Note:If you specified 0 forthe number of cases to scan in the initial dialog box, the ValueLabel gridwill
initially be blank, except forany preexistingvalue labels and/or defined missing values categories forthe
selectedvariable. In addition, the Suggest button for the measurement level will be disabled.
Measurement Level. Valuelabels are primarilyuseful for categorical (nominal andordinal) variables, and
some procedures treat categorical andscale variables differently;so itis sometimes importanttoassign
the correct measurement level.However, by default, all new numeric variables are assigned the scale
measurement level. Thus,many variables thatare in fact categorical may initially be displayedas scale.
If you are unsure of whatmeasurement level toassign to a variable, clickSuggest.
Role. Somedialogs support the ability topre-select variables for analysis based on definedroles.See the
topic “Roles” on page 54 for more information.
Copy Properties. You can copy value labels and other variable properties from anothervariable tothe
currently selected variable or from the currently selectedvariable to one or more other variables.
Unlabeled Values. To create labels forunlabeledvalues automatically, click Automatic Labels.
Variable Label andDisplay Format
You can change the descriptive variable label andthe display format.
v You cannot change the variable's fundamental type (string or numeric).
v For string variables, you can change only the variable label, notthedisplayformat.
v For numeric variables, you can change the numeric type (such as numeric, date,dollar,orcustom
currency), width (maximum number of digits, including anydecimal and/or grouping indicators), and
number of decimal positions.
v For numeric date format, you can select a specific dateformat (such as dd-mm-yyyy, mm/dd/yy, and
v For numeric custom format, you can selectone offive custom currencyformats (CCA through CCE).
Seethe topic “Currency options” on page 203 for moreinformation.
v An asterisk is displayed in the Value column if thespecified width is less than the width of the
scanned values orthe displayed values for preexistingdefined value labels or missing values
v Aperiod (.) is displayed if the scanned values orthe displayed values for preexistingdefined value
labels ormissing values categories are invalidfor theselected display format type. Forexample, an
internal numeric value of less than 86,400 is invalid for a date format variable.
Assigning the Measurement Level
When you click Suggest for the measurement level in theDefine Variable Properties main dialogbox, the
current variable is evaluatedbased on the scanned cases anddefined value labels, anda measurement
level is suggested in the SuggestMeasurement Level dialog box thatopens. The Explanation area
provides a brief description of the criteria usedtoprovide the suggestedmeasurement level.
Note:Values definedas representing missing values are notincluded in the evaluation for measurement
level. For example, theexplanation for thesuggestedmeasurement level mayindicate that the suggestion
is in partbased on thefact that thevariablecontains nonegative values, whereas it may in fact contain
negativevalues--but thosevalues arealready definedas missingvalues.
1. Click Continue to accept the suggestedlevel of measurementorCancel to leavethe measurement
Custom Variable Attributes
The Attributes button in Define Variable Properties opens the Custom VariableAttributes dialog box. In
addition tothe standard variable attributes, such as value labels, missing values, and measurement level,
you can create your own custom variable attributes. Like standard variable attributes, these custom
attributes are saved with IBM SPSS Statistics data files.
Name. Attributenames must follow thesame rules as variablenames.Seethetopic “Variable names” on
page 50 for more information.
Value.The value assigned to the attribute for the selectedvariable.
v Attribute names that begin with a dollar sign are reserved and cannot be modified.You can view the
contents of a reservedattribute byclicking the button in the desiredcell.
v The text Array..., displayed in a Valuecell, indicates that this is an attribute array, an attribute that
contains multiplevalues. Click the button in thecell to display the list of values.
Copying Variable Properties
TheApply Labels andLevel dialog box is displayedwhen you click From Another Variable orTo Other
Variables in the Define VariableProperties main dialog box.It displays all of the scanned variables that
match the current variable's type (numeric orstring). For string variables, the defined width must also
1. Select a single variable from which tocopy value labels andother variable properties (except variable
2. Select one or more variables towhich tocopy value labels andother variable properties.
3. Click Copy to copy the value labels and the measurement level.
v Existing valuelabels and missing value categories for target variable(s) are not replaced.
v Value labels andmissingvaluecategories forvalues notalready definedfor the targetvariable(s) are
added to the set of value labels and missing value categories for the targetvariable(s).
v The measurementlevel forthe target variable is always replaced.
v The role forthe target variable is always replaced.
v If either the sourceortarget variable has a defined range of missing values, missingvalues definitions
Setting measurement level for variables with unknown measurement
For some procedures, measurement level can affect the results ordetermine which features areavailable,
and you cannot access thedialogs for these procedures until all variables have a definedmeasurement
level. The SetMeasurement Level for Unknown dialog allows you to definemeasurement level for any
variables with an unknown measurement level without performing a data pass (which may be
time-consumingfor largedata files).
Under certain conditions, the measurement level forsome or all numeric variables (fields) in a file may
be unknown. Theseconditions include:
v Numeric variables from Excel 95 or later files, text data files, or data base sources prior to the first data
v New numeric variables createdwith transformation commands prior to the first data pass after
creation of those variables.
These conditions apply primarilytoreading data orcreating new variables via commandsyntax. Dialogs
for reading data and creating new transformed variables automatically perform a data pass that sets the
measurement level, based on the default measurement level rules.
Toset the measurement level forvariables with an unknown measurement level
1. In thealertdialog thatappears for the procedure, click AssignManually.
2. From the menus, choose:
Data > Set Measurement Level for Unknown
3. Move variables (fields) from the source list to the appropriate measurement level destination list.
v Nominal. Avariable can be treatedas nominal when its values represent categories with nointrinsic
ranking (forexample, the departmentofthecompany in which an employee works). Examples of
nominal variables includeregion, postal code, andreligious affiliation.
v Ordinal.Avariablecan be treated as ordinal when its values represent categories with someintrinsic
ranking (forexample, levels of service satisfaction from highly dissatisfied to highly satisfied).
Examples of ordinal variables include attitudescores representing degree of satisfaction or confidence
andpreference rating scores.
v Continuous.Avariable can be treated as scale (continuous) when its values represent ordered categories
with a meaningful metric,so that distance comparisons between values are appropriate. Examples of
scale variables include age in years and income in thousands of dollars.
Multiple Response Sets
Custom Tables and the Chart Builder support a special kind of "variable" calleda multiple response set.
Multiple response sets aren'treally "variables" in thenormal sense. You can't see them in the Data Editor,
and other procedures don't recognize them. Multipleresponse sets use multiple variables to record
responses to questions where the respondent can give more than one answer. Multipleresponse sets are
treatedlikecategorical variables, andmost of the things you can dowith categorical variables, you can
also dowith multipleresponse sets.
Multiple response sets are constructed from multiple variables in the data file. Amultiple response set is
aspecial constructwithin a data file. You can define andsave multipleresponse sets in IBM SPSS
Statistics data files, but you cannotimport or export multiple response sets from/to other file formats.
You can copy multipleresponse sets from other IBM SPSS Statistics data files using Copy Data Properties,
which is accessed from theData menu in the Data Editor window.
Defining Multiple Response Sets
Todefine multiple response sets:
1. From the menus, choose:
Data > Define Multiple Response Sets...
2. Select twoormore variables. If your variables are coded as dichotomies, indicate which value you
want to have counted.
3. Enter a unique name foreach multiple response set. Thename can be up to 63 bytes long.A dollar
sign is automatically added to the beginning of the setname.
4. Enter a descriptive label for the set. (This is optional.)
5. Click Add to addthe multiple response set to the list of defined sets.
Amultipledichotomyset typically consists of multiple dichotomous variables: variables with onlytwo
possible values of a yes/no, present/absent, checked/notchecked nature.Although the variables may not
be strictly dichotomous, all of the variables in theset are coded the same way, and the Counted Value
represents the positive/present/checkedcondition.
For example, a survey asks the question, "Which of the following sources do you rely on for news?" and
provides five possible responses. The respondentcan indicatemultiple choices by checking a box next to
each choice. Thefive responses become five variables in the data file,coded 0 for No (not checked) and 1
forYes(checked). In themultiple dichotomy set, the Counted Valueis 1.
The sample data filesurvey_sample.sav already has three definedmultiple responsesets. $mltnewsis a
1. Select (click)$mltnewsin the Mult. Response Sets list.
This displays the variables and settings usedtodefine this multiple response set.
v The Variables in Set list displays thefive variables usedto construct the multipleresponse set.
v The Variable Coding group indicates that the variables are dichotomous.
v The Counted Valueis 1.
2. Select (click)one of the variables in the Variables in Setlist.
3. Right-click the variable and select Variable Informationfrom the pop-up pop-up menu.
4. In theVariable Information window, click the arrow on the Value Labels drop-down listtodisplaythe
entire list of defined value labels.
The valuelabels indicate thatthe variable is a dichotomywith values of0 and 1, representing No and Yes,
respectively.All fivevariables in the list are coded the same way, and the value of 1 (the codefor Yes) is
the counted value for the multiple dichotomy set.
Amultiplecategoryset consists of multiplevariables, all codedthe same way,often with manypossible
response categories. Forexample, a survey item states,"Name up to three nationalities thatbest describe
yourethnic heritage." Theremay be hundreds of possible responses, but for coding purposes the list is
limitedtothe40 most common nationalities, with everything else relegatedto an "other" category. In the
data file,the threechoices become threevariables,each with 41 categories (40 codednationalities andone
In thesample data file, $ethmult and $mltcars aremultiple category sets.
For multiple dichotomies, you can control how sets are labeled.
v Variable labels. Uses thedefined variable labels (orvariablenames forvariables without defined
variable labels) as theset category labels. For example, ifall ofthe variables in the set havethe same
value label (or nodefined value labels)for the counted value (for example, Yes), then you should use
the variable labels as the set category labels.
v Labels of countedvalues. Uses thedefined value labels of the countedvalues as set category labels.
Selectthis option only if all variables have a defined valuelabel forthe countedvalueandthe value
label for the countedvalue is different foreach variable.
v Use variable label as set label. Ifyou select Label of countedvalues, you can also use the variable
label for the first variable in the set with a defined variable label as the set label. If noneof the
variables in the set have defined variable labels, the name of the first variable in the set is usedas the
Documents you may be interested
Documents you may be interested