Information & documentation

M583: SPSS® 11.5 (Statistical Software Package) on the PWF PCs

Introduction

This leaflet is designed to introduce briefly the statistical package SPSS 11.5 for Windows, as implemented on the PCs of the Computing Service's PWF network, and to tell you where to find other sources of information that will help you use the software. (It is not intended to serve as a guide to statistics and data analysis using SPSS.)

A basic knowledge of the PWF and its Windows 2000 operating system is assumed. Those without this knowledge should consult the following web page: http://www.cam.ac.uk/cs/pwf/.

SPSS is a large and powerful software package offering a wide range of statistical techniques coupled with excellent data management facilities and high quality graphics. It is applicable to many different research fields, though it has particular strengths in the areas of social sciences. It is often described as one of the 'friendliest' statistical packages of its type, with a good graphical user interface and plenty of built-in documentation. Advanced features include powerful data transformation, matrix algebra, macro and scripting facilities. There is no built-in limit on the number of variables and cases forming the data (beyond what the system can handle). It can also produce output in HTML and JPEG format for use in web sites.

Besides being available on the PWF, SPSS for Windows is obtainable under site licence from the University Computing Service.

Although SPSS is marketed commercially in modular form (SPSS Base and a number of extra-cost options such as Tables, Trends and Regression Models), these extra modules are included in the university's site licence; they are installed on the PWF and work seamlessly together. The only significance this structure is likely to have for most Cambridge University users, therefore, is that it is reflected in the organization of the SPSS manuals.

1. A practical session with SPSS on the PWF

1.1 Before you start

There are some important differences between using SPSS on the PWF network and using it on a machine fully under your own control. On the PWF special measures are in place to ensure that users' preferences follow them from workstation to workstation using the 'roaming profiles' mechanism, to provide access to a very large number of software packages without making the system impossibly slow, and to minimise the risk of damage (accidental or otherwise).

• In every PWF session, the first time you use SPSS it must be launched from either the Start menu or the PWF Programs folder on the desktop, as described below, and not from an SPSS application file (document) containing data, output, etc. This ensures that the correct setup information is copied to the workstation. If you start SPSS from an application file, you bypass the Novell Applications Launcher program (NAL), which manages the distribution of software to workstations. This can result in SPSS being incorrectly configured on the workstation. So you must start SPSS as described above, even if you close it down straight away and relaunch from an SPSS document (which you may wish to do in order to work in a particular directory).
• Certain SPSS features depend on Microsoft technologies that are not available or are not fully implemented on the PWF, notably ActiveX, and so are themselves not available.

1.2 Login to the PWF and copy the SPSS course data files to your own filespace

Double click on the My Computer icon on the Windows desktop, then on Ux on 'PWF-central' (I:), then on Lessons. You will see a folder icon labelled SPSS Leaflet M583 Data. (be careful to choose the correct folder as there are several with rather similar names). Drag-and-drop this onto the desktop icon for your own filespace Userid on 'Pwf-... (also referred to as drive F:). You will now have your own copy of this folder.

DATA SET: EMPLOYEES

In the United States, a lawsuit alleging discrimination was taken out against a Midwestern bank. The dataset is taken from the evidence used in the case, and contains information about 474 of the employees. Besides age, sex, ethnicity ('minority') and job category, it includes beginning salary, salary as of March 1977, years of education, years of work experience, etc.

1.3 Starting SPSS 11.5

Click on the button at the bottom left hand corner of the screen. This opens the Start menu. PWF Programs appears at the top. If you click on this and then on Spreadsheets Charts and Statistics, followed by SPSS, you will find SPSS 11.5. You can follow the same route through the folders on the desktop, if you prefer. Ignore SPSS 11.5 Production, which is a facility suitable only for advanced users. Double click on SPSS 11.5 to launch SPSS. You will briefly see a banner, then the SPSS "Launch" dialog box.

In this dialog box you are offered a range of choices. Double-click on More Files from Open an existing data source. In the Open File dialog box, click on the scroll button for the Look in list. Navigate to your filespace Userid on 'Pwf-... (also referred to as drive F:) and continue to the SPSS Leaflet M583 Data folder and the Employee data.sav data file. Double-click on this (or select it and then use the Open button).

You will be left in the Data Editor window, which displays the contents of the Employees data file. Take a good look at its contents.

At this stage there is no window for viewing output, but one will be started up as soon as it is needed. A simple SPSS session need use only two windows, one for data and one for output, though it is possible to work with many more than that.

1.4 The SPSS Data Editor window

This has the usual Microsoft Windows features (a control-menu box in the top left corner, minimize, maximize and close buttons in the top right corner, a title bar at the top, and scroll bars to the right and bottom).

From top to bottom you will normally see the following features (though it is possible to change the layout of the Data Editor window to some extent, e.g. by turning off the toolbar display).

• The menu bar, providing most of the facilities available in SPSS.
• The tool bar, giving easy access to frequently used features.
• The location (row and column) and contents of the current cell, if any.
• The data displayed in a spreadsheet-like layout.
• The status bar, telling you what (if anything) SPSS is doing. If SPSS is idle it will contain the information SPSS for Windows Processor is ready.

1.4.1 The Menu Bar

The menu bar contains menus with all the features needed to manage data and other files, to perform a wide range of statistical analysis, and to produce graphs (SPSS documentation refers to them as charts). It also provides editing, control over fonts and other aspects of the appearance of SPSS windows, and customisation of the way SPSS behaves.

1.4.2 The Tool Bar

The tool bar provides instant access to many useful features. On the left-hand side a standard set of tools is always available. The tools on the right are those appropriate to the active window (so they change when you change windows).

To find out what a tool does, use the mouse to place the cursor over the tool button without clicking. The tool description will then appear at the left-hand end of the status bar (described next).

One of the most useful tools is (Dialog recall). This allows you to repeat an earlier operation (either exactly as before, or with modifications) by reopening the appropriate dialog box in its previous state.

1.4.3 Data View

Below the 'spreadsheet' you will see two tabs. The one currently selected is labelled Data View, indicating that the data values are displayed (the other, Variable View, shows how SPSS has been set up to handle the data, and will be discussed later).

In this example there are ten columns of data (known in SPSS as variables), headed id, gender, bdate, etc., though not all are visible in the screenshot shown earlier. These headings are the variable names which SPSS uses to identify the data columns, and are limited to 8 characters (more information about naming variables is given in section 2). Scrolling down, we can see that in this data file there are 474 rows (cases, in SPSS terminology).

Place the cursor over one of the column headings (if the column contains data the heading consists of a variable name). In a second or two a yellow box will appear displaying the corresponding variable label. Because SPSS variable names are restricted to a maximum of 8 characters it is common practice to supply longer and more informative labels for use in output tables, etc.

In the Employees data set, you will see that the variables jobcat and minority, which are categorical, have been coded numerically rather than being entered as text. Gender is represented by a single character ('f' or 'm'). Text data (especially if it consists of more than 8 characters) is unsuitable for many of SPSS's statistical operations, and of course the use of short codes speeds up the data entry process. However, SPSS has a very useful design feature: you can set up value labels for a variable. These make output tables and graphs more readable, and can also be displayed in Data View. From the menu bar, click once on View. In the displayed menu, you will see Value Labels (currently without a tick against it). Single-click on Value Labels.

The appearance of the data changes. Wherever applicable, instead of values the corresponding value labels are displayed. To reverse this, repeat the actions above (there is currently a tick against Value Labels in the View menu, indicating that the option has been selected previously).

1.4.4 The Status Bar

At the bottom of the Data Editor window, there is a grey status bar which provides a range of useful information about what is going on at any time. At startup, the status bar says Starting SPSS for Windows Processor but changes very quickly to SPSS for Windows Processor is ready. Other messages are displayed when appropriate. For example, if SPSS is currently using only a subset of cases (whether selected according to criteria specified by the user or because a random sample has been requested) the message Filter on is displayed.

1.5 Performing a simple analysis: a frequency table

To calculate simple frequency tables from the working data file, click on Analyze in the menu bar, point to Descriptive Statistics and then click on Frequencies... in the resulting submenus (in the rest of this document such a sequence will usually be represented as Analyze => Descriptive Statistics => Frequencies...). This opens the Frequencies dialog box. Its left-hand side lists variables available for use in frequency tables (though not all of these are really suitable for the purpose, as explained later).

Click on the Employment Category entry and then on the button. This moves Employment Category into the box labelled Variable(s), indicating that it has been selected for inclusion in the analysis. Clicking on the OK button causes a frequency table to be constructed and written to the Output Viewer window.

It is possible to obtain frequency tables for more than one variable in a single operation; e.g. if you click on Minority Classification you can transfer it to the selected variables list. Note how the transfer button changes direction when a variable is highlighted in the selected variables list, so the button can then be used to deselect it.. If you make a mistake in choosing a variable, select it in the Variable(s) box and click again on the same button (the arrow will have changed direction, pointing back to the source list).

The various buttons at the bottom of the Frequencies dialog box also allow you to request additional output, etc.

Note that you can obtain information about the meaning of the various buttons and other features of a dialog box by clicking on its Help button.

1.6 The SPSS Output Viewer

SPSS sends its results to the SPSS Output Viewer window. This is a very powerful tool for manipulating and modifying SPSS output. However, for the moment you need concern yourself only with its simpler features.

The Output Viewer window contains two panes (see the illustration in the previous section). The output objects (tables, graphs, etc.), are displayed in the right-hand (output) pane. In the left-hand (outline) pane the individual output objects are represented by icons grouped together in a tree-like layout. The groups are labelled with the name of the SPSS operation that produced them (and indicated by a yellow icon). There is also a yellow icon at the top, representing the entire contents of the window.

The outline pane is used for navigating through the output, and provides a convenient means of selecting, deleting and otherwise managing it. Double-clicking on the icon for a single output object either shows or hides it (the icon appears as an open or closed book, accordingly). Double-clicking on one of the yellow icons representing grouped output collapses or expands the corresponding 'branch'. A collapsed branch is flagged with a plus sign '+' in a box, and an expanded one by a minus '-'.

In the right-hand (output) pane, double-clicking anywhere on an output object opens it for editing. In either pane, you can select an object (e.g. for printing) by single-clicking on it.

The Output Viewer is described in detail in the SPSS Tutorial, under the heading Working with output. For more information on this very extensive subject see Help => Topics => Contents and look under Output management.

1.7 Summarising data: Descriptive statistics (means, standard deviations)

Frequency tables are useful for variables which are essentially categorical or for which the values are at fixed intervals, and preferably not too numerous. However, for numeric data measured at irregular intervals on a scale it is more useful to describe their distribution in terms of the mean (their 'typical' value) and standard deviation (their 'spread' about the mean).

To do this, use Analyze => Descriptive Statistics => Descriptives... . This opens the Descriptives dialog box (which is very similar to that for frequency tables). Select and transfer Current salary to the Variable(s) box, then do the same for Previous Experience, and click on OK.

The Options button provides a choice of a range of statistics (mean, sum, standard deviation, minimum and maximum, etc.). There is also a Save standardised values as variables button which can be used to generate Z scores (otherwise known as standard scores, in which values are expressed as the deviation from the sample mean, measured in standard deviation units).

1.8 Entering data values in the Data Editor window

THE 'TASTE' DATA SET

In testing food products for palatability, General Foods employed a 7-point scale from -3 (terrible to +3 (excellent) with 0 representing "average". Their standard method for testing palatability was to conduct a taste test with 50 persons - 25 men and 25 women. The experiment reported here involved the effects on palatability of a course versus fine screen and of a low versus high concentration of a liquid component. Four groups of 50 consumers each were recruited from local churches and club groups. Persons were assigned randomly to the four treatment groups as they were recruited. The experiment was replicated four times, so that there were 16 groups of 50 consumers each in the entire experiment.

Empty the Data Editor window by selecting File => New => Data... from the menu bar. (You could have obtained a similarly empty window when launching SPSS, if you had chosen Type in Data in SPSS's opening dialog box instead of picking an existing data file, or if you had clicked on the Cancel button.)

The appearance of the empty Data Editor window is shown below. The cells are all empty, the case numbers and column headings are greyed out, and every column has the heading var. The cell at the top left corner is highlighted, ready to receive a data value. You can type data directly into the empty spreadsheet, either before or after defining the variables in detail. In this example the values will be entered first.

Unless SPSS has previously been told otherwise in Variable View (described later in this section), it decides whether a variable is to be defined as numeric or as non-numeric according to the first value inserted into a column. If you try to enter a non-numeric value in a column which has been defined as numeric, SPSS will just beep at you to warn you of the mistake, and not accept the value. It is not permissible to mix different types of data in a single column. See below for more information about SPSS variables. (In earlier versions of SPSS it was always necessary to define string variables explicitly.) If you have anything other than simple numbers or text (i.e. dates, times, amounts in dollars, etc.) these must be defined explicitly beforehand.

For the moment you need only concern yourself with simple numeric and text (also known as alphanumeric or string) variables. String variables with 8 characters or less can be used in some statistical procedures; longer ones not at all.

1.8.1 Entering some data values

The data to be entered is as follows: score, sieve and liq are the variable names.

Use the Tab key after each value in a row to move to the next cell, and the Enter key to start a new row. You can also use the arrow keys to move around the spreadsheet. As you enter data in each column of an empty spreadsheet without predefined variables, you will see the column headings (the variable names) change to var00001, var00002, etc. As you enter new cases (rows) you will see that the corresponding case numbers (row numbers) are no longer greyed out.

To alter an existing data value, select the cell and edit in the usual manner (use the arrow keys to move around the spreadsheet).

1.8.2 Rules for naming variables

Make sure you learn these rules about SPSS variable names as soon as possible:

• Variable names may each be used only once in an SPSS data file (otherwise SPSS would not know which variable you intended when you used the name).
• They may consist of no more than eight characters.
• They must normally begin with a letter or the @ (at) symbol. There are two exceptions. System variables created by SPSS itself start with $ (dollar) and scratch variables (special temporary variables which as a beginner you can ignore) start with # (hash).
• They may contain only letters, numbers, period '.', underline '*', dollar '$', or at '@'. Note especially that variable names may not contain spaces.
• They are not case-sensitive (it makes no difference whether the letters are uppercase or lowercase), so REFNO, refno, RefNo, and Refno may all be used for the same variable.
• The following may not be used as variable names, because they are also used as SPSS keywords and could cause confusion: ALL, AND, BY, EQ, GE, GT, LE, LT, NE, NOT, OR, TO, WITH.

Here are some examples of acceptable variable names:

 AGE  V1001  ht@10  Db.Press  A_93SEP 

The eight-character limit for names may seem an inconvenient restriction. However, more than adequate compensation is provided by SPSS's facility for using variable labels. Variable labels are optional (you do not have to declare them when creating a data file), but are extremely useful as they can provide a more complete description of each variable. Once they have been set up, as in the Road Construction Bids data set used earlier, you will see them appear if you hover the mouse over each column heading in turn. When SPSS produces statistical tables the variable labels improve their readability.

1.8.3 Types of Variable

• Numeric variables. All numeric values are treated as having decimal places, even if read in as integers. They are represented internally in floating point format, as n.nnnn... x 2nnn, so very large and very small numbers can be handled (to get an idea of the largest number that SPSS can cope with, take 1796 and add 305 zeros after it!). The precision used is 16 significant figures. The appearance of numeric values in the Data Editor and in output depends on the format specified for them, e.g. numeric with so many decimal places (or none), currency, etc. Date variables and time variables are simply numeric variables with special formats assigned to them. Some procedures display derived statistics in a fixed format which is 'hard-wired' and cannot be changed.
• String variables (containing text) may be up to 255 characters long. However, most SPSS statistical procedures cannot use string variables with more than 8 characters; such variables will not be listed in the dialog boxes for these procedures. For this reason a distinction is often made between short string and long string variables.

1.8.4 Defining or redefining variables

To provide your own variable names, labels, and so on, select any cell in the column of interest, and either double-click on it or select View => Variables from the menu bar. This displays the variable as definitions in Variable View, with the following columns.

To change the variable name, select the name and edit or replace it. Other columns offer drop-down lists and other devices to help you make the settings.

To change the number of decimal places to zero, proceed as follows. For each variable select the Decimals cell, then use the tool which appears at the right hand end of the cell to reduce the number from 2 to 0.

1.8.5 Adding labels to variables

To add variable labels and value labels to a variable, use the Label or Values cell in Variable View, as appropriate. For this data set, suitable labels are:

Typing in a new variable label in the Label cell is straightforward. However, in the case of value labels first select the Values cell, then click on the grey area at its right hand end. A dialog box is then opened.

You must enter both the data value and the corresponding label, then click on the Add button, to build a list of value/label pairs. To amend or remove value/label pairs, click on their entry in the list and use Change or Remove. When everything is to your liking, click on Continue, and then on OK.

1.8.6 Missing value declarations

If the coding for a variable includes values which are to be treated as missing data, you must declare these. To define user-missing value codes for a variable, you can use the Missing cell to open a dialog box (as in the example below). However, for this data set it is not necessary.

1.8.7 Levels of Measurement

SPSS uses another classification of variables, into levels of measurement. Variables may be declared through the Measure cell to be scale (numeric data on an interval or ratio scale), ordinal or nominal. Nominal datacan be either string or numeric . This classification is important when using some of the chart procedures, but may be ignored for the time being.

1.9 Summarising data: crosstabulating two or more categorical variables

Analyze => Descriptive Statistics => Crosstabs ... has a more complicated dialog box than those for frequency tables and simple descriptive statistics. It allows for up to three lists of variables, giving three dimensions: columns, rows, and layers (into which the rows are 'nested').

Frequently two dimensions are all that is wanted (i.e. rows and columns). and it is not necessary to put any variables into the third (layer) box. This is the case for the taste data as there are only two categorical variables (Sieve type and Liquid level), anyway. It is possible to add additional layers, for even more complicated nesting, but this is outside the scope of this introduction to the software.

The Statistics... button allows you to choose from a range of hypothesis tests (Chi-squared tests, etc.). The Exact... button provides alternative methods for calculating the significance of the statistics when the the amount of data is very small or there are many empty cells in the table. Use it only when strictly necessary, because with more than 30 cases the calculations may take a long time. The Cells... button determines whether the cells are to contain a simple frequency count (the default) or percentages and other values, as illustrated below. The Format... button for this procedure only allows you to choose which way the rows are sorted.

1.10 Summarising data: obtaining group means, etc.

Analyze => Compare Means => Means... requires you to select at least two variables, one for each of the Dependent List (whose means are to be displayed) and Independent List (of variables defining the groups), though you can select more if you wish. Obviously, any variables used to define groups should be suitable for the purpose (measurements recorded to several significant figures and/or at irregular intervals will need first to be grouped, for instance).

Adding more variables to the Independent List results in additional sets of breakdown tables. If desired, additional layers of grouping variables may be used to obtain a more detailed (nested) breakdown of the data. To choose from the wide range of statistics available, use the Options... button.

1.11 Saving the data file

Initially the contents of all SPSS windows are held as temporary files. File => Save... (or File => Save As...) is used to make a permanent copy of whatever temporary file is denoted by the active window. So to save the data which you entered in the previous section, you must first select the Data Editor window. Don't forget to do this, as otherwise you may only be saving the contents of the Output Viewer window!

If the contents of the window have not been saved previously, File => Save... will display a dialog box allowing you to choose a file name and location. Note that if you launched SPSS from the button or the PWF Programs and Information desktop folder the drive and directory offered by default may not be suitable! Navigate to a more appropriate one, if necessary, such as a directory in your own PWF filespace, on drive F.

File => Save As... enables you to save to a file with a different name (assuming that the window is displaying an existing file). Again, take care to select the required drive and directory.

In the case of a data file, you can choose different formats. The default is an SPSS 11 data file, but you could specify a dBase or Excel file, for example.

1.12 Leaving SPSS

Exit SPSS by one of the usual Windows methods:

• choose Exit from the File menu;
• use the control-menu box at the top left-hand corner of the Data Editor window to select Close (it must be the Data Editor because this is the only window that SPSS absolutely requires).

You will be asked if you want to save any windows which have been changed during the SPSS session.

2. Some other useful information about SPSS

2.1 Built-in sources of information

2.1.1 The SPSS Tutorial

SPSS for Windows comes with its own online Tutorial. This is offered in the opening dialog box when you launch SPSS (unless you have previously checked the Don't show this dialog in the future box in its bottom left corner). It may also be accessed at any time while SPSS is running from SPSS's own Help menu, which is discussed later. It opens in a web browser, with four navigation buttons in the lower right-hand corner.

Click on the second of these to open the table of contents (to expand the various sections, click on the + symbols beside them).

2.1.2 The Help Menu

There are various kinds of help provided by SPSS besides the Help buttons on dialog boxes. Most are obtainable from a Help pull-down menu in the menu bar of an SPSS window. They include

• The Tutorial;
• Case Studies ('slide shows' explaining various SPSS procedures and their output);
• The Statistics Coach, which helps the beginner to choose a suitable statistical technique (but it covers only a very small subset of SPSS's capabilities);
• The entire contents of the various Syntax Guides for SPSS (in PDF format, to be viewed using Adobe Acrobat). These are of no interest to the beginner but are useful for more experienced users writing programs in the SPSS command language;
• A link to the SPSS web site.

Help may be obtained by other means, e.g.

• Press the F1 function key (at the top of the keyboard) to open a Help window;
• Begin by double-clicking on a pivot table or chart in the Output Viewer (it also allows you to modify the table or chart, but that is outside the scope of this leaflet). This causes a Results Coach to be offered both in the Help menu and as a button on the Formatting toolbar; you can also right-click on part of the pivot table to obtain a definition or explanation (select What's this? from the popup menu).

2.2 Launching SPSS from an SPSS application file

If SPSS is already running, or if SPSS has been used earlier in the same PWF session, you can double-click on any existing SPSS application file to open it into an SPSS window of the appropriate type.

The principal SPSS filename extensions and their (large) icons are as shown in the following table. Windows uses a file's extension to determine how it should be treated. In Windows these extensions may be hidden from view (you will sometimes find that a PWF workstation has been left in this state by a previous user). It is therefore advisable to learn to distinguish at least the first two icons, which are those most commonly encountered.

File Type	File Name	Icon
SPSS data file	*.sav
SPSS output file	*.spo
SPSS syntax file	*.sps
SPSS (standard) chart template file	*.sct
SPSS (interactive) ChartLooks file	*.clo
SPSS TableLooks file	*.tlo
SPSS production job file	*.spp

2.3 Printing SPSS output

Printing output directly from SPSS for Windows follows normal Windows procedures. Make sure that the Output Viewer is the active window, if necessary by selecting it from the Window menu. Use File => Print... to print all or part of its contents. By default, SPSS will print everything that is displayed when you scroll the output pane from top to bottom. To inspect the appearance of the output on the page before printing it, use File => Print Preview (a facility available only for output, and not for data, etc.).

To avoid excessive amounts of output, either edit the contents of the window to delete anything you do not want, or select the particular object or objects that you wish to print before using File => Print... (which will then assume that you wish to print only your selection). To select multiple output objects you can use the Enter and Shift keys with the mouse button as described earlier in Helpful Tip 1 (selecting the objects in the left-hand, Outline pane).

Alternatively, rather than printing directly from SPSS you may prefer to transfer part of your output to Microsoft Word or some other word processor as described later.

2.4 Saving SPSS output

SPSS uses its own special format when saving SPSS Viewer files (with the .spo filename extension). By default, File => Save saves the entire contents of the Viewer, but you can select and save only part if you wish.

Saved .spo files can be read back into SPSS later, either by using File => Open... and specifying Viewer document (*.spo) as the file type, or by double-clicking on the file's icon.

2.5 Copying output to Microsoft Word and other applications

There are two approaches to doing this, depending on whether you want merely to convert the SPSS output to a file in another format, or to insert SPSS output piecemeal into a file.

2.5.1 Bulk transfer of output into a new file

This is done by selecting the Output Viewer window and using File => Export... The formats available are HTML (for the Web), Text, MS Word and MS Excel; selected in the Export Output dialog box via File Type. The default is HTML. By default, SPSS charts are not exported (and they can only be exported to HTML and MS Word; though in a future version of SPSS it may be possible to export them to MS Excel). Use the Browse button to choose a folder in which to create the new file and then give the file a suitable name. You can choose to export all output objects, only the visible objects, or objects that have been selected prior to opening the dialog box.

2.5.2 Transfer of output into an existing document

Individual items of SPSS output (objects) may be transferred to chosen locations in a MS Word document by selecting them and using Edit => Copy in SPSS, and then selecting the insertion point in the Word document and using Edit => Paste. (Drag-and-drop between SPSS and Word behaves in exactly the same way.) The format in which they are then transferred to MS Word depends on the type of object:

To copy multiple (rather than single) objects you must select them all and use Edit => Copy objects rather than Edit => Copy

When pivot tables are pasted into Word as RTF (using Edit => Paste) the table layout occasionally need some re-adjustment in Word , though this is much less of a problem than with earlier versions of SPSS.

It is also possible to use Edit => Paste special => Picture to transfer a pivot table as a picture (Windows metafile) which retains the exact appearance of the original table. The disadvantage of this, though, is that it cannot then be edited as text. Alternatively, you can transfer the table to MS Excel (see below) and tidy it up there before transferring it to Word.

In the SPSS Tutorial you should look at the section Working with Output => Using Results in Other Applications. For more detailed information, see Help => Topics => Contents and follow a similar path.

Pivot tables can also be copied and pasted into an Excel spreadsheet. In this case, because of the use of BIFF format, the full precision of the results is maintained rather than just the number of decimal places displayed by SPSS. If you wish, therefore, further analysis may be performed in Excel.

2.6 Transferring Data between Microsoft Excel and SPSS

To read an Excel file, use File => Open, navigate to the required directory and select Excel (*.xls) from the Files of Type list; until you do this, the file will not be shown. When you have selected the file and clicked on the Open button, an Opening Excel Data Source dialog box will appear. If your spreadsheet includes column headings in the first row, ensure that the Read Variable Names box is checked (otherwise the first row will be treated as data and SPSS will invent variable names itself. Select the worksheet you require (if the file contains more than one) and click on OK.

2.6.1 Notes on preparing Excel files for use with SPSS

Excel provides a convenient and popular means of full-screen data entry and editing data intended for SPSS, because it is available on many machines which lack SPSS. The steps involved are also an object lesson in the transfer of data between two packages, inasmuch as it is necessary to understand how both of them work.

Check the following, to ensure compatibility with SPSS:

• The spreadsheet should be laid out in a manner similar to that used by SPSS, i.e. with columns containing variables and rows containing cases.
• If you have column headings in the first row, these should be valid SPSS variable names, and unique. Otherwise SPSS will do its best to create variable names from them, but you may not like the result very much (it will chop them down to 8 characters, or fewer if it has to add a number at the end to make the name unique; spaces will be converted to underlines, and so on).
• Numeric data should be formatted as numbers (not left in the default format).
• Dates and times must be formatted explicitly as dates and times, which are special instances of numeric formats. Otherwise they may not be correctly interpreted by SPSS.
• Use only fixed width fonts (e.g. Courier and Courier New) rather than variable width fonts. Columns should be wide enough to show the whole of every cell.

2.6.2 Transferring data from Excel to SPSS

To read an Excel file, use File => Open=> Data, navigate to the required directory and select Excel (*.xls) from the Files of Type list (until you do this the file will not be shown). When you have selected the file and clicked on the Open button, an Opening Excel Data Source dialog box will appear. If your spreadsheet includes column headings in the first row, make sure that Read Variable Names is selected. Select the worksheet you require (if the file contains more than one) and click on OK.

2.6.3 Transferring data from SPSS to Excel

From the Data Editor window, use File =>Save as... and select Excel (*.xls) from the Save as Type list. If you want your spreadsheet to include SPSS variable names as column headings in the first row, take care that Write variable names to spreadsheet is selected. Navigate to the appropriate directory, enter the file name and click on OK.

Although some variable definition information (date/time formats and the distinction between string and numeric data) is preserved in the transfer. Other variable definition and formatting information (variable labels, value labels, missing value codes, etc.) cannot be preserved because they rely on concepts that are alien to Excel.

3. Finding out more about SPSS

3.1 DIY course materials from Computing Service Reception

SPSS 11.0 Guide to Data Analysis (book with floppy disk)

This teach-yourself course consists of the book SPSS 11.0 Guide to Data Analysis, by Marija J. Norusis, and a CD. It may be borrowed from Computing Service Reception (a returnable deposit of £50 is required). The Guide gives an introduction both to data analysis and SPSS version 11. Topics covered include describing data, testing hypotheses and examining relationships using SPSS. Chapter exercises provide comprehensive examples of basic statistical techniques. The data files used are available on the PWF, and instructions for copying them are pasted inside the book's cover.

3.2 Publications

Although SPSS 11.5 has quite detailed online documentation, a lot of problems are best solved through access to printed documentation. It is strongly recommended that you make sure you have easy access at least to the base manuals, and if at all possible to those for any other parts of SPSS on which you rely.

The SPSS 11.0 Guide to Data Analysis by Marija J. Norusis described earlier is highly recommended as an introduction to the practical use of SPSS, and is well worth considering for individual or library purchase.

The official SPSS publications appropriate for SPSS 11.5 are listed here. Note that some manuals have not been produced in a version 11.5 edition. Copies of most of these manuals are available in the Phoenix Teaching Room and other Computing Service User Libraries.

• SPSS Base 11.5 User's Guide
• SPSS Base 10.0 Applications Guide
• SPSS Interactive Graphics 11.5
• SPSS 11.5 Syntax Reference Guide (2 vols)
• SPSS 11.5 Brief Guide
• SPSS 11.0 Brief Guide for Macintosh
• SPSS Advanced Models 11.5
• SPSS Regression Models 11.0
• SPSS Categories 11.0
• SPSS Tables 11.5
• SPSS Conjoint 8.0
• SPSS Trends 10.0
• SPSS Missing Value Analysis 7.5
• SPSS Exact Tests 7.0 for Windows
• SPSS 11.0 Guide to Data Analysis (See above. This is available on loan from Computing Service Reception as a Teach-Yourself course)

If you use a manual written for an earlier version, you should be aware that the Analyze menu used to be called Statistics.

There are many 'third party' publications on SPSS. Some are a version or two behind the latest SPSS (and some may cover a more recent version than you are using). They range from simple SPSS primers to excellent textbooks in applied statistics that happen to make use of SPSS. Always check to make sure that they are what you require (it is unlikely that anything published prior to 1997/8 will be useful, except in specialist areas). Look in the academic bookshops to see what they have in stock. For a more comprehensive listing it is worth investigating the major web-based booksellers (but you should try to see copies of the books somewhere before buying, if possible). The list given here is very far from complete (note that it is given in alphabetical order by author).

• Babbie, E. and Halley, F. Adventures in Social Research: Data Analysis using SPSS 11.0 for Windows. Sage Pubns, 2003. [Examples based on General Social Survey data]
• Bryman, A. and Cramer, D. Quantitative data analysis with SPSS Release 10 for Windows. A Guide for Social Scientists. Taylor & Francis, Routledge, 2001. [Strong on statistical coverage]
• Coakes, S. and Steed, L. SPSS Analysis Without Anguish. Jacaranda, John Wiley, 2001.
• Field, A. Discovering Statistics using SPSS for Windows. Advanced Techniques for the Beginner. Sage Pubns, 2000. [This takes the statistics further than most other SPSS books.]
• Foster, J.J. Data Analysis using SPSS 8 to 10 for Windows. A Beginner's Guide. Sage, 2000. [A good feature is that although it uses the SPSS Manager, the resulting SPSS commands are also shown].
• Howitt, D. and Cramer, D. A Guide to Computing Statistics with SPSS 11 for Windows. Pearson Higher Education, Prentice Hall, 2002. Paperback.
• Kinnear, P.R. and Gray, C.D. SPSS for Windows made Simple. 4th edn, Taylor & Francis, 2000. [Based on SPSS 10. ]
• Pallant, J. SPSS Survival Manual. Open University Press, 2001.
• Weinberg, S.L. and Abramowitz, S.K. Data Analysis for the Behavioral Sciences using SPSS. Cambridge University Press, 2002. Paperback.

3.3 Help from the Internet

Under the terms of the University's SPSS licence you are not allowed to approach the manufacturers directly for technical support (only the official Technical Contacts at the University Computing Service are allowed to do that), but there is plenty of technical help out there on the Internet.

The SPSS Inc. Web page is at

• http://www.spss.com/

Besides product information the website offers useful online technical advice: see

• http://support.spss.com/

For support you have to log in to go any further, but there is a guest account and password available for new or occasional users. Once you have logged in, the menu on the left of the page includes a Troubleshooting heading which gives access to SPSS's searchable technical database (Resolution Search). In old documentation you may see this described as the SPSS Answernet.

The Usenet newsgroup for SPSS

• comp.soft-sys.stat.spss

is a useful source of advice from all over the world, if you care to post your problems to it (but try hard to solve them first, to avoid wasting peoples' time, and perhaps search the archives as described next). It is well worth browsing, as the staff of SPSS Inc. monitor it and frequently respond to queries.

The Google Groups web page

• http://groups.google.com/

maintains a searchable archive of discussion lists and newsgroups (including comp.soft-sys.stat.spss of course). You can use this to search for postings about SPSS. From the home page you can do a simple search by using a search expression such as spss & "logistic regression". Alternatively (for instance if you want your search to go be restricted to the comp.soft-sys.stat.spss newsgroup) you can opt for an advanced search. On the homepage click on Advanced Search, or else type the URL

• http://groups.google.com/advanced_search?q=&

Then fill in the resulting form to specify logistic regression (say) as the search term, comp.soft-sys.stat.spss as the newsgroup and perhaps at least a starting date as well. You can even search for postings by your favourite author! These search pages are well worth bookmarking.

If you want to see an example of a 'third party' SPSS 11 tutorial on the web go to

• http://www.indiana.edu/~statmath/stat/spss/win/index.html

There are also useful tutorials for slightly earlier versions (whose main difference is in the way that variable definitions are set up) at

• http://www.ats.ucla.edu/stat/spss/
• http://psych.utoronto.ca/courses/c1/spss/toc.htm

Of the personal sites containing SPSS information, a particularly useful one is Raynald Levesque's at

• http://pages.infinit.net/rlevesqu/

3.4 Computing Service courses

SPSS: Basic Part 1 and SPSS: Basic Part 2 provide an introduction to some essential concepts and are typically given seven times per year. Each lasts one afternoon. They tend to be oversubscribed and early booking is strongly advised (places are limited). There is a follow-up course SPSS: Beyond the Basics (one afternoon), usually given shortly afterwards, which is recommended for SPSS users with anything other than the simplest data.

SPSS: Further Use is a one-afternoon course given once per year and is intended for those who already have moderate SPSS experience and wish to learn more advanced techniques in order to increase their efficiency.

Another one-afternoon course given once per year, Simple Questionnaire Design and Data Entry, is likely be of interest to some SPSS users, although it is not based on a particular software package.

The title of this document is: M583: SPSS® 11.5 (Statistical Software Package) on the PWF PCs
URL: http://www.cam.ac.uk/cs/docs/leaflets/m583/