desmat model [, colinf defcon(string) ]
desmat : any_stata_command [using] [if,in,weights_for_command] model [, verbose defcon(string) desrep(string) [command_options] ]
desmat is used to generate a design matrix, i.e. a set of dummy variables based on categorical and/or continuous variables. These dummy variables _x_* can then be used in any appropriate Stata procedure. desmat therefore serves the same purpose as xi, but allows different types of parameterizations than the indicator contrast (i.e. dummy variables with a fixed reference category). In addition, desmat allows the specification of higher order interaction effects and an easier specification of the reference category. After estimating a model, desrep can be used to produce a compact overview of the estimates with informative labels. In addition, the program destest can be used to perform a Wald test on model terms.
Like xi, desmat can be used as either as a command or as a command prefix. When used as a command, desmat generates a set of dummy variables for use by subsequent Stata programs. When used as a command prefix, the model is estimated after the dummy variables are generated and the results are presented using desrep.
A model consists of one or more terms separated by spaces. A term can be a single variable, two or more variables joined by period(s), or two or more variables joined by asterisk(s). A period is used to specify an interaction effect as such, whereas an asterisk indicates hierarchical notation, in which both the interaction effect itself plus all possible nested interactions and main effects are included. For example, the term vote*educ*race is expanded to vote educ vote.educ race vote.race educ.race vote.educ.race.
All variables in the model will be treated as categorical unless specified as continuous using the pzat characteristic (discussed below) or by specifying a contrast for the term (also discussed below). Alternatively, a variable can be prefixed by an @ to flag it as a continuous variable. For example:
desmat: regress brate @medage @medagesq region
The variables medage and medagesq will be treated as continuous variables. The variable region will be treated as categorical and dummy variables will be generated using its first category as reference category.
When desmat is used as a command prefix, weights, if, or in options may be specified in the usual manner and will be passed on to the procedure in question. Any options besides verbose, defcon and desrep will be passed on to the procedure as well.
If using filename is specified then the results will be written to a tab-delimited ascii file. The default extension for filename is .out (cf. outshee2). See desrep for further details
defcon specifies a default contrast to be used in the model. See the section on contrasts below for details. By default, desmat generates dummy variables using the first category as reference category.
For compatibility with earlier versions of desmat, a default parameterization may be specified as an option rather than an argument for the defcon option. This option is only available when desmat is used as a command by itself.
colinf lets desmat report which variables are dropped because of collinearity. desmat will generate duplicate dummy variables if the same variable is specified twice in a model, e.g. in interaction terms. desmat subsequently uses the Stata facilities for removing collinear variables to delete these duplicates. The information on which variables are dropped will therefore usually be uninteresting. If variables are being dropped because they are actually collinear rather than duplicates, the colinf can be used to find out where the problems are.
verbose prints information on the design matrix generated and the regular output of the Stata command being executed when desmat is used as a command prefix.
desrep passes option on to desrep, which displays after the model has been estimated. Note that most of these options can be specified using global macro variables; see desrep for details. An exception could be the exp option. desrep displays linear coefficients even if the procedure prints exponential coefficients, e.g. the odds-ratios produced by logistic. Specify:
desmat: logistic vote memb educ*race [fw=pop], desrep(exp all)
to display odds-ratios. See desrep for further details.
By default, desmat generates dummy variables using the first category as the reference category, as does xi. However, it can also use different types of restrictions (contrasts) and different reference categories when generating the dummy variables. A restriction of some type is required for the effects of categorical variables to be identifiable. The restriction used does not affect the fit of the model but does determine the meaning of the parameters. A common restriction and the one used by xi is to drop the dummy variable for a reference category. The parameters for that variable are then relative to the reference category. Another common constraint is the deviation contrast, in which parameters have a sum of zero. One parameter can therefore be dropped as redundant during estimation and found afterwards using minus the sum of the estimated parameters, or by re-estimating the model using a different omitted category. Bock (1975) and Finn (1974) discuss other types of parameterizations (or contrasts) and the technical details in implementing them.
A contrast can be specified as a name, of which the first three characters are significant, optionally followed by a specification of the reference category in parentheses (no spaces). The reference category should refer to the category number, not the category value. So for a variable with values 0 to 3, the specification dev(1) indicates that the deviation contrast is to be used with the first category (i.e. 0) as the reference. If no reference category is specified or the category specified is less than 1 then the first category is used as reference category. If the reference category specified is larger than the number of categories then the highest category is used. Note that for certain types of contrasts, the reference specifiation has a different meaning.
The available contrasts are:
ind(ref) specifies the indicator contrast, i.e. dummy variables with ref as reference category. This is the contrast used by xi and the default contrast for desmat.
full specifies a full contrast, i.e. dummy variables are included for all categories and no restrictions are imposed. Because of this, desmat also does not check for collinearity due to duplicat dummy variables in e.g. interaction terms.
dir specifies a direct effect. This is used to include continuous variables in the model.
dev(ref) specifies the deviation contrast. Parameters sum to zero over the categories of the variable. The parameter for ref is omitted as redundant, but can be found from minus the sum of the estimated parameters.
sim(ref) specifies the simple contrast with ref as the reference category. The highest order effects are the same as indicator contrast effects, but lower order effects and the constant will be different.
dif(ref) specifies the difference contrast, for variables with ordered categories. Parameters are relative to the following category. If the first letter of ref is b then the backward difference contrast is used instead, and parameters are relative to the previous category.
hel(ref) specifies the Helmert contrast, which is again used for variables with ordered categories. Parameters represents the contrast between that category and the mean of the subsequent categories. If the first letter of ref is b then the reverse Helmert contrast is used and parameters are relative to the mean of the preceding categoriees.
orp(ref) specifies orthogonal polynomials of degree ref. The first parameter is a linear effect, the second quadratic, etc. This option calls orthpoly to generate the design (sub)matrix.
use(ref) specifies a user-defined contrast. ref refers to an R by C contrast matrix, where C is the number of categories and R < C. If rownames are specified for this matrix, these names will be used as variable labels for the resulting dummy variables. [Single lowercase letters as names for the contrast matrix cause problems at the moment, e.g use(c). Use uppercase names or more than one letter, e.g. use(cc) or use(C)]
Specifying contrasts using the defcon option
The defcon option can be used to specify a different contrast than ind(1) for all variables in all terms, e.g.
desmat: logistic vote memb educ*race [fw=pop], desrep(exp all) defcon(dev(99))
The deviation contrast will now be used with the highest category as the redundant category.
The global variable $D_CON can be used to specify a default contrast for the current Stata session. For example:
global D_CON "dev(99)"
will cause desmat to use the deviation contrast for the duration of the Stata session. By specifing this command in their profile.do, users can specify a different contrast for all desmat models. The $D_CON global variable is overridden by the defcon option if this is specified.
Specifying contrasts using the pzat characteristic
A pzat characteristic can be assigned to a variable to specifify a contrast to be used for that variable. For example, to use the backward difference contrast for education but the default indicator contrast for the other variables, use:
char educ[pzat] dif(b) desmat logistic vote memb educ*race [fw=pop], desrep(exp all)
The pzat characteristic will override the contrast specified by the defcon option. So in
char educ[pzat] dif(b) desmat: logistic vote memb educ*race [fw=pop], desrep(exp all) defcon(dev(99))
The difference contrast will be used for all variables except educ.
Specifying contrasts in the model specification
It is also possible to specify contrasts in the model specification, on a variable by variable basis if so desired. This is done by appending =con(ref) to a single variable, =con(ref).con(ref) to an interaction effect, and =con(ref)*con(ref) to an interaction using hierarchical notation. A somewhat contrived example:
desmat race=ind(1) educ=hel memb vote vote.memb=dif.dev(1), defcon(ind(99))
The variable race will use the indicator contrast with the first category as reference. The variable educ will use the helmert contrast, vote will use the difference contrast in its interaction with memb, whereas memb will use the deviation contrast in its interaction with vote. The main effects of memb and vote will use the default contrast, which is specified here as the indicator contrast with the highest contrast as reference. Interpreting this mishmash of parameterizations would be quite a chore of course.
A variable's pzat characteristic overrides the defcon option, but is itself overridden by a specification in the model. For example:
char educ[pzat] dif(b) desmat vote*memb vote*educ*race=dev(99)*orp(1)*dev(99) educ*race*memb, defcon(d > ev(99))
educ will use a first degree polynomial restriction in the vote*educ*race term and a backward difference contrast elsewhere. All other variables will use the deviation contrast.
Specifying contrasts in the model statement will tend to look messy and provides an overkill in flexibility. Use of the pzat characteristic in conjunction with the defcon option and the @ prefix to flag continuous variables will usually be preferable.
When used as a command, or in command prefix mode in conjunction with the verbose option, desmat produces a legend of dummy variables it has produced, the model term these pertain to, and the contrast used. The showtrms command can be used afterwards to generate this legend for the last model generated by desmat. This can be useful when desmat is used as a command prefix to check on the types of contrasts being used.
When used as a command rather than a command prefix, the dummy variables generated by desmat can be included in any Stata procedure as _x_*. After estimating the model,the companion program desrep can be used to present the results with descriptive labels.
desmat creates global macro variables $term1, $term2, etc. for each terms in the model. The program destest can be used to perform a Wald test on model terms.
Note that either in command mode or command prefix mode, desmat produces a set of dummy variables _x_*. These variables must be present for destest and showtrms to work. The commands:
drop _x_* macro drop term*
can be used to cleanup after desmat if so desired.
Hendrickx, J. 1999. dm73: Using categorical variables in Stata. Stata Technical Bulletin 52: 2-8. Reprinted in Stata Technical Bulletin Reprints, vol. 9. pp. 51-59.
--. 2000. dm73.1: Contrasts for categorical variables: update. Stata Technical Bulletin 54: 7. Reprinted in Stata Technical Bulletin Reprints, vol. 9. pp. 60-61.
--. 2001. dm73.2: Contrasts for categorical variables: update. Stata Technical Bulletin 59: 2-5.
Direct comments to: John Hendrickx
desmat is available at SSC-IDEAS. Use findit desmat to locate the latest version.
Aso see Manual: [R] xi [U] Commands for dealing with categorical variables On-line: help for desrep, destest, showtrms, xi