Universals
The 35 names of system variables begin with a dollar sign ($). You cannot modify a system variable or alter its print or write format. Except for these restrictions, you can use system variables anywhere that a normal variable is used in the transformation language. System variables are not available for procedures.
$CASENUM Current case sequence number. For each case, $CASENUM is the number of cases read up to and including that case. The format is F8.0. The value of $CASENUM is not necessarily the row number in a Data Editor window (available in windowed environments), and the value changes if the file is sorted or new cases are inserted before the end of the file. System-missing value. The system-missing value displays as a period (.) or whatever is used as the decimal point. Current date in number of days from October 14, 1582 (day 1 of the Gregorian calendar). The format is F6.0. Current date in international date format with two-digit year. The format is A9 in the form dd-mmm-yy. Current date in international date format with four-digit year. The format is A11 in the form dd-mmm-yyyy. Current date and time. $TIME represents the number of seconds from midnight, October 14, 1582, to the date and time when the transformation command is executed. The format is F20. You can display this as a date in a number of different date formats. You can also use it in date and time functions. The current page length. The format is F11.0. For more information, see SET. The current page width. The format is F3.0. For more information, see SET.
$SYSMIS $JDATE $DATE $DATE11 $TIME
$LENGTH $WIDTH
Variable Types and Formats
SPSS recognizes two basic variable types:
String. Also referred to alphanumeric. String values are stored as codes listed in the SPSS
character set. For more information, see IMPORT/EXPORT Character Sets on p. 1934.
Numeric. Numeric values are stored internally as double-precision floating-point numbers.
Variable formats determine how SPSS reads raw data into storage and how it displays and writes out values. For example, all dates and times are stored internally as numeric values, but you can use date and time format specifications to both read and display date and time values in standard date and time formats. The following sections provide details on how formats are specified and how those formats affect how data are read, displayed, and written.
Input and Output Formats
Values are read according to their input format and displayed according to their output format. The input and output formats differ in several ways. The input format is either specified or implied on the DATA LIST, GET DATA, or other data definition commands. It is in effect only when SPSS builds cases in an active dataset.
36 Universals
Output formats are automatically generated from input formats, with output formats expanded to include punctuation characters, such as decimal indicators, grouping symbols, and dollar signs. For example, an input format of DOLLAR7.2 will generate an output format of DOLLAR10.2 to accommodate the dollar sign, grouping symbol (comma), and decimal indicator (period). The formats (specified or default) on NUMERIC, STRING, COMPUTE, or other commands that create new variables are output formats. You must specify adequate widths to accommodate all punctuation characters. The output format is in effect during the entire working session (unless explicitly changed) and is saved in the dictionary of an SPSS-format data file. Output formats for numeric variables can be changed with FORMATS, PRINT FORMATS, and WRITE FORMATS. The width for string variables cannot be changed with command syntax. However, you can use STRING to declare a new variable with the desired format and then use COMPUTE to copy values from the existing string variable into the new variable. The format type cannot be changed from string to numeric, or vice versa, with command syntax. However, you can use RECODE to recode values from one variable into another variable of a different type.
String Variable Formats
The values of string variables can contain numbers, letters, and special characters and can be up to 32,767 characters long. SPSS differentiates between long strings and short strings. A short string is a string variable with a maximum width of eight bytes. A long string is a string variable with a maximum width greater than eight bytes. Long strings cannot have user-missing values, and some procedures that accept short string variables do not accept long string variables. System-missing values cannot be generated for string variables, since any character is a legal string value. When a transformation command that creates or modifies a string variable yields a missing or undefined result, a null string is assigned. The variable displays as blanks and is not treated as missing. String formats are used to read and write string variables. The input values can be alphanumeric characters (A format) or the hexadecimal representation of alphanumeric characters (AHEX format). For fixed-format raw data, the width can be explicitly specified on commands such as DATA LIST and GET DATA or implied if column-style specifications are used. For freefield data, the default width is 1; if the input string may be longer, w must be explicitly specified. Input strings shorter than the specified width are right-padded with blanks. The output format for a string variable is always A. The width is determined by the input format or the format assigned on the STRING command. Once defined, the width of a string variable cannot be changed.
37 Universals
A Format (Standard Characters)
The A format is used to read standard characters. Characters can include letters, numbers, punctuation marks, blanks, and most other characters on your keyboard. Numbers entered as values for string variables cannot be used in calculations unless you convert them to numeric format with the NUMBER function. For more information, see String/Numeric Conversion Functions on p. 79. Fixed data: With fixed-format input data, any punctuation--including leading, trailing, and embedded blanks--within the column specifications is included in the string value. For example, a string value of
Mr. Ed
(with one embedded blank) is distinguished from a value of
Mr. Ed
(with two embedded blanks). It is also distinguished from a string value of
MR. ED
(all upper case), and all three are treated as separate values. These can be important considerations for any procedures, transformations, or data selection commands involving string variables. Consider the following example:
DATA LIST FIXED /ALPHAVAR 1-10 (A). BEGIN DATA Mr. Ed Mr. Ed MR. ED Mr. Ed Mr. Ed END DATA. AUTORECODE ALPHAVAR /INTO NUMVAR. LIST.
AUTORECODE recodes the values into consecutive integers. The following figure shows the recoded values.
Figure 2-3 Different string values illustrated
ALPHAVAR Mr. Ed Mr. Ed MR. ED Mr. Ed Mr. Ed NUMVAR 4 4 2 3 1
38 Universals
AHEX Format (Hexadecimal Characters)
The AHEX format is used to read the hexadecimal representation of standard characters. Each set of two hexadecimal characters represents one standard character. For codes used on different operating systems, see IMPORT/EXPORT Character Sets on p. 1934. The w specification refers to columns of the hexadecimal representation and must be an even number. Leading, trailing, and embedded blanks are not allowed, and only valid hexadecimal characters can be used in input values. For some operating systems (e.g., IBM CMS), letters in hexadecimal values must be upper case. The default output format for variables read with the AHEX input format is the A format. The default width is half the specified input width. For example, an input format of AHEX14 generates an output format of A7. Used as an output format, the AHEX format displays the printable characters in the hexadecimal characters specific to your system. The following commands run on a UNIX system--where A=41 (decimal 65), a=61 (decimal 97), and so on--produce the output shown below:
DATA LIST FIXED /A,B,C,D,E,F,G,H,I,J,K,L,M,N,O,P,Q,R,S,T,U,V,W,X,Y,Z 1-26 (A). FORMATS ALL (AHEX2). BEGIN DATA ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz END DATA. LIST.
Figure 2-4 Display of hexadecimal representation of the character set with AHEX format
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A
Numeric Variable Formats
By default, if no format is explicitly specified, commands that read raw data--such as DATA LIST and GET DATA--assume that variables are numeric with an F format type. The default width depends on whether the data are in fixed or freefield format. For a discussion of fixed data and freefield data, see DATA LIST on p. 461. Numeric variables created by COMPUTE, COUNT, or other commands that create numeric variables are assigned a format type of F8.2(or the default numeric format defined on SET FORMAT). If a data value exceeds its width specification, SPSS makes an attempt to display some value nevertheless. It first rounds the decimals, then takes out punctuation characters, then tries scientific notation, and if there is still not enough space, produces asterisks (***), indicating that a value is present but cannot be displayed in the assigned width.
39 Universals
The output format does not affect the value stored in the file. A numeric value is always stored in double precision. For default numeric (F) format and scientific notation (E) format, the decimal indicator of the input data from text data sources (read by commands such as DATA LIST and GET DATA) must match the SPSS locale decimal indicator (period or comma). Use SET DECIMAL to set the decimal indicator. Use SHOW DECIMAL to display the current decimal indicator.
F, N, and E Formats
The following table lists the formats most commonly used to read in and write out numeric data. Format names are followed by total width (w) and an optional number of decimal positions (d). For example, a format of F5.2 represents a numeric value with a total width of 5, including two decimal positions and a decimal indicator.
Table 2-1 Common numeric formats
Format Description type
Sample Sample Output for fixed format input input Format Value
Output for freefield input Format Value F5.0 1234 1* F6.2 1234.0 1.23 F5.0 123 123 F6.2 12345 . E10.3 1.234E+06 1.234E+03
Fw.d
Standard numeric
F5.0
1234 1.234
F5.0
1234 1*
F5.2
1234 1.234
F6.2
12.34 1.23
Nw.d
Restricted numeric
N5.0
00123 123
F5.0
123 .
N5.2
12345 12.34
F6.2
123.45 . 1.234E+06 1.234E+03
Ew.d
Scientific notation
E8.0
1234E3 E10.3 1234
* Only the display is truncated. The value is stored in full precision. System-missing value. Scientific notation is accepted in input data with F, COMMA, DOLLAR, DOT, and PCT formats. The
same rules apply as specified below. For fixed data: If a value has no coded decimal point but the input format specifies decimal positions, the rightmost positions are interpreted as implied decimal digits. For example, if the input F format specifies two decimal digits, the value 1234 is interpreted as 12.34; however, the value 123.4 is still interpreted as 123.4.
40 Universals
With the N format, decimal places can only be implied. Only unsigned integers are allowed as input values. Values not padded with leading zeros to the specified width or those containing decimal points are assigned the system-missing value. This format is useful for reading and checking values that should be integers containing leading zeros. The E format reads all forms of scientific notation. If the sign is omitted, + is assumed. If the sign (+ or ) is specified before the exponent, the E or D can be omitted. A single space is permitted after the E or D and/or after the sign. If both the sign and the letter E or D are omitted, implied decimal places are assumed. For example, 1.234E3, 1.234+3, 1.234E+3, 1.234D3, 1.234D+3, 1.234E 3, and 1234 are all legitimate values. Only the last value can imply decimal places.
E format input values can be up to 40 characters wide and include up to 15 decimal positions.
The default output width (w) for the E format is either the specified input width or the number of specified decimal positions plus 7 (d+7), whichever is greater. The minimum width is 10 and the minimum decimal places are 3. For freefield data:
F format w and d specifications do not affect how data are read. They only determine the
output formats (expanded, if necessary). 1234 is always read as 1234 in freefield data, but a specified F5.2 format will be expanded to F6.2 and the value will be displayed as 1234.0 (the last decimal place is rounded because of lack of space). When the N format is used for freefield data, input values with embedded decimal indicators are assigned the system-missing value, but integer input values without leading zeroes are treated as valid. For example, with an input format of N5.0, a value of 123 is treated the same as a value of 00123, but a value of 12.34 is assigned the system-missing value. The E format for freefield data follows the same rules as for fixed data except that no blank space is permitted in the value. Thus, 1.234E3 and 1.234+3 are allowed, but the value 1.234 3 will cause mistakes when the data are read. The default output E format and the width and decimal place limitations are the same as with fixed data.
N (Restricted Numeric) Output Format
N format input values are assigned an F output format. To display, print, and write N format values with leading zeroes, use the FORMATS command to specify N as the output format. For more information, see FORMATS on p. 653.
COMMA, DOT, DOLLAR, and PCT Formats
The numeric formats listed below read and write data with embedded punctuation characters and symbols, such as commas, dots, and dollar and percent signs. The input data may or may not contain such characters. The data values read in are stored as numbers but displayed using the appropriate formats.
DOLLAR. Numeric values with a leading dollar sign, a comma used as the grouping separator,
and a period used as the decimal indicator. For example, $1,234.56.
41 Universals
COMMA. Numeric values with a comma used as the grouping separator and a period used as
decimal indicator. For example, 1,234.56.
DOT. Numeric values with a period used as the grouping separator and a comma used as the
decimal indicator. For example, 1.234,56.
PCT. Numeric values with a trailing percent sign. For example, 123.45%.
The input data values may or may not contain the punctuation characters allowed by the specified format, but the data values may not contain characters not allowed by the format. For example, with a DOLLAR input format, input values of 1234.56, 1,234.56, and $1,234.56 are all valid and stored internally as the same value--but with a COMMA input format, the input value with a leading dollar sign would be assigned the system-missing value.
DATA LIST LIST (" ") /dollarVar (DOLLAR9.2) commaVar (COMMA9.2) dotVar (DOT9.2) pctVar (PCT9.2). BEGIN DATA 1234 1234 1234 1234 $1,234.00 1,234.00 1.234,00 1234.00% END DATA. LIST. Figure 2-5 Output illustrating DOLLAR, COMMA, DOT, and PCT formats
dollarVar $1,234.00 $1,234.00 commaVar 1,234.00 1,234.00 dotVar 1.234,00 1.234,00 pctVar 1234.00% 1234.00%
Other formats that use punctuation characters and symbols are date and time formats and custom currency formats. For more information on date and time formats, see Date and Time Formats on p. 44. Custom currency formats are output formats only, and are defined with the SET command.
Binary and Hexadecimal Formats
SPSS is capable of reading and writing data in formats used by a number of programming languages such as PL/I, COBOL, FORTRAN, and Assembler. The data can be binary, hexadecimal, or zoned decimal. Formats described in this section can be used both as input formats and output formats, but with fixed data only. The described formats are not available on all systems. Consult the SPSS Base User's Guide for your version of SPSS for details. The default output format for all formats described in this section is an equivalent F format, allowing the maximum number of columns for values with symbols and punctuation. To change the default, use FORMATS or WRITE FORMATS. IBw.d (integer binary): The IB format reads fields that contain fixed-point binary (integer) data. The data might be generated by COBOL using COMPUTATIONAL data items, by FORTRAN using INTEGER*2 or INTEGER*4, or by Assembler using fullword and halfword items. The general format is a signed binary number that is 16 or 32 bits in length.
42 Universals
The general syntax for the IB format is IBw.d, where w is the field width in bytes (omitted for column-style specifications) and d is the number of digits to the right of the decimal point. Since the width is expressed in bytes and the number of decimal positions is expressed in digits, d can be greater than w. For example, both of the following commands are valid:
DATA LIST FIXED /VAR1 (IB4.8). DATA LIST FIXED /VAR1 1-4 (IB,8).
Widths of 2 and 4 represent standard 16-bit and 32-bit integers, respectively. Fields read with the IB format are treated as signed. For example, the one-byte binary value 11111111 would be read as 1. PIBw.d (positive integer binary): The PIB format is essentially the same as IB except that negative numbers are not allowed. This restriction allows one additional bit of magnitude. The same one-byte value 11111111 would be read as 255. PIBHEXw (hexadecimal of PIB): The PIBHEX format reads hexadecimal numbers as unsigned integers and writes positive integers as hexadecimal numbers. The general syntax the for PIBHEX format is PIBHEXw, where w indicates the total number of hexadecimal characters. The w specification must be an even number with a maximum of 16. For input data, each hexadecimal number must consist of the exact number of characters. No signs, decimal points, or leading and trailing blanks are allowed. For some operating systems (such as IBM CMS), hexadecimal characters must be upper case. The following example illustrates the kind of data that the PIBHEX format can read:
DATA LIST FIXED /VAR1 1-4 (PIBHEX) VAR2 6-9 (PIBHEX) VAR3 11-14 (PIBHEX). BEGIN DATA 0001 0002 0003 0004 0005 0006 0007 0008 0009 000A 000B 000C 000D 000E 000F 00F0 0B2C FFFF END DATA. LIST.
The values for VAR1, VAR2, and VAR3 are listed in the figure below. The PIBHEX format can also be used to write decimal values as hexadecimal numbers, which may be useful for programmers.
Figure 2-6 Output displaying values read in PIBHEX format
VAR1 1 4 7 10 13 240 VAR2 2 5 8 11 14 2860 VAR3 3 6 9 12 15 65535
43 Universals
Zw.d (zoned decimal): The Z format reads data values that contain zoned decimal data. Such numbers may be generated by COBOL systems using DISPLAY data items, by PL/I systems using PICTURE data items, or by Assembler using zoned decimal data items. In zoned decimal format, one digit is represented by one byte, generally hexadecimal F1 representing 1, F2 representing 2, and so on. The last byte, however, combines the sign for the number with the last digit. In the last byte, hexadecimal A, F, or C assigns +, and B, D, or E assigns . For example, hexadecimal D1 represents 1 for the last digit and assigns the minus sign () to the number. The general syntax of the Z format is Zw.d, where w is the total number of bytes (which is the same as columns) and d is the number of decimals. For input data, values can appear anywhere within the column specifications. Both leading and trailing blanks are allowed. Decimals can be implied by the input format specification or explicitly coded in the data. Explicitly coded decimals override the input format specifications. The following example illustrates how the Z format reads zoned decimals in their printed forms on IBM mainframe and PC systems. The printed form for the sign zone (A to I for +1 to +9, and so on) may vary from system to system.
DATA LIST FIXED /VAR1 1-5 (Z) VAR2 7-11 (Z,2) VAR3 13-17 (Z) VAR4 19-23 (Z,2) VAR5 25-29 (Z) VAR6 31-35 (Z,2). BEGIN DATA 1234A 1234A 1234B 1234B 1234C 1234C 1234D 1234D 1234E 1234E 1234F 1234F 1234G 1234G 1234H 1234H 1234I 1234I 1234J 1234J 1234K 1234K 1234L 1234L 1234M 1234M 1234N 1234N 1234O 1234O 1234P 1234P 1234Q 1234Q 1234R 1234R 1234{ 1234{ 1234} 1234} 1.23M 1.23M END DATA. LIST.
The values for VAR1 to VAR6 are listed in the following figure.
Figure 2-7 Output displaying values read in Z format
VAR1 VAR2 VAR3 VAR4 VAR5 VAR6
12341 123.41 12342 123.42 12343 123.43 12344 123.44 12345 123.45 12346 123.46 12347 123.47 12348 123.48 12349 123.49 -12341 -123.41 -12342 -123.42 -12343 -123.43 -12344 -123.44 -12345 -123.45 -12346 -123.46 -12347 -123.47 -12348 -123.48 -12349 -123.49 12340 123.40 -12340 -123.40 -1 -1.23
The default output format for the Z format is the equivalent F format, as shown in the figure. The default output width is based on the input width specification plus one column for the sign and one column for the implied decimal point (if specified). For example, an input format of Z4.0 generates an output format of F5.0, and an input format of Z4.2 generates an output format of F6.2.
44 Universals
Pw.d (packed decimal): The P format is used to read fields with packed decimal numbers. Such numbers are generated by COBOL using COMPUTATIONAL3 data items and by Assembler using packed decimal data items. The general format of a packed decimal field is two four-bit digits in each byte of the field except the last. The last byte contains a single digit in its four leftmost bits and a four-bit sign in its rightmost bits. If the last four bits are 1111 (hexadecimal F), the value is positive; if they are 1101 (hexadecimal D), the value is negative. One byte under the P format can represent numbers from 9 to 9. The general syntax of the P format is Pw.d, where w is the number of bytes (not digits) and d is the number of digits to the right of the implied decimal point. The number of digits in a field is (2*w1). PKw.d (unsigned packed decimal): The PK format is essentially the same as P except that there is no sign. That is, even the rightmost byte contains two digits, and negative data cannot be represented. One byte under the PK format can represent numbers from 0 to 99. The number of digits in a field is 2*w. RBw (real binary): The RB format is used to read data values that contain internal format floating-point numbers. Such numbers are generated by COBOL using COMPUTATIONAL1 or COMPUTATIONAL2 data items, by PL/I using FLOATING DECIMAL data items, by FORTRAN using REAL or REAL*8 data items, or by Assembler using floating-point data items. The general syntax of the RB format is RBw, where w is the total number of bytes. The width specification must be an even number between 2 and 8. Normally, a width specification of 8 is used to read double-precision values, and a width of 4 is used to read single-precision values. RBHEXw (hexadecimal of RB): The RBHEX format interprets a series of hexadecimal characters as a number that represents a floating-point number. This representation is system-specific. If the field width is less than twice the width of a floating-point number, the value is right-padded with binary zeros. For some operating systems (for example, IBM CMS), letters in hexadecimal values must be upper case. The general syntax of the RBHEX format is RBHEXw, where w indicates the total number of columns. The width must be an even number. The values are real (floating-point) numbers. Leading and trailing blanks are not allowed. Any data values shorter than the specified input width must be padded with leading zeros.
Date and Time Formats
Date and time formats are both input and output formats. Like numeric formats, each input format generates a default output format, automatically expanded (if necessary) to accommodate display width. Internally, all date and time format values are stored as a number of seconds: date formats (e.g., DATE, ADATE, SDATE, DATETIME) are stored as the number of seconds since October 14,
45 Universals
1582; time formats (TIME, DTIME) are stored as a number of seconds that represents a time interval (e.g., 10:00:00 is stored internally as 36000, which is 60 seconds x 60 minutes x 10 hours). All date and time formats have a minimum input width, and some have a different minimum output. Wherever the input minimum width is less than the output minimum, SPSS expands the width automatically when displaying or printing values. However, when you specify output formats, you must allow enough space for displaying the date and time in the format you choose. Input data shorter than the specified width are correctly evaluated as long as all the necessary elements are present. For example, with the TIME format, 1:2, 01 2, and 01:02 are all correctly evaluated even though the minimum width is 5. However, if only one element (hours or minutes) is present, you must use a time function to aggregate or convert the data. For more information, see Date and Time Functions on p. 68. If a date or time value cannot be completely displayed in the specified width, values are truncated in the output. For example, an input time value of 1:20:59 (1 hour, 20 minutes, 59 seconds) displayed with a width of 5 will generate an output value of 01:20, not 01:21. The truncation of output does not affect the numeric value stored in the working file. The following table shows all available date and time formats, where w indicates the total number of columns and d (if present) indicates the number of decimal places for fractional seconds. The example shows the output format with the minimum width and default decimal positions (if applicable). The format allowed in the input data is much less restrictive. For more information, see Input Data Specification on p. 46.
Table 2-2 Date and time formats
Format type
Description
Min w In Out 9 11 8 10 8 10 5 7 8 10 6 8
Max w Max d General form
Example
DATEw
International date American date European date Julian date
9 10 8 10 8 10 5 7
40
dd-mmm-yy dd-mmm-yyyy
28-OCT-90 28-OCT-1990 10/28/90 10/28/1990 28.10.90 28.10.1990 90301 1990301 90/10/28 1990/10/28 4 Q 90 4 Q 1990
ADATEw
40
mm/dd/yy mm/dd/yyyy
EDATEw
40
dd.mm.yy dd.mm.yyyy
JDATEw
40
yyddd yyyyddd
SDATEw
Sortable date* Quarter and year
8 10 4 6
40
yy/mm/dd yyyy/mm/dd
QYRw
40
q Q yy q Q yyyy
46 Universals
Format type
Description
Min w In Out 6 8 8 10 2 3 5 10 1 13 17 22
Max w Max d General form mmm yy mmm yyyy 40 ww WK yy ww WK yyyy 40 40 40 40 40 40 40 40 16 16 16 (name of the day) (name of the month) hh:mm hh:mm:ss.s dd hh:mm dd hh:mm:ss.s dd-mmm-yyyy hh:mm dd-mmm-yyyy hh:mm:ss.s
Example
MOYRw
Month and year Week and year Day of the week Month Time
6 8 6 8 2 3 5 10
40
OCT 90 OCT 1990 43 WK 90 43 WK 1990 SU JAN 01:02 01:02:34.75 20 08:03 20 08:03:00 20-JUN-1990 08:03 20-JUN-1990 08:03:00
WKYRw
WKDAYw MONTHw TIMEw TIMEw.d DTIMEw DTIMEw.d DATETIMEw DATETIMEw.d
Days and time Date and time
1 13 17 22
* All date and time formats produce sortable data. SDATE, a date format used in a number of
Asian countries, can be sorted in its character form and is used as a sortable format by many programmers.
Input Data Specification
The following general rules apply to date and time input formats: The century value for two-digit years is defined by the SET EPOCH value. By default, the century range begins 69 years prior to the current year and ends 30 years after the current year. Whether all four digits or only two digits are displayed in output depends on the width specification on the format. Dashes, periods, commas, slashes, or blanks can be used as delimiters in the input values. For example, with the DATE format, the following input forms are all acceptable:
28-OCT-90 28/10/1990 28.OCT.90 28 October, 1990
The displayed values, however, will be the same: 28-OCT-90 or 28-OCT-1990, depending on whether the specified width allows 11 characters in output. The JDATE format does not allow internal delimiters and requires leading zeros for day values of less than 100 and two-digit-year values of less than 10. For example, for January 1, 1990, the following two specifications are acceptable:
90001 1990001
47 Universals
However, neither of the following is acceptable:
90 1 90/1
Months can be represented in digits...