1 Instructions for SPSS 21 1 Introduction... 2 1.1 Opening the SPSS program... 2 1.2 General... 2 2 Data inputting and processing... 2 2.1 Manual input and data processing... 2 2.2 Saving data... 3 2.3 Transforming variables... 3 2.4 Selecting a sub-group or exclude data... 4 2.5 Combining files... 5 3 Descriptive statistics... 5 3.1 For nominal or ordinal variables (qualitative)... 5 3.1.1 Pie chart... 5 3.1.2 Bar chart... 5 3.1.3 Three-dimensional Bar chart divided into groups... 6 3.1.4 Bar chart for multiple response... 6 3.1.5 Frequency tables... 6 3.2 For metric and continuous variables (scale)... 7 3.2.1 Histogram... 7 3.2.2 Bar chart for means... 7 3.2.3 Statistics (mean, standard deviation etc.)... 7 3.2.4 Explore: Outliers, skewness, kurtosis, normal distribution, etc.... 8 3.3 Edit graphs... 8 3.4 Edit frequency tables... 8 4 Statistical analyses... 9 4.1 CHI 2 -goodness-of-fit-test... 9 4.2 CHI 2 -test of independence in contingency tables... 9 4.2.1 CHI 2 -TEST of independence with Multiple response sets... 9 4.3 One-sample T-TEST... 10 4.4 Paired-samples T-TEST... 10 4.5 Independent-samples T-TEST... 10 4.6 SIGN TEST and WILCOXON'S TEST for paired-samples... 10 4.7 MANN WHITNEY'S U-TEST (& Wilcoxon's rank sum test)... 11 4.8 RUNS TEST for randomnes... 11 4.9 ANOVA (one factor model)... 11 4.10 ANOVA (more than one factor incl. interactions)... 11 4.11 KRUSKAL-WALLIS' TEST... 12 4.12 HOTELLING'S T 2 for two independent samples... 12 4.13 HOTELLING'S T 2 for paired samples = one-sample... 12 4.14 MANOVA... 12 4.15 DISCRIMINANT ANALYSIS... 13 4.16 LOGISTIC REGRESSION... 13 4.17 REGRESSION ANALYSIS... 13 4.18 CORRELATIONS (Pearson and/or Spearman)... 14 4.19 CANONICAL CORRELATION... 14 4.20 FACTOR ANALYSIS... 14 4.21 CRONBACH S ALPHA... 15 4.22 CLUSTER ANALYSIS (K-MEANS)... 15 4.23 CLUSTER ANALYSIS (HIERARCHIC)... 15 5 Interpreting outputs - Help... 16 6 Printouts... 16 6.1 Printouts can be resized and copied as images (to e.g. Word and PowerPoint)... 16 6.2 Printouts can be exported to Word... 16 6.3 Printouts can be printed... 16 6.4 Printouts can be saved in SPSS... 16
2 1 Introduction 1.1 Opening the SPSS program Open SPSS by choosing Programs IBM SPSS Statistics. IBM SPSS Statistics 21 NOTE! The program is rather large and will take a while to load. If a dialogue box appears with the question; "What do you want to do?" choose Type in data if you are going to start inputting data. 1.2 General SPSS uses two windows: one for data storage and processing (Data Editor), and one for showing results (Output Window opens automatically when there are results to present). The functions and calculations are made through the menus (not within cells as in e.g. Excel). Analysing data material with SPSS requires three steps: 1) Input of data into the Data Editor 2) Choice of procedure/printout from the menus 3) Results in the Output Window. 2 Data inputting and processing Data input takes place within the Data Editor. Data can also: - consist of a previously made SPSS file: choose File Open Data ( -Files of type: SPSS (name.sav) ) - consist of a text file: choose File Read Text Data => follow instructions - be imported from other programs, e.g. Excel choose File Open Data -Files of type Excel (name.xls) - be pasted into the table copy the material in the original file, choose the area where the material shall be pasted into SPSS, and choose Paste. However, the usual method is to directly input the data into SPSS. 2.1 Manual input and data processing The Data Editor window consists of two views (chosen through the tabs on the bottom left): The variables are defined in Variable view (name, number of decimals etc), while data input occurs in Data view. A) Start by defining the variables: Click on Variable view at the bottom of the page =>variable page opens. Each row in Variable view corresponds to a variable. Each column describes the features of the variable: NAME: a short variable name, must start with a letter and cannot contain spaces, dots or other symbols TYPE: numeric indicates that data is inputted as numbers (other special alternatives as e.g. date, currency, and string can be chosen by first clicking on Numeric ) WIDTH: maximum number of characters, numbers, in the variable (can be increased) DECIMALS: number of decimals in the variable LABEL: a longer variable name can be given here containing any characters
3 VALUES: defining the coding used for nominal or ordinal variables (click on None and on the dots, define the codes, see the example below) MISSING: none indicates that unanswered questions are left empty in the data file COLUMNS: column width shown on screen (can be increased) ALIGNMENT: right (alternative positioning: left or centre) MEASURE: defines data level: nominal, ordinal or scale (change by clicking on Scale ) ROLE: Input (alternatively: target, both etc.). Some dialogs support predefined roles. NOTE! After a variable is named (NAME), its other properties will automatically receive the default settings. In most cases it is not necessary to change more than MEASURE, and define VALUES (numerical codes) for nominal and ordinal variables. Example: Define the Loyal Customer' variable (yes/no => nominal) NAME: Loyalcustomer TYPE: Numeric (since number codes are being used for the input. See Values below) DECIMALS: 0 LABEL: A loyal customer VALUES: Define how a loyal customer has been coded (first click on None ) Value: 1 Label: Yes Add Value: 0 Label: No Add MISSING: None (unanswered questions will be left empty) MEASURE: Nominal (qualitative variable without ranking = nominal level) ROLE: Input B) Input data: Click on Data view at the bottom of the page =>data page opens. Each column in Data view corresponds to a question (a variable). Each row corresponds to a case, i.e. an answer to the question (an observation). Input your data using numerical values and codes. => By choosing: View - Value Labels the number codes will be replaced with text, e.g. Loyalcustomer Loyalcustomer 1 yes 1 yes 0 no 1 yes 0 no 0 no NOTE! If mistakes are made at the data input stage, one single value can be replaced by writing on top, or erased by pressing Delete. In order to erase the entire row (observation), the row must be chosen from the row number to the left and then the Delete button must be pressed. If only the cells in one row are chosen, the values will be erased, but the observation (row) will remain as an empty answer, which will affect the sample size. 2.2 Saving data When data input is ready, save the data file. Choose File Save => name.sav Alternatively it can be saved as e.g. an Excel files: File - Save as => name.xls 2.3 Transforming variables In the Transform menu you ll find options for recoding variables, creating new variables from old ones or e.g. dividing scale variables into classes.
4 A) Create a new variable by using another variable (recoding, class dividing etc.): TRANSFORM Recode Into Different Variables (both the new and the original will be saved) Example: divide an age variable into classes => To divide age into e.g. two classes (under 30 and 30 or above) choose: Input variable: age (choose from the list to the left, click ) Output Variable Name: ageclass (give a name for the new variable) Change. Old and New Values Choose one of Range option buttons at a time in order to define the limits for each interval. Define the corresponding class number for the interval (e.g. 0 29 becomes 1, 30 becomes 2) and press Add. Old Values: New Values: Old -> New: Define class 1: * Range: Lowest through value:29 * Value: 1 Add Define class 2: * All other values * Value: 2 Add Add also: * System-missing * System-missing Add Continue => a new variable, ageclass, has been created last in the data file. Move to Variable view and define VALUES for ageclass according to the coding, and change MEASURE=ordinal for the new variable. B) Numerical calculations (sum, logarithm, change of a scale etc.) TRANSFORM - Compute: Example 1. Change a scale going from 1=often to 5=never to go from 5=often to 1=never by subtracting the values from 6: Target variable: newvar (a new name is given for the variable) Numeric Expression: 6 the old variable (choose from the list to the left, click ) Example 2. Calculate age from year of birth => subtract the year of birth from present year Target variable: age (a new name is given for the variable) Numeric Expression: 2012 yearofbirth (choose yearofbirth from the list to the left, click ) 2.4 Selecting a sub-group or exclude data Sometimes it is meaningful to analyse only a chosen part of the data. In order to choose observations that fulfil certain criteria, conditional clauses can be used. Data Select Cases * If condition is satisfied => IF Choose the filter variable from the list to the left, click and define the condition for values to be chosen (which sub-group of the data) E.g. income < 60 all observations with income < 60 will be chosen. Ex: income > 10 AND income < 60 (all observations with 10< income < 60 will be chosen) Continue - Check that the excluded variables are only filtered (Filtered) not erased (Deleted)
2.5 Combining files When you have data in two different files, it can sometimes be necessary to combine the material into one file. One possibility is to simply copy the material from one file to the other by using Copy- Paste functions. Another possibility is to use Data Merge Files. Merge Files gives you the opportunity to specifically decide how the files will be combined. Use Add cases when two files with different observations for the same variables will be combined Add variables when two files with different variables for the same observations will be combined 5 3 Descriptive statistics 3.1 For nominal or ordinal variables (qualitative) Qualitative variables are preferably presented with pie charts, bar charts and frequency tables (also %) as well as with the mode, median (for ordinal level) etc. 3.1.1 Pie chart Choose from the menus at the top of the page: Graphs - Legacy Dialogs Pie *Summaries for groups of cases - Define Slices represents: N of cases (alt. % of cases) Define slices by: choose a qualitative variable from the list to the left, click If you want to make a pie chart for different sub-groups, choose also: Panel by: Choose a grouping variable from the list to the left, and click either as Rows or Columns In section 3.3 you will find instructions for editing the graph. 3.1.2 Bar chart Choose from the menus at the top of the page: Graphs - Legacy Dialogs Bar Simple *Summaries for groups of cases => Define Bars represent: N of cases (alt. % of cases) Category Axis: choose a qualitative variable from the list to the left, click If you want to make bar charts for different sub-groups, choose also: Panel by: Choose a grouping variable from the list and click either as Rows or Columns In section 3.3 you will find instructions for editing the graph.
6 Alternatively you can choose bar chart with cluster division for making bar charts for different sub-groups: Graphs - Legacy Dialogs Bar Clustered *Summaries for groups of cases => Define Bars represent: % of cases (alt. N of cases) Category axis: choose a qualitative variable from the list to the left, click Define cluster by: choose a grouping variable from the list to the left, click Note that the graph will look differently if you change the order of the variables (category/cluster) because the percentages are calculated inside the clusters. In section 3.3 you will find instructions for editing the graph. 3.1.3 Three-dimensional Bar chart divided into groups 3 D Bar chart X axis represents: * Groups of cases Z axis represents: * Groups of cases Define X category axis (horizontal axis): choose a qualitative variable, click Z category axis (depth axis): choose a qualitative grouping variable, click In section 3.3 you will find instructions for editing the graph. 3.1.4 Bar chart for multiple response Choose from the menus at the top of the page: Graphs - Legacy Dialogs Bar Simple *Summaries of separate variables =>Define Bars represent: Choose all the variables belonging to the multiple response question in the list to the left, click Highlight (i.e. select) the variables Mean(X1), Mean(X2) etc. - Change statistic * Percentage above - Value: 0, (if your variables are coded 1=yes/0=no dummies) (If the variables are coded e.g. 1=yes, 2=no, choose * Percentage below - Value: 2) Continue In section 3.3 you will find instructions for editing the graph. 3.1.5 Frequency tables Analyze Descriptive Statistics Frequencies Variables: choose qualitative variables from the list to the left, click Statistics: e.g. Minimum, Maximum, Mode for ordinal variables Continue In section 3.4 you will find instructions for editing frequency tables.
7 3.2 For metric and continuous variables (scale) Quantitative, continuous variables are preferably presented with a histogram as well as with the mean (average), median, min, max, standard deviation etc. 3.2.1 Histogram Graphs - Legacy Dialogs Histogram Variable: choose quantitative variables from the list to the left, click In section 3.3 you will find instructions for editing the graph. 3.2.2 Bar chart for means Graphs - Legacy Dialogs Bar Simple *Summaries of separate variables - Define Bars represent: choose metric variables from the list to the left, click (If you want to present e.g. medians instead of means you can change this by highlighting the variables Mean(x 1 ), Mean(x 2 ) etc. - Change statistic Median of values - Continue) Category Axis: choose a qualitative grouping variable, click If you want to make bar charts for sub-groups, choose also: Panel by: Choose a grouping variable from the list and click either as Rows or Columns In section 3.3 you will find instructions for editing the graph. 3.2.3 Statistics (mean, standard deviation etc.) Analyze Reports Case Summaries Variables: choose quantitative variables from the list to the left, click Grouping variable: leave empty, or choose a grouping variable if you want to have statistics per group click Exclude: Display cases Statistics: e.g. Mean, Median, Minimum, Maximum - Continue In section 3.3 you will find instructions for editing the graph.
8 3.2.4 Explore: Outliers, skewness, kurtosis, normal distribution, etc. Analyze Descriptive statistics Explore Dependent list: choose the quantitative variables you want to investigate, click Statistics: *descriptives, *outliers, *percentiles Continue Plots: *histogram, *factor levels together, *normality plots with tests Continue 3.3 Edit graphs In order to edit a graph, start by double clicking the graph => a Chart Editor window will open. Below you will find examples of different useful editing options. Elements Show Data Labels => fills in the frequency N or % for each category Close. NOTE! If you have a panelled chart and have chosen to represent the bars or slices as % of cases, the percentages shown in the graph are calculated on the entire material (not per group). If you have chosen to represent them as N of cases the percentages shown in the graph are calculated per sub-group. Edit - Properties: Here you can e.g. - change colours, patterns etc. by first clicking the category you want to edit (or the entire picture) and then choosing Fill & Border, - Apply Close - create shadings and depth (3-D) in the graphs by first clicking on the chart (the pie or bar) and then choosing Depth & Angel, - Apply Close - change chart type with Variables - Element Type, - Apply Close - change class width in a histogram graphs by first clicking directly on the boxes and then choosing Binning X Axis: *Custom, - Apply Close Options - Transpose chart => turns a vertical bar chart horizontally Options Show Grid Lines => inserts a grid behind a bar chart Edit - X or Y: here you can decide the group order on the X-axis, name the end values for the Y-axis etc. Note that if you have transposed a bar chart, X is now on the vertical axis and Y is on the horizontal axis. Edit 3-D Rotation: here you can rotate a three-dimensional histogram See also chapter 6.1 for resizing graphs! 3.4 Edit frequency tables Start by double clicking the table. Then right click directly on the table. Below you will find examples of different useful editing options. Choose from the menu list: Table Looks to choose between different table samples
9 Table Properties to edit a table, e.g. - General: change column width - Cell Formats: change font, font size, font colour, background colour, decimals etc - Borders: border width - Cell Properties to change font, colour, decimals etc. in a specific cell. Click first on the cell you want to edit and choose then Cell Properties and make the changes Pivoting Trays to change places between rows and columns. Drag the coloured arrowboxes from one side to another (Columns to Rows and vice versa). 4 Statistical analyses 4.1 CHI 2 -goodness-of-fit-test NONPARAMETRIC TEST - LEGACY DIALOGS CHI SQUARE TEST VARIABLE LIST: X (a qualitative variable from the list) EXPECTED VALUES: * all categories equal or * values: Add the expected proportions on in turn (note! Sum=1) 4.2 CHI 2 -test of independence in contingency tables DESCRIPTIVE STATISTICS CROSS TABS ROW: X 1 (usually a "background" or cause variable) COLUMN: X 2 (usually a result variable) STATISTICS: * Chi-square CONTINUE CELLS: * Observed Percentage: e.g. * Row (for easier result interpretation) CONTINUE 4.2.1 CHI 2 -TEST of independence with Multiple response sets This is done in two steps: 1) Start by connecting all variables from a multiple response question (= multiple response set). Choose from the menus at the top of the page: DATA - DEFINE MULTIPLE RESPONSE SETS... VARIABLES IN SET: Choose and click from the list to the left all the variables belonging to the multiple response question VARIABLE CODING: * Dichotomies (when variables have two categories, e.g. yes/no) Counted value: 1 (if 1 means "yes", i.e. the value to be noted) SET NAME: XXX (name the multiple response question) (SET LABEL: a longer, descriptive name if necessary) ADD
2) Continue with the test of independence: TABLES - CUSTOM TABLES... RESET - All Tabs * Drag a qualitative grouping variable from the list to the Rows bar in the work field * Drag the multiple response question XXX (last in the list to the left) to the Columns bar in the work field In order to obtain the group percentages double click on the row variable in the table => a Summary Statistics page will open: ROW N %, choose and click APPLY TO SELECTION - Choose the tab: TEST STATISTICS :* Test of independence (Chi-square) * Include multiple response variables in tests - 10 4.3 One-sample T-TEST COMPARE MEANS ONE-SAMPLE T-TEST TEST VARIABLE: Y-variable (metric) TEST VALUE: the value you test the mean against OPTIONS: decide the confidence level for an interval 4.4 Paired-samples T-TEST COMPARE MEANS PAIRED SAMPLES T-TEST PAIRED VARIABLES: Y 1 and Y 2 are chosen as variable 1 and 2 OPTIONS: decide the confidence level for an interval 4.5 Independent-samples T-TEST COMPARE MEANS INDEPENDENT SAMPLES T-TEST TEST VARIABLE: Y-variable (metric) GROUPING VARIABLE: X-variable (qualitative with 2 groups) DEFINE GROUPS: define how the groups are coded in your file (e.g. 1, 2) 4.6 SIGN TEST and WILCOXON'S TEST for paired-samples NONPARAMETRIC TEST - LEGACY DIALOGS 2-RELATED SAMPLES TEST PAIR: Y 1 and Y 2 are chosen as variable 1 and 2 *Wilcoxon or *Sign
11 4.7 MANN WHITNEY'S U-TEST (& Wilcoxon's rank sum test) NONPARAMETRIC TEST - LEGACY DIALOGS 2- INDEPENDENT SAMPLES TEST VARIABLE: Y-variable (ordinal or metric) GROUPING VARIABLE: X-variable (qualitative with 2 groups) DEFINE GROUPS: define how the groups are coded in your file (e.g. 1, 2) * Mann-Whitney U 4.8 RUNS TEST for randomnes NONPARAMETRIC TEST - LEGACY DIALOGS RUNS TEST VARIABLE: Y-variable (ordinal or metric) CUT POINT: *median (alternatively mean, mode or custom) 4.9 ANOVA (one factor model) COMPARE MEANS ONE-WAY ANOVA DEPENDENT: Y-variable (metric) FACTOR: X-variable (grouping factor) OPTIONS: * Descriptive * Homogeneity of variance test * Means plot 4.10 ANOVA (more than one factor incl. interactions) GENERAL LINEAR MODEL UNIVARIATE DEPENDENT: Y-variable (metric) FIXED FACTORS: X-variables (grouping factors) MODEL: *Full factorial, if interactions also are included (or: *Custom, if e.g. interaction terms are excluded: Build Terms: choose the x-variables one at a time Change "Interactions" to "Main effects") Sum of Squares: Type III (alt. I or II) *include intercept OPTIONS - Display: * Descriptive statistics * Homogeneity test => Continue PLOTS: Horizontal: X 1 Separate line: X 2 Add Continue
12 4.11 KRUSKAL-WALLIS' TEST NONPARAMETRIC TEST - LEGACY DIALOGS K-INDEPENDENT SAMPLES TEST VARIABLE: Y-variable (ordinal or metric) GROUPING VARIABLE: X-variable (qualitative with 2 or more groups) DEFINE RANGE: define smallest and largest value for the grouping variable * Kruskal-Wallis H 4.12 HOTELLING'S T 2 for two independent samples GENERAL LINEAR MODEL MULTIVARIATE DEPENDENT: Y-variables (metric) FIXED FACTOR: X-variable (qualitative with 2 groups) OPTIONS - Display: * Descriptive statistics * Homogeneity test => Continue 4.13 HOTELLING'S T 2 for paired samples = one-sample When you test paired samples you should start by calculating the differences Y diff between your variables using the Transform module (see chapter 2.2 B). Hotelling's T 2 is testing the differences against zero (i.e. whether there exist differences or not) GENERAL LINEAR MODEL MULTIVARIATE DEPENDENT: Y diff -variables (or simple Y variables in one sample tests) OPTIONS - Display: * Descriptive statistics * Homogeneity test => Continue 4.14 MANOVA GENERAL LINEAR MODEL MULTIVARIATE DEPENDENT: Y-variables (metric) FIXED FACTORS: X-variables (grouping factors) MODEL: *Full factorial, if interactions also are included (or: *Custom, if e.g. interaction terms are excluded Build Terms: choose the x-variables one at a time Change "Interactions" to "Main effects") Sum of Squares: Type III (alt. I or II) *include intercept => Continue OPTIONS - Display: * Descriptive statistics * Homogeneity test => Continue PLOTS: Horizontal: X 1 Separate line: X 2 (when more than one group variable) Add => Continue
4.15 DISCRIMINANT ANALYSIS CLASSIFY DISCRIMINANT GROUPING VARIABLE: Y-variable (grouping variable) DEFINE RANGE: min-max INDEPENDENT: X-variables (metric + dummies) * enter independent together (or: * use stepwise method, if stepwise selection process is required) STATISTICS, good to choose at least the following: *means *univariate ANOVA *Box's M *Fisher s (METHOD: define method if stepwise procedure is chosen) CLASSIFY PRIOR PROB.:*all groups equal or: *compute from group size (consider costs due to wrong classification when choosing prior prob) DISPLAY: * summary table (table for right classification) PLOTS: * Separate-groups (the discriminant function per group) 4.16 LOGISTIC REGRESSION REGRESSION BINARY LOGISTIC DEPENDENT: Y-variable (grouping variable) COVARIATES: X-variables (metric + dummies) OPTIONS: *classification plots *Hosmer-Lemeshow goodness-of-fit Display: *at last step Classification cut off: (you can change the classification cut-off value from 0,5 to e.g. 0,75 => group into A when e.g. p(a) > 0,75) CONTINUE 4.17 REGRESSION ANALYSIS REGRESSION LINEAR DEPENDENT: Y-variable (metric) INDEPENDENT: X-variables (metric + dummies) Method: Enter (or Stepwise for stepwise procedure) STATISTICS: *Estimates *Model fit *Descriptives (in order to obtain mean etc.) *Collinearity diagnostics (in order to detect multicollinearity) PLOTS: Y= ZRESID X= ZPRED (in order to detect possible heteroscedasticity) *normal probability plot 13
14 4.18 CORRELATIONS (Pearson and/or Spearman) CORRELATE BIVARIATE VARIABLES: X-variables (metric) Correlation coefficients: *Pearson or *Spearman Test of significance: *Two-tailed or *One-tailed 4.19 CANONICAL CORRELATION Canonical correlation is made within SPSS using a macro-procedure according to the following: 1) Open data file to be used 2) Open a syntax window: FILE NEW SYNTAX 3) Write the following text: include 'T:\SPSS\Canonical correlation.sps'. cancorr set1=y 1 y 2... y q / set2=x 1 x 2... x p /. Note that the variables are given with spaces, the variable sets are separated with / and the first and last row end with a full stop. The path above (T:\SPSS\Canonical correlation.sps) is for the server at Hanken. If you are running the programme from your own computer, you must check where the file Canonical correlation.sps is saved on your computer. 4) Start the running with the RUN - All -command (in the syntax window) 4.20 FACTOR ANALYSIS DIMENSION REDUCTION FACTOR VARIABLES: X-variables (quantitative) DESCRIPTIVES: Statistics: *initial solution Correlation Matrix: * KMO and Bartlett's test (suitability test) * Reproduced (investigates the unique part) EXTRACTION Method: (choose extraction method) Extract: *eigenvalue over 1 or *Fixed number of factors: # Display: *unrotated factor solution (can be excluded) *scree plot Maximum iterations=25 (can be increased if necessary) ROTATION Method: *varimax (e.g.) Display: *rotated solution Maximum iterations=25 (can be increased if necessary) (SCORES:*save as variables) OPTIONS - Coefficient Display Format: *sorted by size => sorts the loadings according to size
15 4.21 CRONBACH S ALPHA SCALE RELIABILITY ANALYSIS ITEMS: X-variables (metric) MODEL: Alpha 4.22 CLUSTER ANALYSIS (K-MEANS) CLASSIFY K-MEANS CLUSTER VARIABLES: X-variables (quantitative) (LABEL CASES BY: a "string"-variable in order to identify cases) NUMBERS OF CLUSTERS: # (a number must be given) ITERATE: maximum iterations=10 (can be increased if necessary) (SAVE: *cluster membership) OPTIONS *initial cluster centres (can be excluded) *ANOVA table (*cluster info for each case => lists the cluster belonging for each case, therefore recommended only for small samples) 4.23 CLUSTER ANALYSIS (HIERARCHIC) CLASSIFY HIERARCHICAL CLUSTER VARIABLES: X-variables (quantitative) (LABEL CASES BY: a "string"-variable in order to identify cases) CLUSTER: *Cases DISPLAY: *Statistics & *Plots METHOD: (choose method and distance measure) PLOTS: *Dendrogram (tree structure) Icicle: *none or: *all clusters (same as dendrogram *vertical (height fits better than width)) STATISTICS: *Agglomeration schedule (defines how clusters have been paired together, as well as the distance between them) Cluster membership: * single solution, number of clusters: # (lists the cluster belonging for each case, therefore recommended only for small samples) (SAVE: Cluster membership *single solution, number of clusters: # (defines which cluster solution will be saved, i.e. the number of clusters))
16 5 Interpreting outputs - Help You can get help with the interpretation of printouts in SPSS by choosing: Help Case studies Statistics Base: Choose an analysis or test and proceed by clicking The content of each table will be explained. 6 Printouts 6.1 Printouts can be resized and copied as images (to e.g. Word and PowerPoint) You can resize graphs in SPSS before copying, by dragging from the corners. This will change only the size of the graph itself, while the text size remains the same and will therefore be readable in the target document. You can also enlarge or reduce tables and graphs in the target document, but this will change the size of the whole object (including texts). Graphs and tables can be copied by right-clicking on the objects in turn and choose: Copy Special: Image (JPG, PNG) To paste an image in e.g. Word choose: Paste Special: Picture (JPG or PNG) Images cannot be edited in the target document. 6.2 Printouts can be exported to Word Graphs and tables can be exported by choosing File Export and All visible output or Selection => a word-document is created containing your tables and graphs. All tables can be edited in the Word-document. Note that some tables are too wide to fit in a Word-document, and are therefore recommended to be copied as images instead (see section 5.1). 6.3 Printouts can be printed If you send the printout to the printer (File - Print) choose: All visible output if you want to print everything from the Output window Selected output if you only want to print a selected table or graph 6.4 Printouts can be saved in SPSS Printouts can be saved as SPSS files: File Save => name.spv