5.1.1 Defining Variables

Variables come in two basic types, viz: numeric and string. Variables such as age, height and satisfaction are numeric, whereas name is a string variable. String variables are best reserved for commentary data to assist the human observer. However they can also be used for nominal or categorical data.

The following example defines two variables forename and height, and reads data into them by manual input:

PSPP> data list list /forename (A12) height.
PSPP> begin data.
data> Ahmed 188
data> Bertram 167
data> Catherine 134.231
data> David 109.1
data> end data
PSPP>

There are several things to note about this example.

• The words ‘data list list’ are an example of the DATA LIST command. See DATA LIST. It tells PSPP to prepare for reading data. The word ‘list’ intentionally appears twice. The first occurrence is part of the DATA LIST call, whilst the second tells PSPP that the data is to be read as free format data with one record per line.
• The ‘/’ character is important. It marks the start of the list of variables which you wish to define.
• The text ‘forename’ is the name of the first variable, and ‘(A12)’ says that the variable forename is a string variable and that its maximum length is 12 bytes. The second variable’s name is specified by the text ‘height’. Since no format is given, this variable has the default format. Normally the default format expects numeric data, which should be entered in the locale of the operating system. Thus, the example is correct for English locales and other locales which use a period (‘.’) as the decimal separator. However if you are using a system with a locale which uses the comma (‘,’) as the decimal separator, then you should in the subsequent lines substitute ‘.’ with ‘,’. Alternatively, you could explicitly tell PSPP that the height variable is to be read using a period as its decimal separator by appending the text ‘DOT8.3’ after the word ‘height’. For more information on data formats, see Input and Output Formats.
• Normally, PSPP displays the prompt PSPP> whenever it’s expecting a command. However, when it’s expecting data, the prompt changes to data> so that you know to enter data and not a command.
• At the end of every command there is a terminating ‘.’ which tells PSPP that the end of a command has been encountered. You should not enter ‘.’ when data is expected (ie. when the data> prompt is current) since it is appropriate only for terminating commands.

5.1.2 Listing the data

Once the data has been entered, you could type

PSPP> list /format=numbered.

to list the data. The optional text ‘/format=numbered’ requests the case numbers to be shown along with the data. It should show the following output:

Note that the numeric variable height is displayed to 2 decimal places, because the format for that variable is ‘F8.2’. For a complete description of the LIST command, see LIST.

5.1.3 Reading data from a text file

The previous example showed how to define a set of variables and to manually enter the data for those variables. Manual entering of data is tedious work, and often a file containing the data will be have been previously prepared. Let us assume that you have a file called mydata.dat containing the ascii encoded data:

Ahmed          188.00
Bertram        167.00
Catherine      134.23
David          109.10
              .
              .
              .
Zachariah      113.02

You can can tell the DATA LIST command to read the data directly from this file instead of by manual entry, with a command like:

PSPP> data list file='mydata.dat' list /forename (A12) height.

Notice however, that it is still necessary to specify the names of the variables and their formats, since this information is not contained in the file. It is also possible to specify the file’s character encoding and other parameters. For full details refer to see DATA LIST.

5.1.4 Reading data from a pre-prepared PSPP file

When working with other PSPP users, or users of other software which uses the PSPP data format, you may be given the data in a pre-prepared PSPP file. Such files contain not only the data, but the variable definitions, along with their formats, labels and other meta-data. Conventionally, these files (sometimes called “system” files) have the suffix .sav, but that is not mandatory. The following syntax loads a file called my-file.sav.

PSPP> get file='my-file.sav'.

You will encounter several instances of this in future examples.

5.1.5 Saving data to a PSPP file.

If you want to save your data, along with the variable definitions so that you or other PSPP users can use it later, you can do this with the SAVE command.

The following syntax will save the existing data and variables to a file called my-new-file.sav.

PSPP> save outfile='my-new-file.sav'.

If my-new-file.sav already exists, then it will be overwritten. Otherwise it will be created.

5.1.6 Reading data from other sources

Sometimes it’s useful to be able to read data from comma separated text, from spreadsheets, databases or other sources. In these instances you should use the GET DATA command (see GET DATA).

5.1.7 Exiting PSPP

Use the FINISH command to exit PSPP:

PSPP> finish.

5.2.1 Identifying incorrect data

Data from real sources is rarely error free. PSPP has a number of procedures which can be used to help identify data which might be incorrect.

The DESCRIPTIVES command (see DESCRIPTIVES) is used to generate simple linear statistics for a dataset. It is also useful for identifying potential problems in the data. The example file physiology.sav contains a number of physiological measurements of a sample of healthy adults selected at random. However, the data entry clerk made a number of mistakes when entering the data. The following example illustrates the use of DESCRIPTIVES to screen this data and identify the erroneous values:

PSPP> get file='/usr/local/share/pspp/examples/physiology.sav'.
PSPP> descriptives sex, weight, height.

For this example, PSPP produces the following output:

The most interesting column in the output is the minimum value. The weight variable has a minimum value of less than zero, which is clearly erroneous. Similarly, the height variable’s minimum value seems to be very low. In fact, it is more than 5 standard deviations from the mean, and is a seemingly bizarre height for an adult person.

We can look deeper into these discrepancies by issuing an additional EXAMINE command:

PSPP> examine height, weight /statistics=extreme(3).

This command produces the following additional output (in part):

From this new output, you can see that the lowest value of height is 179 (which we suspect to be erroneous), but the second lowest is 1598 which we know from DESCRIPTIVES is within 1 standard deviation from the mean. Similarly, the lowest value of weight is negative, but its second lowest value is plausible. This suggests that the two extreme values are outliers and probably represent data entry errors.

The output also identifies the case numbers for each extreme value, so we can see that cases 30 and 38 are the ones with the erroneous values.

5.2.2 Dealing with suspicious data

If possible, suspect data should be checked and re-measured. However, this may not always be feasible, in which case the researcher may decide to disregard these values. PSPP has a feature whereby data can assume the special value ‘SYSMIS’, and will be disregarded in future analysis. See Handling missing observations. You can set the two suspect values to the ‘SYSMIS’ value using the RECODE command.

PSPP> recode height (179 = SYSMIS).
PSPP> recode weight (LOWEST THRU 0 = SYSMIS).

The first command says that for any observation which has a height value of 179, that value should be changed to the SYSMIS value. The second command says that any weight values of zero or less should be changed to SYSMIS. From now on, they will be ignored in analysis. For detailed information about the RECODE command see RECODE.

If you now re-run the DESCRIPTIVES or EXAMINE commands from the previous section, you will see a data summary with more plausible parameters. You will also notice that the data summaries indicate the two missing values.

5.2.3 Inverting negatively coded variables

Data entry errors are not the only reason for wanting to recode data. The sample file hotel.sav comprises data gathered from a customer satisfaction survey of clients at a particular hotel. The following commands load the file and display its variables and associated data:

PSPP> get file='/usr/local/share/pspp/examples/hotel.sav'.
PSPP> display dictionary.

It yields the following output:

The output shows that all of the variables v1 through v5 are measured on a 5 point Likert scale, with 1 meaning “Strongly disagree” and 5 meaning “Strongly agree”. However, some of the questions are positively worded (v1, v2, v4) and others are negatively worded (v3, v5). To perform meaningful analysis, we need to recode the variables so that they all measure in the same direction. We could use the RECODE command, with syntax such as:

recode v3 (1 = 5) (2 = 4) (4 = 2) (5 = 1).

However an easier and more elegant way uses the COMPUTE command (see COMPUTE). Since the variables are Likert variables in the range (1 … 5), subtracting their value from 6 has the effect of inverting them:

compute var = 6 - var.

The following section uses this technique to recode the variables v3 and v5. After applying COMPUTE for both variables, all subsequent commands will use the inverted values.

5.2.4 Testing data consistency

A sensible check to perform on survey data is the calculation of reliability. This gives the statistician some confidence that the questionnaires have been completed thoughtfully. If you examine the labels of variables v1, v3 and v4, you will notice that they ask very similar questions. One would therefore expect the values of these variables (after recoding) to closely follow one another, and we can test that with the RELIABILITY command (see RELIABILITY). The following example shows a PSPP session where the user recodes negatively scaled variables and then requests reliability statistics for v1, v3, and v4.

PSPP> get file='/usr/local/share/pspp/examples/hotel.sav'.
PSPP> compute v3 = 6 - v3.
PSPP> compute v5 = 6 - v5.
PSPP> reliability v1, v3, v4.

This yields the following output:

As a rule of thumb, many statisticians consider a value of Cronbach’s Alpha of 0.7 or higher to indicate reliable data.

Here, the value is 0.81, which suggests a high degree of reliability among variables v1, v3 and v4, so the data and the recoding that we performed are vindicated.

5.2.5 Testing for normality

Many statistical tests rely upon certain properties of the data. One common property, upon which many linear tests depend, is that of normality — the data must have been drawn from a normal distribution. It is necessary then to ensure normality before deciding upon the test procedure to use. One way to do this uses the EXAMINE command.

In the following example, a researcher was examining the failure rates of equipment produced by an engineering company. The file repairs.sav contains the mean time between failures (mtbf) of some items of equipment subject to the study. Before performing linear analysis on the data, the researcher wanted to ascertain that the data is normally distributed.

PSPP> get file='/usr/local/share/pspp/examples/repairs.sav'.
PSPP> examine mtbf
                /statistics=descriptives.

This produces the following output:

A normal distribution has a skewness and kurtosis of zero. The skewness of mtbf in the output above makes it clear that the mtbf figures have a lot of positive skew and are therefore not drawn from a normally distributed variable. Positive skew can often be compensated for by applying a logarithmic transformation, as in the following continuation of the example:

PSPP> compute mtbf_ln = ln (mtbf).
PSPP> examine mtbf_ln
                /statistics=descriptives.

which produces the following additional output:

The COMPUTE command in the first line above performs the logarithmic transformation:

compute mtbf_ln = ln (mtbf).

Rather than redefining the existing variable, this use of COMPUTE defines a new variable mtbf_ln which is the natural logarithm of mtbf. The final command in this example calls EXAMINE on this new variable. The results show that both the skewness and kurtosis for mtbf_ln are very close to zero. This provides some confidence that the mtbf_ln variable is normally distributed and thus safe for linear analysis. In the event that no suitable transformation can be found, then it would be worth considering an appropriate non-parametric test instead of a linear one. See NPAR TESTS, for information about non-parametric tests.

5.3.1 Testing for differences of means

A common statistical test involves hypotheses about means. The T-TEST command is used to find out whether or not two separate subsets have the same mean.

A researcher suspected that the heights and core body temperature of persons might be different depending upon their sex. To investigate this, he posed two null hypotheses based on the data from physiology.sav previously encountered:

• The mean heights of males and females in the population are equal.
• The mean body temperature of males and females in the population are equal.

For the purposes of the investigation the researcher decided to use a p-value of 0.05.

In addition to the T-test, the T-TEST command also performs the Levene test for equal variances. If the variances are equal, then a more powerful form of the T-test can be used. However if it is unsafe to assume equal variances, then an alternative calculation is necessary. PSPP performs both calculations.

For the height variable, the output shows the significance of the Levene test to be 0.33 which means there is a 33% probability that the Levene test produces this outcome when the variances are equal. Had the significance been less than 0.05, then it would have been unsafe to assume that the variances were equal. However, because the value is higher than 0.05 the homogeneity of variances assumption is safe and the “Equal Variances” row (the more powerful test) can be used. Examining this row, the two tailed significance for the height t-test is less than 0.05, so it is safe to reject the null hypothesis and conclude that the mean heights of males and females are unequal.

For the temperature variable, the significance of the Levene test is 0.58 so again, it is safe to use the row for equal variances. The equal variances row indicates that the two tailed significance for temperature is 0.20. Since this is greater than 0.05 we must reject the null hypothesis and conclude that there is insufficient evidence to suggest that the body temperature of male and female persons are different.

The syntax for this analysis is:

PSPP> get file='/usr/local/share/pspp/examples/physiology.sav'.
PSPP> recode height (179 = SYSMIS).
PSPP> t-test group=sex(0,1) /variables = height temperature.

PSPP produces the following output for this syntax:

The T-TEST command tests for differences of means. Here, the height variable’s two tailed significance is less than 0.05, so the null hypothesis can be rejected. Thus, the evidence suggests there is a difference between the heights of male and female persons. However the significance of the test for the temperature variable is greater than 0.05 so the null hypothesis cannot be rejected, and there is insufficient evidence to suggest a difference in body temperature.

5.3.2 Linear Regression

Linear regression is a technique used to investigate if and how a variable is linearly related to others. If a variable is found to be linearly related, then this can be used to predict future values of that variable.

In the following example, the service department of the company wanted to be able to predict the time to repair equipment, in order to improve the accuracy of their quotations. It was suggested that the time to repair might be related to the time between failures and the duty cycle of the equipment. The p-value of 0.1 was chosen for this investigation. In order to investigate this hypothesis, the REGRESSION command was used. This command not only tests if the variables are related, but also identifies the potential linear relationship. See REGRESSION.

A first attempt includes duty_cycle:

PSPP> get file='/usr/local/share/pspp/examples/repairs.sav'.
PSPP> regression /variables = mtbf duty_cycle /dependent = mttr.

This attempt yields the following output (in part):

The coefficients in the above table suggest that the formula mttr = 9.81 + 3.1 \times mtbf + 1.09 \times duty_cycle can be used to predict the time to repair. However, the significance value for the duty_cycle coefficient is very high, which would make this an unsafe predictor. For this reason, the test was repeated, but omitting the duty_cycle variable:

PSPP> regression /variables = mtbf /dependent = mttr.

This second try produces the following output (in part):

This time, the significance of all coefficients is no higher than 0.06, suggesting that at the 0.06 level, the formula mttr = 10.5 + 3.11 \times mtbf is a reliable predictor of the time to repair.

6.7.1 Attributes of Variables

Each variable has a number of attributes, including:

Name

An identifier, up to 64 bytes long. Each variable must have a different name. See Tokens.

Some system variable names begin with ‘$’, but user-defined variables’ names may not begin with ‘$’.

The final character in a variable name should not be ‘.’, because such an identifier will be misinterpreted when it is the final token on a line: FOO. is divided into two separate tokens, ‘FOO’ and ‘.’, indicating end-of-command. See Tokens.

The final character in a variable name should not be ‘_’, because some such identifiers are used for special purposes by PSPP procedures.

As with all PSPP identifiers, variable names are not case-sensitive. PSPP capitalizes variable names on output the same way they were capitalized at their point of definition in the input.

Type

Numeric or string.

Width

(string variables only) String variables with a width of 8 characters or fewer are called short string variables. Short string variables may be used in a few contexts where long string variables (those with widths greater than 8) are not allowed.

Position

Variables in the dictionary are arranged in a specific order. DISPLAY can be used to show this order: see DISPLAY.

Initialization

Either reinitialized to 0 or spaces for each case, or left at its existing value. See LEAVE.

Missing values

Optionally, up to three values, or a range of values, or a specific value plus a range, can be specified as user-missing values. There is also a system-missing value that is assigned to an observation when there is no other obvious value for that observation. Observations with missing values are automatically excluded from analyses. User-missing values are actual data values, while the system-missing value is not a value at all. See Handling missing observations.

Variable label

A string that describes the variable. See VARIABLE LABELS.

Value label

Optionally, these associate each possible value of the variable with a string. See VALUE LABELS.

Print format

Display width, format, and (for numeric variables) number of decimal places. This attribute does not affect how data are stored, just how they are displayed. Example: a width of 8, with 2 decimal places. See Input and Output Formats.

Write format

Similar to print format, but used by the WRITE command (see WRITE).

Measurement level

One of the following:

Nominal

Each value of a nominal variable represents a distinct category. The possible categories are finite and often have value labels. The order of categories is not significant. Political parties, US states, and yes/no choices are nominal. Numeric and string variables can be nominal.

Ordinal

Ordinal variables also represent distinct categories, but their values are arranged according to some natural order. Likert scales, e.g. from strongly disagree to strongly agree, are ordinal. Data grouped into ranges, e.g. age groups or income groups, are ordinal. Both numeric and string variables can be ordinal. String values are ordered alphabetically, so letter grades from A to F will work as expected, but poor, satisfactory, excellent will not.

Scale

Scale variables are ones for which differences and ratios are meaningful. These are often values which have a natural unit attached, such as age in years, income in dollars, or distance in miles. Only numeric variables are scalar.

Variables created by COMPUTE and similar transformations, obtained from external sources, etc., initially have an unknown measurement level. Any procedure that reads the data will then assign a default measurement level. PSPP can assign some defaults without reading the data:

• Nominal, if it’s a string variable.
• Nominal, if the variable has a WKDAY or MONTH print format.
• Scale, if the variable has a DOLLAR, CCA through CCE, or time or date print format.

Otherwise, PSPP reads the data and decides based on its distribution:

• Nominal, if all observations are missing.
• Scale, if one or more valid observations are noninteger or negative.
• Scale, if no valid observation is less than 10.
• Scale, if the variable has 24 or more unique valid values. The value 24 is the default and can be adjusted (see SET SCALEMIN).

Finally, if none of the above is true, PSPP assigns the variable a nominal measurement level.

Custom attributes

User-defined associations between names and values. See VARIABLE ATTRIBUTE.

Role

The intended role of a variable for use in dialog boxes in graphical user interfaces. See VARIABLE ROLE.

6.7.2 Variables Automatically Defined by PSPP

There are seven system variables. These are not like ordinary variables because system variables are not always stored. They can be used only in expressions. These system variables, whose values and output formats cannot be modified, are described below.

$CASENUM

Case number of the case at the moment. This changes as cases are shuffled around.

$DATE

Date the PSPP process was started, in format A9, following the pattern DD-MMM-YY.

$DATE11

Date the PSPP process was started, in format A11, following the pattern DD-MMM-YYYY.

$JDATE

Number of days between 15 Oct 1582 and the time the PSPP process was started.

$LENGTH

Page length, in lines, in format F11.

$SYSMIS

System missing value, in format F1.

$TIME

Number of seconds between midnight 14 Oct 1582 and the time the active dataset was read, in format F20.

$WIDTH

Page width, in characters, in format F3.

6.7.3 Lists of variable names

To refer to a set of variables, list their names one after another. Optionally, their names may be separated by commas. To include a range of variables from the dictionary in the list, write the name of the first and last variable in the range, separated by TO. For instance, if the dictionary contains six variables with the names ID, X1, X2, GOAL, MET, and NEXTGOAL, in that order, then X2 TO MET would include variables X2, GOAL, and MET.

Commands that define variables, such as DATA LIST, give TO an alternate meaning. With these commands, TO define sequences of variables whose names end in consecutive integers. The syntax is two identifiers that begin with the same root and end with numbers, separated by TO. The syntax X1 TO X5 defines 5 variables, named X1, X2, X3, X4, and X5. The syntax ITEM0008 TO ITEM0013 defines 6 variables, named ITEM0008, ITEM0009, ITEM0010, ITEM0011, ITEM0012, and ITEM00013. The syntaxes QUES001 TO QUES9 and QUES6 TO QUES3 are invalid.

After a set of variables has been defined with DATA LIST or another command with this method, the same set can be referenced on later commands using the same syntax.

6.7.4 Input and Output Formats

An input format describes how to interpret the contents of an input field as a number or a string. It might specify that the field contains an ordinary decimal number, a time or date, a number in binary or hexadecimal notation, or one of several other notations. Input formats are used by commands such as DATA LIST that read data or syntax files into the PSPP active dataset.

Every input format corresponds to a default output format that specifies the formatting used when the value is output later. It is always possible to explicitly specify an output format that resembles the input format. Usually, this is the default, but in cases where the input format is unfriendly to human readability, such as binary or hexadecimal formats, the default output format is an easier-to-read decimal format.

Every variable has two output formats, called its print format and write format. Print formats are used in most output contexts; write formats are used only by WRITE (see WRITE). Newly created variables have identical print and write formats, and FORMATS, the most commonly used command for changing formats (see FORMATS), sets both of them to the same value as well. Thus, most of the time, the distinction between print and write formats is unimportant.

Input and output formats are specified to PSPP with a format specification of the form TYPEw or TYPEw.d, where TYPE is one of the format types described later, w is a field width measured in columns, and d is an optional number of decimal places. If d is omitted, a value of 0 is assumed. Some formats do not allow a nonzero d to be specified.

The following sections describe the input and output formats supported by PSPP.

• Basic Numeric Formats
• Custom Currency Formats
• Legacy Numeric Formats
• Binary and Hexadecimal Numeric Formats
• Time and Date Formats
• Date Component Formats
• String Formats

SET CCx="string".

SET CCA="-,,,".  /* Same as COMMA.
SET CCB="-...".  /* Same as DOT.
SET CCC="-,$,,". /* Same as DOLLAR.
SET CCD="-,,%,". /* Like PCT, but groups with commas.

SET CCA=",EUR,,-".   /* Euro.
SET CCB="(,USD ,,)". /* US dollar.
SET CCC="-.R$..".    /* Brazilian real.
SET CCD="-,, NIS,".  /* Israel shekel.
SET CCE="-.Rp'. ..". /* Indonesia Rupiah.

0123456789
{ABCDEFGHI
}JKLMNOPQR

6.7.5 Scratch Variables

Most of the time, variables don’t retain their values between cases. Instead, either they’re being read from a data file or the active dataset, in which case they assume the value read, or, if created with COMPUTE or another transformation, they’re initialized to the system-missing value or to blanks, depending on type.

However, sometimes it’s useful to have a variable that keeps its value between cases. You can do this with LEAVE (see LEAVE), or you can use a scratch variable. Scratch variables are variables whose names begin with an octothorpe (‘#’).

Scratch variables have the same properties as variables left with LEAVE: they retain their values between cases, and for the first case they are initialized to 0 or blanks. They have the additional property that they are deleted before the execution of any procedure. For this reason, scratch variables can’t be used for analysis. To use a scratch variable in an analysis, use COMPUTE (see COMPUTE) to copy its value into an ordinary variable, then use that ordinary variable in the analysis.

7.7.1 Mathematical Functions

Advanced mathematical functions take numeric arguments and produce numeric results.

Function: EXP (exponent) ¶

Returns e (approximately 2.71828) raised to power exponent.

Function: LG10 (number) ¶

Takes the base-10 logarithm of number. If number is not positive, the result is system-missing.

Function: LN (number) ¶

Takes the base-e logarithm of number. If number is not positive, the result is system-missing.

Function: LNGAMMA (number) ¶

Yields the base-e logarithm of the complete gamma of number. If number is a negative integer, the result is system-missing.

Function: SQRT (number) ¶

Takes the square root of number. If number is negative, the result is system-missing.

7.7.2 Miscellaneous Mathematical Functions

Miscellaneous mathematical functions take numeric arguments and produce numeric results.

Function: ABS (number) ¶

Results in the absolute value of number.

Function: MOD (numerator, denominator) ¶

Returns the remainder (modulus) of numerator divided by denominator. If numerator is 0, then the result is 0, even if denominator is missing. If denominator is 0, the result is system-missing.

Function: MOD10 (number) ¶

Returns the remainder when number is divided by 10. If number is negative, MOD10(number) is negative or zero.

Function: RND (number [, mult[, fuzzbits]]) ¶

Rounds number and rounds it to a multiple of mult (by default 1). Halves are rounded away from zero, as are values that fall short of halves by less than fuzzbits of errors in the least-significant bits of number. If fuzzbits is not specified then the default is taken from SET FUZZBITS (see SET FUZZBITS), which is 6 unless overridden.

Function: TRUNC (number [, mult[, fuzzbits]]) ¶

Rounds number to a multiple of mult, toward zero. For the default mult of 1, this is equivalent to discarding the fractional part of number. Values that fall short of a multiple of mult by less than fuzzbits of errors in the least-significant bits of number are rounded away from zero. If fuzzbits is not specified then the default is taken from SET FUZZBITS (see SET FUZZBITS), which is 6 unless overridden.

7.7.3 Trigonometric Functions

Trigonometric functions take numeric arguments and produce numeric results.

Function: ARCOS (number) ¶

Function: ACOS (number) ¶

Takes the arccosine, in radians, of number. Results in system-missing if number is not between -1 and 1 inclusive. This function is a PSPP extension.

Function: ARSIN (number) ¶

Function: ASIN (number) ¶

Takes the arcsine, in radians, of number. Results in system-missing if number is not between -1 and 1 inclusive.

Function: ARTAN (number) ¶

Function: ATAN (number) ¶

Takes the arctangent, in radians, of number.

Function: COS (angle) ¶

Takes the cosine of angle which should be in radians.

Function: SIN (angle) ¶

Takes the sine of angle which should be in radians.

Function: TAN (angle) ¶

Takes the tangent of angle which should be in radians. Results in system-missing at values of angle that are too close to odd multiples of \pi/2. Portability: none.

7.7.4 Missing-Value Functions

Missing-value functions take various numeric arguments and yield various types of results. Except where otherwise stated below, the normal rules of evaluation apply within expression arguments to these functions. In particular, user-missing values for numeric variables are converted to system-missing values.

Function: MISSING (expr) ¶

When expr is simply the name of a numeric variable, returns 1 if the variable has the system-missing value or if it is user-missing. For any other value 0 is returned. If expr takes another form, the function returns 1 if the value is system-missing, 0 otherwise.

Function: NMISS (expr [, expr]…) ¶

Each argument must be a numeric expression. Returns the number of system-missing values in the list, which may include variable ranges using the var1 TO var2 syntax.

Function: NVALID (expr [, expr]…) ¶

Each argument must be a numeric expression. Returns the number of values in the list that are not system-missing. The list may include variable ranges using the var1 TO var2 syntax.

Function: SYSMIS (expr) ¶

Returns 1 if expr has the system-missing value, 0 otherwise.

Function: VALUE (variable) ¶

Function: VALUE (vector(index)) ¶

Prevents the user-missing values of the variable or vector element from being transformed into system-missing values, and always results in its actual value, whether it is valid, user-missing, or system-missing.

7.7.5 Set-Membership Functions

Set membership functions determine whether a value is a member of a set. They take a set of numeric arguments or a set of string arguments, and produce Boolean results.

String comparisons are performed according to the rules given in Relational Operators. User-missing string values are treated as valid values.

Function: ANY (value, set [, set]…) ¶

Returns true if value is equal to any of the set values, and false otherwise. For numeric arguments, returns system-missing if value is system-missing or if all the values in set are system-missing. If value

Function: RANGE (value, low, high [, low, high]…) ¶

Returns true if value is in any of the intervals bounded by low and high inclusive, and false otherwise. low and high must be given in pairs. Returns system-missing if any high is less than its low or, for numeric arguments, if value is system-missing or if all the low-high pairs contain one (or two) system-missing values. A pair does not match value if either low or high is missing, even if value equals the non-missing endpoint.

7.7.6 Statistical Functions

Statistical functions compute descriptive statistics on a list of values. Some statistics can be computed on numeric or string values; other can only be computed on numeric values. Their results have the same type as their arguments. The current case’s weighting factor (see WEIGHT) has no effect on statistical functions.

These functions’ argument lists may include entire ranges of variables using the var1 TO var2 syntax.

Unlike most functions, statistical functions can return non-missing values even when some of their arguments are missing. Most statistical functions, by default, require only 1 non-missing value to have a non-missing return, but CFVAR, SD, and VARIANCE require 2. These defaults can be increased (but not decreased) by appending a dot and the minimum number of valid arguments to the function name. For example, MEAN.3(X, Y, Z) would only return non-missing if all of ‘X’, ‘Y’, and ‘Z’ were valid.

Function: CFVAR (number, number[, …]) ¶

Results in the coefficient of variation of the values of number. (The coefficient of variation is the standard deviation divided by the mean.)

Function: MAX (value, value[, …]) ¶

Results in the value of the greatest value. The values may be numeric or string.

Function: MEAN (number, number[, …]) ¶

Results in the mean of the values of number.

Function: MEDIAN (number, number[, …]) ¶

Results in the median of the values of number. Given an even number of nonmissing arguments, yields the mean of the two middle values.

Function: MIN (number, number[, …]) ¶

Results in the value of the least value. The values may be numeric or string.

Function: SD (number, number[, …]) ¶

Results in the standard deviation of the values of number.

Function: SUM (number, number[, …]) ¶

Results in the sum of the values of number.

Function: VARIANCE (number, number[, …]) ¶

Results in the variance of the values of number.

7.7.7 String Functions

String functions take various arguments and return various results.

Function: CONCAT (string, string[, …]) ¶

Returns a string consisting of each string in sequence. CONCAT("abc", "def", "ghi") has a value of "abcdefghi". The resultant string is truncated to a maximum of 32767 bytes.

Function: INDEX (haystack, needle) ¶

Function: RINDEX (haystack, needle) ¶

Returns a positive integer indicating the position of the first (for INDEX) or last (for RINDEX) occurrence of needle in haystack. Returns 0 if haystack does not contain needle. Returns 1 if needle is the empty string.

Function: INDEX (haystack, needles, needle_len) ¶

Function: RINDEX (haystack, needle, needle_len) ¶

Divides needles into multiple needles, each with length needle_len, which must be a positive integer that evenly divides the length of needles. Searches haystack for the occurrences of each needle and returns a positive integer indicating the byte index of the beginning of the first (for INDEX) or last (for RINDEX) needle it finds. Returns 0 if haystack does not contain any of the needles, or if needles is the empty string.

Function: LENGTH (string) ¶

Returns the number of bytes in string.

Function: LOWER (string) ¶

Returns a string identical to string except that all uppercase letters are changed to lowercase letters. The definitions of “uppercase” and “lowercase” are system-dependent.

Function: LPAD (string, length[, padding]) ¶

Function: RPAD (string, length[, padding]) ¶

If string is at least length bytes long, these functions return string unchanged. Otherwise, they return string padded with padding on the left side (for LPAD) or right side (for RPAD) to length bytes. These functions report an error and return string unchanged if length is missing or bigger than 32767.

The padding argument must not be an empty string and defaults to a space if not specified. If its length does not evenly fit the amount of space needed for padding, the returned string will be shorter than length.

Function: LTRIM (string[, padding]) ¶

Function: RTRIM (string[, padding]) ¶

These functions return string, after removing leading (for LTRIM) or trailing (for RTRIM) copies of padding. If padding is omitted, these functions remove spaces (but not tabs or other white space). These functions return string unchanged if padding is the empty string.

Function: NUMBER (string, format) ¶

Returns the number produced when string is interpreted according to format specifier format. If the format width w is less than the length of string, then only the first w bytes in string are used, e.g. NUMBER("123", F3.0) and NUMBER("1234", F3.0) both have value 123. If w is greater than string’s length, then it is treated as if it were right-padded with spaces. If string is not in the correct format for format, system-missing is returned.

Function: REPLACE (haystack, needle, replacement[, n]) ¶

Returns string haystack with instances of needle replaced by replacement. If nonnegative integer n is specified, it limits the maximum number of replacements; otherwise, all instances of needle are replaced.

Function: STRING (number, format) ¶

Returns a string corresponding to number in the format given by format specifier format. For example, STRING(123.56, F5.1) has the value "123.6".

Function: STRUNC (string, n) ¶

Returns string, first trimming it to at most n bytes, then removing trailing spaces (but not tabs or other white space). Returns an empty string if n is zero or negative, or string unchanged if n is missing.

Function: SUBSTR (string, start) ¶

Returns a string consisting of the value of string from position start onward. Returns an empty string if start is system-missing, less than 1, or greater than the length of string.

Function: SUBSTR (string, start, count) ¶

Returns a string consisting of the first count bytes from string beginning at position start. Returns an empty string if start or count is system-missing, if start is less than 1 or greater than the number of bytes in string, or if count is less than 1. Returns a string shorter than count bytes if start + count - 1 is greater than the number of bytes in string. Examples: SUBSTR("abcdefg", 3, 2) has value "cd"; SUBSTR("nonsense", 4, 10) has the value "sense".

Function: UPCASE (string) ¶

Returns string, changing lowercase letters to uppercase letters.

7.7.8 Time & Date Functions

For compatibility, PSPP considers dates before 15 Oct 1582 invalid. Most time and date functions will not accept earlier dates.

• How times & dates are defined and represented
• Functions that Produce Times
• Functions that Examine Times
• Functions that Produce Dates
• Functions that Examine Dates
• Time and Date Arithmetic

          10 minutes                        600
          1 hour                          3,600
          1 day, 3 hours, 10 seconds     97,210
          40 days                     3,456,000

              15 Oct 1582                86,400
               4 Jul 1776         6,113,318,400
               1 Jan 1900        10,010,390,400
               1 Oct 1978        12,495,427,200
              24 Aug 1995        13,028,601,600

7.7.9 Miscellaneous Functions

Function: LAG (variable[, n]) ¶

variable must be a numeric or string variable name. LAG yields the value of that variable for the case n before the current one. Results in system-missing (for numeric variables) or blanks (for string variables) for the first n cases.

LAG obtains values from the cases that become the new active dataset after a procedure executes. Thus, LAG will not return values from cases dropped by transformations such as SELECT IF, and transformations like COMPUTE that modify data will change the values returned by LAG. These are both the case whether these transformations precede or follow the use of LAG.

If LAG is used before TEMPORARY, then the values it returns are those in cases just before TEMPORARY. LAG may not be used after TEMPORARY.

If omitted, ncases defaults to 1. Otherwise, ncases must be a small positive constant integer. There is no explicit limit, but use of a large value will increase memory consumption.

Function: YRMODA (year, month, day) ¶

year is a year, either between 0 and 99 or at least 1582. Unlike other PSPP date functions, years between 0 and 99 always correspond to 1900 through 1999. month is a month between 1 and 13. day is a day between 0 and 31. A day of 0 refers to the last day of the previous month, and a month of 13 refers to the first month of the next year. year must be in range. year, month, and day must all be integers.

YRMODA results in the number of days between 15 Oct 1582 and the date specified, plus one. The date passed to YRMODA must be on or after 15 Oct 1582. 15 Oct 1582 has a value of 1.

Function: VALUELABEL (variable) ¶

Returns a string matching the label associated with the current value of variable. If the current value of variable has no associated label, then this function returns the empty string. variable may be a numeric or string variable.

7.7.10 Statistical Distribution Functions

PSPP can calculate several functions of standard statistical distributions. These functions are named systematically based on the function and the distribution. The table below describes the statistical distribution functions in general:

PDF.dist (x[, param…])

Probability density function for dist. The domain of x depends on dist. For continuous distributions, the result is the density of the probability function at x, and the range is nonnegative real numbers. For discrete distributions, the result is the probability of x.

CDF.dist (x[, param…])

Cumulative distribution function for dist, that is, the probability that a random variate drawn from the distribution is less than x. The domain of x depends dist. The result is a probability.

SIG.dist (x[, param…)

Tail probability function for dist, that is, the probability that a random variate drawn from the distribution is greater than x. The domain of x depends dist. The result is a probability. Only a few distributions include an SIG function.

IDF.dist (p[, param…])

Inverse distribution function for dist, the value of x for which the CDF would yield p. The value of p is a probability. The range depends on dist and is identical to the domain for the corresponding CDF.

RV.dist ([param…])

Random variate function for dist. The range depends on the distribution.

NPDF.dist (x[, param…])

Noncentral probability density function. The result is the density of the given noncentral distribution at x. The domain of x depends on dist. The range is nonnegative real numbers. Only a few distributions include an NPDF function.

NCDF.dist (x[, param…])

Noncentral cumulative distribution function for dist, that is, the probability that a random variate drawn from the given noncentral distribution is less than x. The domain of x depends dist. The result is a probability. Only a few distributions include an NCDF function.

The individual distributions are described individually below.

• Continuous Distributions
• Discrete Distributions