4 This chapter documents the statistical procedures that PSPP supports so
8 * DESCRIPTIVES:: Descriptive statistics.
9 * FREQUENCIES:: Frequency tables.
10 * EXAMINE:: Testing data for normality.
11 * CORRELATIONS:: Correlation tables.
12 * CROSSTABS:: Crosstabulation tables.
13 * FACTOR:: Factor analysis and Principal Components analysis
14 * MEANS:: Average values and other statistics.
15 * NPAR TESTS:: Nonparametric tests.
16 * T-TEST:: Test hypotheses about means.
17 * ONEWAY:: One way analysis of variance.
18 * QUICK CLUSTER:: K-Means clustering.
19 * RANK:: Compute rank scores.
20 * REGRESSION:: Linear regression.
21 * RELIABILITY:: Reliability analysis.
22 * ROC:: Receiver Operating Characteristic.
32 /MISSING=@{VARIABLE,LISTWISE@} @{INCLUDE,NOINCLUDE@}
33 /FORMAT=@{LABELS,NOLABELS@} @{NOINDEX,INDEX@} @{LINE,SERIAL@}
35 /STATISTICS=@{ALL,MEAN,SEMEAN,STDDEV,VARIANCE,KURTOSIS,
36 SKEWNESS,RANGE,MINIMUM,MAXIMUM,SUM,DEFAULT,
37 SESKEWNESS,SEKURTOSIS@}
38 /SORT=@{NONE,MEAN,SEMEAN,STDDEV,VARIANCE,KURTOSIS,SKEWNESS,
39 RANGE,MINIMUM,MAXIMUM,SUM,SESKEWNESS,SEKURTOSIS,NAME@}
43 The @cmd{DESCRIPTIVES} procedure reads the active dataset and outputs
45 statistics requested by the user. In addition, it can optionally
48 The VARIABLES subcommand, which is required, specifies the list of
49 variables to be analyzed. Keyword VARIABLES is optional.
51 All other subcommands are optional:
53 The MISSING subcommand determines the handling of missing variables. If
54 INCLUDE is set, then user-missing values are included in the
55 calculations. If NOINCLUDE is set, which is the default, user-missing
56 values are excluded. If VARIABLE is set, then missing values are
57 excluded on a variable by variable basis; if LISTWISE is set, then
58 the entire case is excluded whenever any value in that case has a
59 system-missing or, if INCLUDE is set, user-missing value.
61 The FORMAT subcommand affects the output format. Currently the
62 LABELS/NOLABELS and NOINDEX/INDEX settings are not used. When SERIAL is
63 set, both valid and missing number of cases are listed in the output;
64 when NOSERIAL is set, only valid cases are listed.
66 The SAVE subcommand causes @cmd{DESCRIPTIVES} to calculate Z scores for all
67 the specified variables. The Z scores are saved to new variables.
68 Variable names are generated by trying first the original variable name
69 with Z prepended and truncated to a maximum of 8 characters, then the
70 names ZSC000 through ZSC999, STDZ00 through STDZ09, ZZZZ00 through
71 ZZZZ09, ZQZQ00 through ZQZQ09, in that sequence. In addition, Z score
72 variable names can be specified explicitly on VARIABLES in the variable
73 list by enclosing them in parentheses after each variable.
75 The STATISTICS subcommand specifies the statistics to be displayed:
79 All of the statistics below.
83 Standard error of the mean.
89 Kurtosis and standard error of the kurtosis.
91 Skewness and standard error of the skewness.
101 Mean, standard deviation of the mean, minimum, maximum.
103 Standard error of the kurtosis.
105 Standard error of the skewness.
108 The SORT subcommand specifies how the statistics should be sorted. Most
109 of the possible values should be self-explanatory. NAME causes the
110 statistics to be sorted by name. By default, the statistics are listed
111 in the order that they are specified on the VARIABLES subcommand. The A
112 and D settings request an ascending or descending sort order,
122 /FORMAT=@{TABLE,NOTABLE,LIMIT(limit)@}
123 @{AVALUE,DVALUE,AFREQ,DFREQ@}
124 /MISSING=@{EXCLUDE,INCLUDE@}
125 /STATISTICS=@{DEFAULT,MEAN,SEMEAN,MEDIAN,MODE,STDDEV,VARIANCE,
126 KURTOSIS,SKEWNESS,RANGE,MINIMUM,MAXIMUM,SUM,
127 SESKEWNESS,SEKURTOSIS,ALL,NONE@}
129 /PERCENTILES=percent@dots{}
130 /HISTOGRAM=[MINIMUM(x_min)] [MAXIMUM(x_max)]
131 [@{FREQ[(y_max)],PERCENT[(y_max)]@}] [@{NONORMAL,NORMAL@}]
132 /PIECHART=[MINIMUM(x_min)] [MAXIMUM(x_max)]
133 [@{FREQ,PERCENT@}] [@{NOMISSING,MISSING@}]
135 (These options are not currently implemented.)
141 The @cmd{FREQUENCIES} procedure outputs frequency tables for specified
143 @cmd{FREQUENCIES} can also calculate and display descriptive statistics
144 (including median and mode) and percentiles,
145 @cmd{FREQUENCIES} can also output
146 histograms and pie charts.
148 The VARIABLES subcommand is the only required subcommand. Specify the
149 variables to be analyzed.
151 The FORMAT subcommand controls the output format. It has several
156 TABLE, the default, causes a frequency table to be output for every
157 variable specified. NOTABLE prevents them from being output. LIMIT
158 with a numeric argument causes them to be output except when there are
159 more than the specified number of values in the table.
162 Normally frequency tables are sorted in ascending order by value. This
163 is AVALUE. DVALUE tables are sorted in descending order by value.
164 AFREQ and DFREQ tables are sorted in ascending and descending order,
165 respectively, by frequency count.
168 The MISSING subcommand controls the handling of user-missing values.
169 When EXCLUDE, the default, is set, user-missing values are not included
170 in frequency tables or statistics. When INCLUDE is set, user-missing
171 are included. System-missing values are never included in statistics,
172 but are listed in frequency tables.
174 The available STATISTICS are the same as available in @cmd{DESCRIPTIVES}
175 (@pxref{DESCRIPTIVES}), with the addition of MEDIAN, the data's median
176 value, and MODE, the mode. (If there are multiple modes, the smallest
177 value is reported.) By default, the mean, standard deviation of the
178 mean, minimum, and maximum are reported for each variable.
181 PERCENTILES causes the specified percentiles to be reported.
182 The percentiles should be presented at a list of numbers between 0
184 The NTILES subcommand causes the percentiles to be reported at the
185 boundaries of the data set divided into the specified number of ranges.
186 For instance, @code{/NTILES=4} would cause quartiles to be reported.
189 The HISTOGRAM subcommand causes the output to include a histogram for
190 each specified numeric variable. The X axis by default ranges from
191 the minimum to the maximum value observed in the data, but the MINIMUM
192 and MAXIMUM keywords can set an explicit range. Specify NORMAL to
193 superimpose a normal curve on the histogram. Histograms are not
194 created for string variables.
197 The PIECHART adds a pie chart for each variable to the data. Each
198 slice represents one value, with the size of the slice proportional to
199 the value's frequency. By default, all non-missing values are given
200 slices. The MINIMUM and MAXIMUM keywords can be used to limit the
201 displayed slices to a given range of values. The MISSING keyword adds
202 slices for missing values.
204 The FREQ and PERCENT options on HISTOGRAM and PIECHART are accepted
205 but not currently honored.
208 @comment node-name, next, previous, up
212 @cindex Normality, testing for
216 VARIABLES=var_list [BY factor_list ]
217 /STATISTICS=@{DESCRIPTIVES, EXTREME[(n)], ALL, NONE@}
218 /PLOT=@{BOXPLOT, NPPLOT, HISTOGRAM, ALL, NONE@}
220 /COMPARE=@{GROUPS,VARIABLES@}
223 /PERCENTILE=[value_list]=@{HAVERAGE, WAVERAGE, ROUND, AEMPIRICAL, EMPIRICAL @}
224 /MISSING=@{LISTWISE, PAIRWISE@} [@{EXCLUDE, INCLUDE@}]
225 [@{NOREPORT,REPORT@}]
229 The @cmd{EXAMINE} command is used to test how closely a distribution is to a
230 normal distribution. It also shows you outliers and extreme values.
232 The VARIABLES subcommand specifies the dependent variables and the
233 independent variable to use as factors for the analysis. Variables
234 listed before the first BY keyword are the dependent variables.
235 The dependent variables may optionally be followed by a list of
236 factors which tell PSPP how to break down the analysis for each
237 dependent variable. The format for each factor is
243 The STATISTICS subcommand specifies the analysis to be done.
244 DESCRIPTIVES will produce a table showing some parametric and
245 non-parametrics statistics. EXTREME produces a table showing extreme
246 values of the dependent variable. A number in parentheses determines
247 how many upper and lower extremes to show. The default number is 5.
253 The PLOT subcommand specifies which plots are to be produced if any.
254 Available plots are HISTOGRAM, NPPLOT and BOXPLOT.
256 The COMPARE subcommand is only relevant if producing boxplots, and it is only
257 useful there is more than one dependent variable and at least one factor. If
258 /COMPARE=GROUPS is specified, then one plot per dependent variable is produced,
259 containing boxplots for all the factors.
260 If /COMPARE=VARIABLES is specified, then one plot per factor is produced, each
261 each containing one boxplot per dependent variable.
262 If the /COMPARE subcommand is omitted, then PSPP uses the default value of
265 The ID subcommand also pertains to boxplots. If given, it must
266 specify a variable name. Outliers and extreme cases plotted in
267 boxplots will be labelled with the case from that variable. Numeric or
268 string variables are permissible. If the ID subcommand is not given,
269 then the casenumber will be used for labelling.
271 The CINTERVAL subcommand specifies the confidence interval to use in
272 calculation of the descriptives command. The default it 95%.
275 The PERCENTILES subcommand specifies which percentiles are to be calculated,
276 and which algorithm to use for calculating them. The default is to
277 calculate the 5, 10, 25, 50, 75, 90, 95 percentiles using the
280 The TOTAL and NOTOTAL subcommands are mutually exclusive. If NOTOTAL
281 is given and factors have been specified in the VARIABLES subcommand,
282 then then statistics for the unfactored dependent variables are
283 produced in addition to the factored variables. If there are no
284 factors specified then TOTAL and NOTOTAL have no effect.
287 If many dependent variable are given, or factors are given for which
288 there are many distinct values, then @cmd{EXAMINE} will produce a very
289 large quantity of output.
292 @section CORRELATIONS
297 /VARIABLES = varlist [ WITH varlist ]
302 /VARIABLES = varlist [ WITH varlist ]
303 /VARIABLES = varlist [ WITH varlist ]
306 [ /PRINT=@{TWOTAIL, ONETAIL@} @{SIG, NOSIG@} ]
307 [ /STATISTICS=DESCRIPTIVES XPROD ALL]
308 [ /MISSING=@{PAIRWISE, LISTWISE@} @{INCLUDE, EXCLUDE@} ]
312 The @cmd{CORRELATIONS} procedure produces tables of the Pearson correlation coefficient
313 for a set of variables. The significance of the coefficients are also given.
315 At least one VARIABLES subcommand is required. If the WITH keyword is used, then a non-square
316 correlation table will be produced.
317 The variables preceding WITH, will be used as the rows of the table, and the variables following
318 will be the columns of the table.
319 If no WITH subcommand is given, then a square, symmetrical table using all variables is produced.
322 The @cmd{MISSING} subcommand determines the handling of missing variables.
323 If INCLUDE is set, then user-missing values are included in the
324 calculations, but system-missing values are not.
325 If EXCLUDE is set, which is the default, user-missing
326 values are excluded as well as system-missing values.
329 If LISTWISE is set, then the entire case is excluded from analysis
330 whenever any variable specified in any @cmd{/VARIABLES} subcommand
331 contains a missing value.
332 If PAIRWISE is set, then a case is considered missing only if either of the
333 values for the particular coefficient are missing.
334 The default is PAIRWISE.
336 The PRINT subcommand is used to control how the reported significance values are printed.
337 If the TWOTAIL option is used, then a two-tailed test of significance is
338 printed. If the ONETAIL option is given, then a one-tailed test is used.
339 The default is TWOTAIL.
341 If the NOSIG option is specified, then correlation coefficients with significance less than
342 0.05 are highlighted.
343 If SIG is specified, then no highlighting is performed. This is the default.
346 The STATISTICS subcommand requests additional statistics to be displayed. The keyword
347 DESCRIPTIVES requests that the mean, number of non-missing cases, and the non-biased
348 estimator of the standard deviation are displayed.
349 These statistics will be displayed in a separated table, for all the variables listed
350 in any /VARIABLES subcommand.
351 The XPROD keyword requests cross-product deviations and covariance estimators to
352 be displayed for each pair of variables.
353 The keyword ALL is the union of DESCRIPTIVES and XPROD.
361 /TABLES=var_list BY var_list [BY var_list]@dots{}
362 /MISSING=@{TABLE,INCLUDE,REPORT@}
363 /WRITE=@{NONE,CELLS,ALL@}
364 /FORMAT=@{TABLES,NOTABLES@}
369 /CELLS=@{COUNT,ROW,COLUMN,TOTAL,EXPECTED,RESIDUAL,SRESIDUAL,
370 ASRESIDUAL,ALL,NONE@}
371 /STATISTICS=@{CHISQ,PHI,CC,LAMBDA,UC,BTAU,CTAU,RISK,GAMMA,D,
372 KAPPA,ETA,CORR,ALL,NONE@}
375 /VARIABLES=var_list (low,high)@dots{}
378 The @cmd{CROSSTABS} procedure displays crosstabulation
379 tables requested by the user. It can calculate several statistics for
380 each cell in the crosstabulation tables. In addition, a number of
381 statistics can be calculated for each table itself.
383 The TABLES subcommand is used to specify the tables to be reported. Any
384 number of dimensions is permitted, and any number of variables per
385 dimension is allowed. The TABLES subcommand may be repeated as many
386 times as needed. This is the only required subcommand in @dfn{general
389 Occasionally, one may want to invoke a special mode called @dfn{integer
390 mode}. Normally, in general mode, PSPP automatically determines
391 what values occur in the data. In integer mode, the user specifies the
392 range of values that the data assumes. To invoke this mode, specify the
393 VARIABLES subcommand, giving a range of data values in parentheses for
394 each variable to be used on the TABLES subcommand. Data values inside
395 the range are truncated to the nearest integer, then assigned to that
396 value. If values occur outside this range, they are discarded. When it
397 is present, the VARIABLES subcommand must precede the TABLES
400 In general mode, numeric and string variables may be specified on
401 TABLES. In integer mode, only numeric variables are allowed.
403 The MISSING subcommand determines the handling of user-missing values.
404 When set to TABLE, the default, missing values are dropped on a table by
405 table basis. When set to INCLUDE, user-missing values are included in
406 tables and statistics. When set to REPORT, which is allowed only in
407 integer mode, user-missing values are included in tables but marked with
408 an @samp{M} (for ``missing'') and excluded from statistical
411 Currently the WRITE subcommand is ignored.
413 The FORMAT subcommand controls the characteristics of the
414 crosstabulation tables to be displayed. It has a number of possible
419 TABLES, the default, causes crosstabulation tables to be output.
420 NOTABLES suppresses them.
423 PIVOT, the default, causes each TABLES subcommand to be displayed in a
424 pivot table format. NOPIVOT causes the old-style crosstabulation format
428 AVALUE, the default, causes values to be sorted in ascending order.
429 DVALUE asserts a descending sort order.
432 INDEX/NOINDEX is currently ignored.
435 BOX/NOBOX is currently ignored.
438 The CELLS subcommand controls the contents of each cell in the displayed
439 crosstabulation table. The possible settings are:
455 Standardized residual.
457 Adjusted standardized residual.
461 Suppress cells entirely.
464 @samp{/CELLS} without any settings specified requests COUNT, ROW,
465 COLUMN, and TOTAL. If CELLS is not specified at all then only COUNT
468 The STATISTICS subcommand selects statistics for computation:
475 Pearson chi-square, likelihood ratio, Fisher's exact test, continuity
476 correction, linear-by-linear association.
480 Contingency coefficient.
484 Uncertainty coefficient.
500 Spearman correlation, Pearson's r.
507 Selected statistics are only calculated when appropriate for the
508 statistic. Certain statistics require tables of a particular size, and
509 some statistics are calculated only in integer mode.
511 @samp{/STATISTICS} without any settings selects CHISQ. If the
512 STATISTICS subcommand is not given, no statistics are calculated.
514 @strong{Please note:} Currently the implementation of CROSSTABS has the
519 Pearson's R (but not Spearman) is off a little.
521 T values for Spearman's R and Pearson's R are wrong.
523 Significance of symmetric and directional measures is not calculated.
525 Asymmetric ASEs and T values for lambda are wrong.
527 ASE of Goodman and Kruskal's tau is not calculated.
529 ASE of symmetric somers' d is wrong.
531 Approximate T of uncertainty coefficient is wrong.
534 Fixes for any of these deficiencies would be welcomed.
540 @cindex factor analysis
541 @cindex principal components analysis
542 @cindex principal axis factoring
543 @cindex data reduction
546 FACTOR VARIABLES=var_list
548 [ /METHOD = @{CORRELATION, COVARIANCE@} ]
550 [ /EXTRACTION=@{PC, PAF@}]
552 [ /ROTATION=@{VARIMAX, EQUAMAX, QUARTIMAX, NOROTATE@}]
554 [ /PRINT=[INITIAL] [EXTRACTION] [ROTATION] [UNIVARIATE] [CORRELATION] [COVARIANCE] [DET] [KMO] [SIG] [ALL] [DEFAULT] ]
558 [ /FORMAT=[SORT] [BLANK(@var{n})] [DEFAULT] ]
560 [ /CRITERIA=[FACTORS(@var{n})] [MINEIGEN(@var{l})] [ITERATE(@var{m})] [ECONVERGE (@var{delta})] [DEFAULT] ]
562 [ /MISSING=[@{LISTWISE, PAIRWISE@}] [@{INCLUDE, EXCLUDE@}] ]
565 The FACTOR command performs Factor Analysis or Principal Axis Factoring on a dataset. It may be used to find
566 common factors in the data or for data reduction purposes.
568 The VARIABLES subcommand is required. It lists the variables which are to partake in the analysis.
570 The /EXTRACTION subcommand is used to specify the way in which factors (components) are extracted from the data.
571 If PC is specified, then Principal Components Analysis is used. If PAF is specified, then Principal Axis Factoring is
572 used. By default Principal Components Analysis will be used.
574 The /ROTATION subcommand is used to specify the method by which the extracted solution will be rotated.
575 Three methods are available: VARIMAX (which is the default), EQUAMAX, and QUARTIMAX.
576 If don't want any rotation to be performed, the word NOROTATE will prevent the command from performing any
577 rotation on the data. Oblique rotations are not supported.
579 The /METHOD subcommand should be used to determine whether the covariance matrix or the correlation matrix of the data is
580 to be analysed. By default, the correlation matrix is analysed.
582 The /PRINT subcommand may be used to select which features of the analysis are reported:
586 A table of mean values, standard deviations and total weights are printed.
588 Initial communalities and eigenvalues are printed.
590 Extracted communalities and eigenvalues are printed.
592 Rotated communalities and eigenvalues are printed.
594 The correlation matrix is printed.
596 The covariance matrix is printed.
598 The determinant of the correlation or covariance matrix is printed.
600 The Kaiser-Meyer-Olkin measure of sampling adequacy and the Bartlett test of sphericity is printed.
602 The significance of the elements of correlation matrix is printed.
604 All of the above are printed.
606 Identical to INITIAL and EXTRACTION.
609 If /PLOT=EIGEN is given, then a ``Scree'' plot of the eigenvalues will be printed. This can be useful for visualizing
610 which factors (components) should be retained.
612 The /FORMAT subcommand determined how data are to be displayed in loading matrices. If SORT is specified, then the variables
613 are sorted in descending order of significance. If BLANK(@var{n}) is specified, then coefficients whose absolute value is less
614 than @var{n} will not be printed. If the keyword DEFAULT is given, or if no /FORMAT subcommand is given, then no sorting is
615 performed, and all coefficients will be printed.
617 The /CRITERIA subcommand is used to specify how the number of extracted factors (components) are chosen. If FACTORS(@var{n}) is
618 specified, where @var{n} is an integer, then @var{n} factors will be extracted. Otherwise, the MINEIGEN setting will
619 be used. MINEIGEN(@var{l}) requests that all factors whose eigenvalues are greater than or equal to @var{l} are extracted.
620 The default value of @var{l} is 1. The ECONVERGE and ITERATE settings have effect only when iterative algorithms for factor
621 extraction (such as Principal Axis Factoring) are used. ECONVERGE(@var{delta}) specifies that iteration should cease when
622 the maximum absolute value of the communality estimate between one iteration and the previous is less than @var{delta}. The
623 default value of @var{delta} is 0.001.
624 The ITERATE(@var{m}) setting sets the maximum number of iterations to @var{m}. The default value of @var{m} is 25.
626 The @cmd{MISSING} subcommand determines the handling of missing variables.
627 If INCLUDE is set, then user-missing values are included in the
628 calculations, but system-missing values are not.
629 If EXCLUDE is set, which is the default, user-missing
630 values are excluded as well as system-missing values.
632 If LISTWISE is set, then the entire case is excluded from analysis
633 whenever any variable specified in the @cmd{VARIABLES} subcommand
634 contains a missing value.
635 If PAIRWISE is set, then a case is considered missing only if either of the
636 values for the particular coefficient are missing.
637 The default is LISTWISE.
648 [ BY @{varlist@} [BY @{varlist@} [BY @{varlist@} @dots{} ]]]
651 [ BY @{varlist@} [BY @{varlist@} [BY @{varlist@} @dots{} ]]] ]
653 [/CELLS = [MEAN] [COUNT] [STDDEV] [SEMEAN] [SUM] [MIN] [MAX] [RANGE]
654 [VARIANCE] [KURT] [SEKURT]
655 [SKEW] [SESKEW] [FIRST] [LAST]
656 [HARMONIC] [GEOMETRIC]
661 [/MISSING = [TABLE] [INCLUDE] [DEPENDENT]]
664 You can use the MEANS command to calculate the arithmetic mean and similar
665 statistics, either for the dataset as a whole or for categories of data.
667 The simplest form of the command is
671 @noindent which calculates the mean, count and standard deviation for @var{v}.
672 If you specify a grouping variable, for example
674 MEANS @var{v} BY @var{g}.
676 @noindent then the means, counts and standard deviations for @var{v} after having
677 been grouped by @var{g} will be calculated.
678 Instead of the mean, count and standard deviation, you could specify the statistics
679 in which you are interested:
681 MEANS @var{x} @var{y} BY @var{g}
682 /CELLS = HARMONIC SUM MIN.
684 This example calculates the harmonic mean, the sum and the minimum values of @var{x} and @var{y}
687 The CELLS subcommand specifies which statistics to calculate. The available statistics
691 @cindex arithmetic mean
694 The count of the values.
696 The standard deviation.
698 The standard error of the mean.
700 The sum of the values.
706 The difference between the maximum and minimum values.
710 The first value in the category.
712 The last value in the category.
716 The standard error of the skewness.
720 The standard error of the kurtosis.
722 @cindex harmonic mean
725 @cindex geometric mean
729 In addition, three special keywords are recognized:
732 This is the same as MEAN COUNT STDDEV
734 All of the above statistics will be calculated.
736 No statistics will be calculated (only a summary will be shown).
740 More than one @dfn{table} can be specified in a single command.
741 Each table is separated by a @samp{/}. For
745 @var{c} @var{d} @var{e} BY @var{x}
746 /@var{a} @var{b} BY @var{x} @var{y}
747 /@var{f} BY @var{y} BY @var{z}.
749 has three tables (the @samp{TABLE =} is optional).
750 The first table has three dependent variables @var{c}, @var{d} and @var{e}
751 and a single categorical variable @var{x}.
752 The second table has two dependent variables @var{a} and @var{b},
753 and two categorical variables @var{x} and @var{y}.
754 The third table has a single dependent variables @var{f}
755 and a categorical variable formed by the combination of @var{y} and @var{z}.
758 By default values are omitted from the analysis only if missing values
759 (either system missing or user missing)
760 for any of the variables directly involved in their calculation are
762 This behaviour can be modified with the /MISSING subcommand.
763 Three options are possible: TABLE, INCLUDE and DEPENDENT.
765 /MISSING = TABLE causes cases to be dropped if any variable is missing
766 in the table specification currently being processed, regardless of
767 whether it is needed to calculate the statistic.
769 /MISSING = INCLUDE says that user missing values, either in the dependent
770 variables or in the categorical variables should be taken at their face
771 value, and not excluded.
773 /MISSING = DEPENDENT says that user missing values, in the dependent
774 variables should be taken at their face value, however cases which
775 have user missing values for the categorical variables should be omitted
776 from the calculation.
782 @cindex nonparametric tests
787 nonparametric test subcommands
792 [ /STATISTICS=@{DESCRIPTIVES@} ]
794 [ /MISSING=@{ANALYSIS, LISTWISE@} @{INCLUDE, EXCLUDE@} ]
796 [ /METHOD=EXACT [ TIMER [(n)] ] ]
799 NPAR TESTS performs nonparametric tests.
800 Non parametric tests make very few assumptions about the distribution of the
802 One or more tests may be specified by using the corresponding subcommand.
803 If the /STATISTICS subcommand is also specified, then summary statistics are
804 produces for each variable that is the subject of any test.
806 Certain tests may take a long time to execute, if an exact figure is required.
807 Therefore, by default asymptotic approximations are used unless the
808 subcommand /METHOD=EXACT is specified.
809 Exact tests give more accurate results, but may take an unacceptably long
810 time to perform. If the TIMER keyword is used, it sets a maximum time,
811 after which the test will be abandoned, and a warning message printed.
812 The time, in minutes, should be specified in parentheses after the TIMER keyword.
813 If the TIMER keyword is given without this figure, then a default value of 5 minutes
818 * BINOMIAL:: Binomial Test
819 * CHISQUARE:: Chisquare Test
820 * COCHRAN:: Cochran Q Test
821 * FRIEDMAN:: Friedman Test
822 * KENDALL:: Kendall's W Test
823 * KOLMOGOROV-SMIRNOV:: Kolmogorov Smirnov Test
824 * KRUSKAL-WALLIS:: Kruskal-Wallis Test
825 * MANN-WHITNEY:: Mann Whitney U Test
826 * MCNEMAR:: McNemar Test
827 * MEDIAN:: Median Test
829 * SIGN:: The Sign Test
830 * WILCOXON:: Wilcoxon Signed Ranks Test
835 @subsection Binomial test
837 @cindex binomial test
840 [ /BINOMIAL[(p)]=var_list[(value1[, value2)] ] ]
843 The /BINOMIAL subcommand compares the observed distribution of a dichotomous
844 variable with that of a binomial distribution.
845 The variable @var{p} specifies the test proportion of the binomial
847 The default value of 0.5 is assumed if @var{p} is omitted.
849 If a single value appears after the variable list, then that value is
850 used as the threshold to partition the observed values. Values less
851 than or equal to the threshold value form the first category. Values
852 greater than the threshold form the second category.
854 If two values appear after the variable list, then they will be used
855 as the values which a variable must take to be in the respective
857 Cases for which a variable takes a value equal to neither of the specified
858 values, take no part in the test for that variable.
860 If no values appear, then the variable must assume dichotomous
862 If more than two distinct, non-missing values for a variable
863 under test are encountered then an error occurs.
865 If the test proportion is equal to 0.5, then a two tailed test is
866 reported. For any other test proportion, a one tailed test is
868 For one tailed tests, if the test proportion is less than
869 or equal to the observed proportion, then the significance of
870 observing the observed proportion or more is reported.
871 If the test proportion is more than the observed proportion, then the
872 significance of observing the observed proportion or less is reported.
873 That is to say, the test is always performed in the observed
876 PSPP uses a very precise approximation to the gamma function to
877 compute the binomial significance. Thus, exact results are reported
878 even for very large sample sizes.
883 @subsection Chisquare Test
885 @cindex chisquare test
889 [ /CHISQUARE=var_list[(lo,hi)] [/EXPECTED=@{EQUAL|f1, f2 @dots{} fn@}] ]
893 The /CHISQUARE subcommand produces a chi-square statistic for the differences
894 between the expected and observed frequencies of the categories of a variable.
895 Optionally, a range of values may appear after the variable list.
896 If a range is given, then non integer values are truncated, and values
897 outside the specified range are excluded from the analysis.
899 The /EXPECTED subcommand specifies the expected values of each
901 There must be exactly one non-zero expected value, for each observed
902 category, or the EQUAL keywork must be specified.
903 You may use the notation @var{n}*@var{f} to specify @var{n}
904 consecutive expected categories all taking a frequency of @var{f}.
905 The frequencies given are proportions, not absolute frequencies. The
906 sum of the frequencies need not be 1.
907 If no /EXPECTED subcommand is given, then then equal frequencies
912 @subsection Cochran Q Test
914 @cindex Cochran Q test
918 [ /COCHRAN = varlist ]
921 The Cochran Q test is used to test for differences between three or more groups.
922 The data for @var{varlist} in all cases must assume exactly two distinct values (other than missing values).
924 The value of Q will be displayed and its Asymptotic significance based on a chi-square distribution.
927 @subsection Friedman Test
929 @cindex Friedman test
932 [ /FRIEDMAN = varlist ]
935 The Friedman test is used to test for differences between repeated measures when there is no indication that the distributions are normally distributed.
937 A list of variables which contain the measured data must be given. The procedure prints the sum of ranks for each variable, the test statistic and its significance.
940 @subsection Kendall's W Test
942 @cindex Kendall's W test
943 @cindex coefficient of concordance
946 [ /KENDALL = varlist ]
949 The Kendall test investigates whether an arbitrary number of related samples come from the
951 It is identical to the Friedman test except that the additional statistic W, Kendall's Coefficient of Concordance is printed.
952 It has the range [0,1] --- a value of zero indicates no agreement between the samples whereas a value of
953 unity indicates complete agreement.
956 @node KOLMOGOROV-SMIRNOV
957 @subsection Kolmogorov-Smirnov Test
958 @vindex KOLMOGOROV-SMIRNOV
960 @cindex Kolmogorov-Smirnov test
963 [ /KOLMOGOROV-SMIRNOV (@{NORMAL [@var{mu}, @var{sigma}], UNIFORM [@var{min}, @var{max}], POISSON [@var{lambda}], EXPONENTIAL [@var{scale}] @}) = varlist ]
966 The one sample Kolmogorov-Smirnov subcommand is used to test whether or not a dataset is
967 drawn from a particular distribution. Four distributions are supported, @i{viz:}
968 Normal, Uniform, Poisson and Exponential.
970 Ideally you should provide the parameters of the distribution against which you wish to test
971 the data. For example, with the normal distribution the mean (@var{mu})and standard deviation (@var{sigma})
972 should be given; with the uniform distribution, the minimum (@var{min})and maximum (@var{max}) value should
974 However, if the parameters are omitted they will be imputed from the data. Imputing the
975 parameters reduces the power of the test so should be avoided if possible.
977 In the following example, two variables @var{score} and @var{age} are tested to see if
978 they follow a normal distribution with a mean of 3.5 and a standard deviation of 2.0.
981 /KOLMOGOROV-SMIRNOV (normal 3.5 2.0) = @var{score} @var{age}.
983 If the variables need to be tested against different distributions, then a separate
984 subcommand must be used. For example the following syntax tests @var{score} against
985 a normal distribution with mean of 3.5 and standard deviation of 2.0 whilst @var{age}
986 is tested against a normal distribution of mean 40 and standard deviation 1.5.
989 /KOLMOGOROV-SMIRNOV (normal 3.5 2.0) = @var{score}
990 /KOLMOGOROV-SMIRNOV (normal 40 1.5) = @var{age}.
993 The abbreviated subcommand K-S may be used in place of KOLMOGOROV-SMIRNOV.
996 @subsection Kruskal-Wallis Test
997 @vindex KRUSKAL-WALLIS
999 @cindex Kruskal-Wallis test
1002 [ /KRUSKAL-WALLIS = varlist BY var (lower, upper) ]
1005 The Kruskal-Wallis test is used to compare data from an
1006 arbitrary number of populations. It does not assume normality.
1007 The data to be compared are specified by @var{varlist}.
1008 The categorical variable determining the groups to which the
1009 data belongs is given by @var{var}. The limits @var{lower} and
1010 @var{upper} specify the valid range of @var{var}. Any cases for
1011 which @var{var} falls outside [@var{lower}, @var{upper}] will be
1014 The mean rank of each group as well as the chi-squared value and significance
1015 of the test will be printed.
1016 The abbreviated subcommand K-W may be used in place of KRUSKAL-WALLIS.
1020 @subsection Mann-Whitney U Test
1021 @vindex MANN-WHITNEY
1023 @cindex Mann-Whitney U test
1024 @cindex U, Mann-Whitney U
1027 [ /MANN-WHITNEY = varlist BY var (group1, group2) ]
1030 The Mann-Whitney subcommand is used to test whether two groups of data come from different populations.
1031 The variables to be tested should be specified in @var{varlist} and the grouping variable, that determines to which group the test variables belong, in @var{var}.
1032 @var{Var} may be either a string or an alpha variable.
1033 @var{Group1} and @var{group2} specify the
1034 two values of @var{var} which determine the groups of the test data.
1035 Cases for which the @var{var} value is neither @var{group1} or @var{group2} will be ignored.
1037 The value of the Mann-Whitney U statistic, the Wilcoxon W, and the significance will be printed.
1038 The abbreviated subcommand M-W may be used in place of MANN-WHITNEY.
1041 @subsection McNemar Test
1043 @cindex McNemar test
1046 [ /MCNEMAR varlist [ WITH varlist [ (PAIRED) ]]]
1049 Use McNemar's test to analyse the significance of the difference between
1050 pairs of correlated proportions.
1052 If the @code{WITH} keyword is omitted, then tests for all
1053 combinations of the listed variables are performed.
1054 If the @code{WITH} keyword is given, and the @code{(PAIRED)} keyword
1055 is also given, then the number of variables preceding @code{WITH}
1056 must be the same as the number following it.
1057 In this case, tests for each respective pair of variables are
1059 If the @code{WITH} keyword is given, but the
1060 @code{(PAIRED)} keyword is omitted, then tests for each combination
1061 of variable preceding @code{WITH} against variable following
1062 @code{WITH} are performed.
1064 The data in each variable must be dichotomous. If there are more
1065 than two distinct variables an error will occur and the test will
1069 @subsection Median Test
1074 [ /MEDIAN [(value)] = varlist BY variable (value1, value2) ]
1077 The median test is used to test whether independent samples come from
1078 populations with a common median.
1079 The median of the populations against which the samples are to be tested
1080 may be given in parentheses immediately after the
1081 /MEDIAN subcommand. If it is not given, the median will be imputed from the
1082 union of all the samples.
1084 The variables of the samples to be tested should immediately follow the @samp{=} sign. The
1085 keyword @code{BY} must come next, and then the grouping variable. Two values
1086 in parentheses should follow. If the first value is greater than the second,
1087 then a 2 sample test is performed using these two values to determine the groups.
1088 If however, the first variable is less than the second, then a @i{k} sample test is
1089 conducted and the group values used are all values encountered which lie in the
1090 range [@var{value1},@var{value2}].
1094 @subsection Runs Test
1099 [ /RUNS (@{MEAN, MEDIAN, MODE, value@}) = varlist ]
1102 The /RUNS subcommand tests whether a data sequence is randomly ordered.
1104 It works by examining the number of times a variable's value crosses a given threshold.
1105 The desired threshold must be specified within parentheses.
1106 It may either be specified as a number or as one of MEAN, MEDIAN or MODE.
1107 Following the threshold specification comes the list of variables whose values are to be
1110 The subcommand shows the number of runs, the asymptotic significance based on the
1114 @subsection Sign Test
1119 [ /SIGN varlist [ WITH varlist [ (PAIRED) ]]]
1122 The /SIGN subcommand tests for differences between medians of the
1124 The test does not make any assumptions about the
1125 distribution of the data.
1127 If the @code{WITH} keyword is omitted, then tests for all
1128 combinations of the listed variables are performed.
1129 If the @code{WITH} keyword is given, and the @code{(PAIRED)} keyword
1130 is also given, then the number of variables preceding @code{WITH}
1131 must be the same as the number following it.
1132 In this case, tests for each respective pair of variables are
1134 If the @code{WITH} keyword is given, but the
1135 @code{(PAIRED)} keyword is omitted, then tests for each combination
1136 of variable preceding @code{WITH} against variable following
1137 @code{WITH} are performed.
1140 @subsection Wilcoxon Matched Pairs Signed Ranks Test
1141 @comment node-name, next, previous, up
1143 @cindex wilcoxon matched pairs signed ranks test
1146 [ /WILCOXON varlist [ WITH varlist [ (PAIRED) ]]]
1149 The /WILCOXON subcommand tests for differences between medians of the
1151 The test does not make any assumptions about the variances of the samples.
1152 It does however assume that the distribution is symetrical.
1154 If the @code{WITH} keyword is omitted, then tests for all
1155 combinations of the listed variables are performed.
1156 If the @code{WITH} keyword is given, and the @code{(PAIRED)} keyword
1157 is also given, then the number of variables preceding @code{WITH}
1158 must be the same as the number following it.
1159 In this case, tests for each respective pair of variables are
1161 If the @code{WITH} keyword is given, but the
1162 @code{(PAIRED)} keyword is omitted, then tests for each combination
1163 of variable preceding @code{WITH} against variable following
1164 @code{WITH} are performed.
1167 @comment node-name, next, previous, up
1174 /MISSING=@{ANALYSIS,LISTWISE@} @{EXCLUDE,INCLUDE@}
1175 /CRITERIA=CIN(confidence)
1183 (Independent Samples mode.)
1184 GROUPS=var(value1 [, value2])
1188 (Paired Samples mode.)
1189 PAIRS=var_list [WITH var_list [(PAIRED)] ]
1194 The @cmd{T-TEST} procedure outputs tables used in testing hypotheses about
1196 It operates in one of three modes:
1198 @item One Sample mode.
1199 @item Independent Groups mode.
1204 Each of these modes are described in more detail below.
1205 There are two optional subcommands which are common to all modes.
1207 The @cmd{/CRITERIA} subcommand tells PSPP the confidence interval used
1208 in the tests. The default value is 0.95.
1211 The @cmd{MISSING} subcommand determines the handling of missing
1213 If INCLUDE is set, then user-missing values are included in the
1214 calculations, but system-missing values are not.
1215 If EXCLUDE is set, which is the default, user-missing
1216 values are excluded as well as system-missing values.
1217 This is the default.
1219 If LISTWISE is set, then the entire case is excluded from analysis
1220 whenever any variable specified in the @cmd{/VARIABLES}, @cmd{/PAIRS} or
1221 @cmd{/GROUPS} subcommands contains a missing value.
1222 If ANALYSIS is set, then missing values are excluded only in the analysis for
1223 which they would be needed. This is the default.
1227 * One Sample Mode:: Testing against a hypothesized mean
1228 * Independent Samples Mode:: Testing two independent groups for equal mean
1229 * Paired Samples Mode:: Testing two interdependent groups for equal mean
1232 @node One Sample Mode
1233 @subsection One Sample Mode
1235 The @cmd{TESTVAL} subcommand invokes the One Sample mode.
1236 This mode is used to test a population mean against a hypothesized
1238 The value given to the @cmd{TESTVAL} subcommand is the value against
1239 which you wish to test.
1240 In this mode, you must also use the @cmd{/VARIABLES} subcommand to
1241 tell PSPP which variables you wish to test.
1243 @node Independent Samples Mode
1244 @comment node-name, next, previous, up
1245 @subsection Independent Samples Mode
1247 The @cmd{GROUPS} subcommand invokes Independent Samples mode or
1249 This mode is used to test whether two groups of values have the
1250 same population mean.
1251 In this mode, you must also use the @cmd{/VARIABLES} subcommand to
1252 tell PSPP the dependent variables you wish to test.
1254 The variable given in the @cmd{GROUPS} subcommand is the independent
1255 variable which determines to which group the samples belong.
1256 The values in parentheses are the specific values of the independent
1257 variable for each group.
1258 If the parentheses are omitted and no values are given, the default values
1259 of 1.0 and 2.0 are assumed.
1261 If the independent variable is numeric,
1262 it is acceptable to specify only one value inside the parentheses.
1263 If you do this, cases where the independent variable is
1264 greater than or equal to this value belong to the first group, and cases
1265 less than this value belong to the second group.
1266 When using this form of the @cmd{GROUPS} subcommand, missing values in
1267 the independent variable are excluded on a listwise basis, regardless
1268 of whether @cmd{/MISSING=LISTWISE} was specified.
1271 @node Paired Samples Mode
1272 @comment node-name, next, previous, up
1273 @subsection Paired Samples Mode
1275 The @cmd{PAIRS} subcommand introduces Paired Samples mode.
1276 Use this mode when repeated measures have been taken from the same
1278 If the @code{WITH} keyword is omitted, then tables for all
1279 combinations of variables given in the @cmd{PAIRS} subcommand are
1281 If the @code{WITH} keyword is given, and the @code{(PAIRED)} keyword
1282 is also given, then the number of variables preceding @code{WITH}
1283 must be the same as the number following it.
1284 In this case, tables for each respective pair of variables are
1286 In the event that the @code{WITH} keyword is given, but the
1287 @code{(PAIRED)} keyword is omitted, then tables for each combination
1288 of variable preceding @code{WITH} against variable following
1289 @code{WITH} are generated.
1293 @comment node-name, next, previous, up
1297 @cindex analysis of variance
1302 [/VARIABLES = ] var_list BY var
1303 /MISSING=@{ANALYSIS,LISTWISE@} @{EXCLUDE,INCLUDE@}
1304 /CONTRAST= value1 [, value2] ... [,valueN]
1305 /STATISTICS=@{DESCRIPTIVES,HOMOGENEITY@}
1306 /POSTHOC=@{BONFERRONI, GH, LSD, SCHEFFE, SIDAK, TUKEY, ALPHA ([value])@}
1309 The @cmd{ONEWAY} procedure performs a one-way analysis of variance of
1310 variables factored by a single independent variable.
1311 It is used to compare the means of a population
1312 divided into more than two groups.
1314 The dependent variables to be analysed should be given in the @code{VARIABLES}
1316 The list of variables must be followed by the @code{BY} keyword and
1317 the name of the independent (or factor) variable.
1319 You can use the @code{STATISTICS} subcommand to tell PSPP to display
1320 ancilliary information. The options accepted are:
1323 Displays descriptive statistics about the groups factored by the independent
1326 Displays the Levene test of Homogeneity of Variance for the
1327 variables and their groups.
1330 The @code{CONTRAST} subcommand is used when you anticipate certain
1331 differences between the groups.
1332 The subcommand must be followed by a list of numerals which are the
1333 coefficients of the groups to be tested.
1334 The number of coefficients must correspond to the number of distinct
1335 groups (or values of the independent variable).
1336 If the total sum of the coefficients are not zero, then PSPP will
1337 display a warning, but will proceed with the analysis.
1338 The @code{CONTRAST} subcommand may be given up to 10 times in order
1339 to specify different contrast tests.
1340 The @code{MISSING} subcommand defines how missing values are handled.
1341 If LISTWISE is specified then cases which have missing values for
1342 the independent variable or any dependent variable will be ignored.
1343 If ANALYSIS is specified, then cases will be ignored if the independent
1344 variable is missing or if the dependent variable currently being
1345 analysed is missing. The default is ANALYSIS.
1346 A setting of EXCLUDE means that variables whose values are
1347 user-missing are to be excluded from the analysis. A setting of
1348 INCLUDE means they are to be included. The default is EXCLUDE.
1350 Using the @code{POSTHOC} subcommand you can perform multiple
1351 pairwise comparisons on the data. The following comparison methods
1355 Least Significant Difference.
1357 Tukey Honestly Significant Difference.
1365 The Games-Howell test.
1369 The optional syntax @code{ALPHA(@var{value})} is used to indicate
1370 that @var{value} should be used as the
1371 confidence level for which the posthoc tests will be performed.
1372 The default is 0.05.
1375 @comment node-name, next, previous, up
1376 @section QUICK CLUSTER
1377 @vindex QUICK CLUSTER
1379 @cindex K-means clustering
1383 QUICK CLUSTER var_list
1384 [/CRITERIA=CLUSTERS(@var{k}) [MXITER(@var{max_iter})]]
1385 [/MISSING=@{EXCLUDE,INCLUDE@} @{LISTWISE, PAIRWISE@}]
1388 The @cmd{QUICK CLUSTER} command performs k-means clustering on the
1389 dataset. This is useful when you wish to allocate cases into clusters
1390 of similar values and you already know the number of clusters.
1392 The minimum specification is @samp{QUICK CLUSTER} followed by the names
1393 of the variables which contain the cluster data. Normally you will also
1394 want to specify @samp{/CRITERIA=CLUSTERS(@var{k})} where @var{k} is the
1395 number of clusters. If this is not given, then @var{k} defaults to 2.
1397 The command uses an iterative algorithm to determine the clusters for
1398 each case. It will continue iterating until convergence, or until @var{max_iter}
1399 iterations have been done. The default value of @var{max_iter} is 2.
1401 The @cmd{MISSING} subcommand determines the handling of missing variables.
1402 If INCLUDE is set, then user-missing values are considered at their face
1403 value and not as missing values.
1404 If EXCLUDE is set, which is the default, user-missing
1405 values are excluded as well as system-missing values.
1407 If LISTWISE is set, then the entire case is excluded from the analysis
1408 whenever any of the clustering variables contains a missing value.
1409 If PAIRWISE is set, then a case is considered missing only if all the
1410 clustering variables contain missing values. Otherwise it is clustered
1411 on the basis of the non-missing values.
1412 The default is LISTWISE.
1416 @comment node-name, next, previous, up
1423 [VARIABLES=] var_list [@{A,D@}] [BY var_list]
1424 /TIES=@{MEAN,LOW,HIGH,CONDENSE@}
1425 /FRACTION=@{BLOM,TUKEY,VW,RANKIT@}
1427 /MISSING=@{EXCLUDE,INCLUDE@}
1429 /RANK [INTO var_list]
1430 /NTILES(k) [INTO var_list]
1431 /NORMAL [INTO var_list]
1432 /PERCENT [INTO var_list]
1433 /RFRACTION [INTO var_list]
1434 /PROPORTION [INTO var_list]
1436 /SAVAGE [INTO var_list]
1439 The @cmd{RANK} command ranks variables and stores the results into new
1442 The VARIABLES subcommand, which is mandatory, specifies one or
1443 more variables whose values are to be ranked.
1444 After each variable, @samp{A} or @samp{D} may appear, indicating that
1445 the variable is to be ranked in ascending or descending order.
1446 Ascending is the default.
1447 If a BY keyword appears, it should be followed by a list of variables
1448 which are to serve as group variables.
1449 In this case, the cases are gathered into groups, and ranks calculated
1452 The TIES subcommand specifies how tied values are to be treated. The
1453 default is to take the mean value of all the tied cases.
1455 The FRACTION subcommand specifies how proportional ranks are to be
1456 calculated. This only has any effect if NORMAL or PROPORTIONAL rank
1457 functions are requested.
1459 The PRINT subcommand may be used to specify that a summary of the rank
1460 variables created should appear in the output.
1462 The function subcommands are RANK, NTILES, NORMAL, PERCENT, RFRACTION,
1463 PROPORTION and SAVAGE. Any number of function subcommands may appear.
1464 If none are given, then the default is RANK.
1465 The NTILES subcommand must take an integer specifying the number of
1466 partitions into which values should be ranked.
1467 Each subcommand may be followed by the INTO keyword and a list of
1468 variables which are the variables to be created and receive the rank
1469 scores. There may be as many variables specified as there are
1470 variables named on the VARIABLES subcommand. If fewer are specified,
1471 then the variable names are automatically created.
1473 The MISSING subcommand determines how user missing values are to be
1474 treated. A setting of EXCLUDE means that variables whose values are
1475 user-missing are to be excluded from the rank scores. A setting of
1476 INCLUDE means they are to be included. The default is EXCLUDE.
1478 @include regression.texi
1482 @section RELIABILITY
1488 /SCALE (@var{name}) = @{var_list, ALL@}
1489 /MODEL=@{ALPHA, SPLIT[(N)]@}
1490 /SUMMARY=@{TOTAL,ALL@}
1491 /MISSING=@{EXCLUDE,INCLUDE@}
1494 @cindex Cronbach's Alpha
1495 The @cmd{RELIABILTY} command performs reliability analysis on the data.
1497 The VARIABLES subcommand is required. It determines the set of variables
1498 upon which analysis is to be performed.
1500 The SCALE subcommand determines which variables reliability is to be
1501 calculated for. If it is omitted, then analysis for all variables named
1502 in the VARIABLES subcommand will be used.
1503 Optionally, the @var{name} parameter may be specified to set a string name
1506 The MODEL subcommand determines the type of analysis. If ALPHA is specified,
1507 then Cronbach's Alpha is calculated for the scale. If the model is SPLIT,
1508 then the variables are divided into 2 subsets. An optional parameter
1509 @var{N} may be given, to specify how many variables to be in the first subset.
1510 If @var{N} is omitted, then it defaults to one half of the variables in the
1511 scale, or one half minus one if there are an odd number of variables.
1512 The default model is ALPHA.
1514 By default, any cases with user missing, or system missing values for
1516 in the VARIABLES subcommand will be omitted from analysis.
1517 The MISSING subcommand determines whether user missing values are to
1518 be included or excluded in the analysis.
1520 The SUMMARY subcommand determines the type of summary analysis to be performed.
1521 Currently there is only one type: SUMMARY=TOTAL, which displays per-item
1522 analysis tested against the totals.
1530 @cindex Receiver Operating Characteristic
1531 @cindex Area under curve
1534 ROC @var{var_list} BY @var{state_var} (@var{state_value})
1535 /PLOT = @{ CURVE [(REFERENCE)], NONE @}
1536 /PRINT = [ SE ] [ COORDINATES ]
1537 /CRITERIA = [ CUTOFF(@{INCLUDE,EXCLUDE@}) ]
1538 [ TESTPOS (@{LARGE,SMALL@}) ]
1539 [ CI (@var{confidence}) ]
1540 [ DISTRIBUTION (@{FREE, NEGEXPO @}) ]
1541 /MISSING=@{EXCLUDE,INCLUDE@}
1545 The @cmd{ROC} command is used to plot the receiver operating characteristic curve
1546 of a dataset, and to estimate the area under the curve.
1547 This is useful for analysing the efficacy of a variable as a predictor of a state of nature.
1549 The mandatory @var{var_list} is the list of predictor variables.
1550 The variable @var{state_var} is the variable whose values represent the actual states,
1551 and @var{state_value} is the value of this variable which represents the positive state.
1553 The optional subcommand PLOT is used to determine if and how the ROC curve is drawn.
1554 The keyword CURVE means that the ROC curve should be drawn, and the optional keyword REFERENCE,
1555 which should be enclosed in parentheses, says that the diagonal reference line should be drawn.
1556 If the keyword NONE is given, then no ROC curve is drawn.
1557 By default, the curve is drawn with no reference line.
1559 The optional subcommand PRINT determines which additional tables should be printed.
1560 Two additional tables are available.
1561 The SE keyword says that standard error of the area under the curve should be printed as well as
1563 In addition, a p-value under the null hypothesis that the area under the curve equals 0.5 will be
1565 The COORDINATES keyword says that a table of coordinates of the ROC curve should be printed.
1567 The CRITERIA subcommand has four optional parameters:
1569 @item The TESTPOS parameter may be LARGE or SMALL.
1570 LARGE is the default, and says that larger values in the predictor variables are to be
1571 considered positive. SMALL indicates that smaller values should be considered positive.
1573 @item The CI parameter specifies the confidence interval that should be printed.
1574 It has no effect if the SE keyword in the PRINT subcommand has not been given.
1576 @item The DISTRIBUTION parameter determines the method to be used when estimating the area
1578 There are two possibilities, @i{viz}: FREE and NEGEXPO.
1579 The FREE method uses a non-parametric estimate, and the NEGEXPO method a bi-negative
1580 exponential distribution estimate.
1581 The NEGEXPO method should only be used when the number of positive actual states is
1582 equal to the number of negative actual states.
1583 The default is FREE.
1585 @item The CUTOFF parameter is for compatibility and is ignored.
1588 The MISSING subcommand determines whether user missing values are to
1589 be included or excluded in the analysis. The default behaviour is to
1591 Cases are excluded on a listwise basis; if any of the variables in @var{var_list}
1592 or if the variable @var{state_var} is missing, then the entire case will be