statistical packages. Except in two instances, we always require that you always work with a "data set," by which we mean a set of contiguous columns that correspond to statistical variables, with variable names in the first row of the data set. You should imagine such a data set as a range that is surrounded by blank rows and columns on all sides (or possibly by the column labels and row labels on the left and the top). In other words, you shouldn't let extraneous stuff bump up next to your data set. This notion of an "isolated" data set range is crucial to the way StatPro works.</P>
<P>There are two other notes about the structure of your data sets. First, the variable names in the top row must adhere to Excel's naming rules for ranges. Basically, this means that names should start with a letter, they shouldn't use characters other than letters, numbers, and the underscore (_) character (although we allow spaces and periods), they shouldn't be the single letter R or C, and they shouldn't look like cell addresses (such as X1). Obviously, it's best to use names that are meaningful, such as Revenue or Variable_Cost. Second, the lengths of the columns in your data set do not have to be equal. For example, you could have two variables, Weight_Men and Weight_Women, with different numbers of observations. If this is the case, however, the add-in will treat the blank cells in the shorter columns as missing data, and it will enter asterisks in these cells.</P>
<P>How do we make use of this structure? It's very simple. Each time you run a procedure, you must first place the cursor anywhere within the data set. (You're always given a second chance to do so if you forget.) Then the procedure looks to the first row of the data set, reads the variable names, names each column with that variable name, fills in blank cells with asterisks, and then accomplishes its tasks. If you don't follow the structure explained in the previous two paragraphs, StatPro will very likely give you an error message or produce unanticipated results.</P>
<P>Why do we impose this structure? In our opinion, one of the main strengths of Excel (or any other spreadsheet package) is also its main weakness when it comes to statistical tasks: It is too flexible! You can put anything anywhere and get away with it. It appears to us that developers of statistical add-ins