Fit Linear Models Defined By Model Syntax

Fit a list of linear models defined by model syntax.

Usage

many_lm(models, data, na_omit_all = TRUE, ...)

# S3 method for class 'lm_list_lmhelprs'
getCall(x, ...)

# S3 method for class 'lm_list_lmhelprs'
update(object, ..., evaluate = TRUE)

# S3 method for class 'lm_list_lmhelprs'
coef(object, y = NULL, ...)

# S3 method for class 'lm_list_lmhelprs'
vcov(object, y = NULL, vcov_args = list(), ...)

# S3 method for class 'lm_list_lmhelprs'
confint(
  object,
  parm = NULL,
  level = 0.95,
  ci_fun = stats::confint,
  ci_args = list(),
  ...
)

# S3 method for class 'lm_list_lmhelprs'
nobs(object, ...)

# S3 method for class 'lm_list_lmhelprs'
model.frame(formula, ...)

# S3 method for class 'lm_list_lmhelprs'
variable.names(object, ...)

Arguments

models: Character. Model syntax. See Details.
data: The data frame. Must be supplied if na_omit_all is TRUE. If na_omit_all is FALSE, it can be omitted (though not suggested).
na_omit_all: How missing data is handled across models. If TRUE, the default, then only cases with no missing data on all variables used at least one of the models will be retained (i.e., listwise deletion). If FALSE, then missing data will be handled in each model separately by lm().
...: Additional arguments. For many_lm(), these arguments will be passed to lm(). For the update method of the output of many_lm(), these arguments will used to update the call to many_lm(). For other methods, these arguments will be passed to methods if applicable.
x: For the getCall method, this is the output of many_lm().
object: For methods of the output of many_lm(), it is the output of many_lm().
evaluate: If TRUE, the updated call will be evaluated. If FALSE, the updated call will be returned.
y: For some methods, if set to a character vector, only the coefficients of the models of the response variables listed in y will be returned. If NULL, all coefficients will be returned.
vcov_args: A named list of arguments to be passed to stats::vcov() when computing the variance-covariance matrices of the regression coefficients. Default is list(), an empty list.
parm: The parameters for which the confidence intervals will be returned. If NULL, the confidence intervals for all coefficients will be returned.
level: The level of confidence of the confidence levels.
ci_fun: The function to be used to form the confidence intervals for regression coefficients. Default is stats::confint
ci_args: A named list of arguments to be passed to ci_fun. Default is list(level = .95), requesting 95% confidence intervals.
formula: For the model.frame method, this should be the output of many_lm().

Value

A list of the output of lm(). The class is lm_list_lmhelprs.

The getCall method returns a call stored in the output of many_lm().

For the update method of the output of many_lm(), if evaluate is TRUE, the updated output of many_lm() will be returned. If evaluate is FALSE, the updated call will be returned.

The coef method returns a numeric vector of the coefficients.

The vcov method returns the variance-covariance matrix of the parameter estimates of the models.

The confint method returns a two-column matrix of the confidence intervals.

The nobs method returns the number of cases used in the models.

The model.frame method of the output of many_lm() returns a model frame, formed by merging the model frames of all models.

The variable.names method returns a character vector of unique variable names in all the stored models.

Details

This function extracts linear model formulas from a model syntax (a character vector), fits each of them by lm(), and stores the results in a list.

Lines with the first non-whitespace character "#" are treated as comments and ignored.

Each line must be a valid formula for lm().

Listwise deletion

If na_omit_all is TRUE, the default, then cases with missing data on at least one of the variables used in the model will be removed. Each call to lm() will have subset set to an integer vector of cases not removed (i.e., cases retained)

Handling the `subset` argument

If subset is used when calling this function, it will also be used to select cases.

Note that the subset argument in the call in each model will be replaced by a numeric vector of cases retained, determined by both missing data and the original value of the subset.

The getCall method of the output of many_lm() extracts the stored call.

The output of many_lm() has a update method.

The coef method extracts the regression coefficients in the lavaan style: y ~ x, y the response variable and x a term.

The vcov method is used to extract the variance-covariance matrices of the models.

The confint method computes the confidence intervals for the coefficients.

The nobs method extracts the number of cases actually used in the analysis. Should be identical for all models.

Unlike the corresponding method for the output of stats::lm(), the model.frame method of the output of many_lm() does only one thing: extracts the stored model frame. It is because many_lm() is for fitting several models, with several formulas instead of only one.

The variable.names method of the output of many_lm() extract the variable names of all models.

Author

Shu Fai Cheung https://orcid.org/0000-0002-9871-9448

Examples


data(data_test1)
mod <- "x3 ~ x2 + x1
        x4 ~ x3
        x5 ~ x4*x1"
out <- many_lm(mod, data_test1)
summary(out)
#> Call:
#> many_lm(models = mod, data = data_test1)
#> 
#> Model:
#> x3 ~ x2 + x1
#> <environment: 0x55b6ed1a1280>
#>             Estimate Std. Error t value Pr(>|t|)    
#> (Intercept)  -0.0835     0.0968   -0.86  0.39067    
#> x2           -0.0494     0.0896   -0.55  0.58289    
#> x1            0.3801     0.0955    3.98  0.00013 ***
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> R-square = 0.143. Adjusted R-square = 0.125. F(2, 97) = 8.079, p < .001
#> 
#> Model:
#> x4 ~ x3
#> <environment: 0x55b6ed1a1280>
#>             Estimate Std. Error t value Pr(>|t|)  
#> (Intercept)  -0.1144     0.0866   -1.32    0.190  
#> x3            0.2156     0.0846    2.55    0.012 *
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> R-square = 0.062. Adjusted R-square = 0.053. F(1, 98) = 6.489, p = 0.012
#> 
#> Model:
#> x5 ~ x4 * x1
#> <environment: 0x55b6ed1a1280>
#>             Estimate Std. Error t value Pr(>|t|)   
#> (Intercept) -0.10644    0.10399   -1.02   0.3086   
#> x4           0.15124    0.11589    1.31   0.1950   
#> x1           0.27872    0.09885    2.82   0.0058 **
#> x4:x1        0.00945    0.12017    0.08   0.9375   
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> R-square = 0.108. Adjusted R-square = 0.080. F(3, 96) = 3.878, p = 0.012