Package 'pwlmm'

Title: PWIGLS for Two-Level Multivariate and Multilevel Linear Models
Description: Estimates two-level multilevel linear model and two-level multivariate linear multilevel model with weights following Probability Weighted Iterative Generalised Least Squares approach. For details see Veiga et al.(2014) <doi:10.1111/rssc.12020>.
Authors: Alinne Veiga [aut, cre] , João Messa [aut, trl, com]
Maintainer: Alinne Veiga <[email protected]>
License: GPL (>= 2)
Version: 1.1.1
Built: 2024-11-27 06:49:04 UTC
Source: CRAN

Help Index


datamultv data

Description

Longitudinal data set with the five occasions for each reference person. Generated from the Continuous PNAD (IBGE/Brazil) from the first quarter of 2018 until the first quarter of 2019 following households included in the sample at the first quarter of 2018.

Usage

data(datamultv)

Format

A data.frame with 1685 observations and the following 13 variables.

Y

the logarithm of the monthly income in the main job for people aged 14 and over (only for people who received cash, products or goods in the main job) plus 1

X1

age of the resident in the reference date centered around 40

X2

indicator variable for male residents

X3

indicator variable for residents of white color or race

X4

the logarithm of hours actually worked in the reference week in all jobs for people aged 14 and over plus 1

X5

years of study (people aged 5 or over) standardized for elementary school 9 YEARS SYSTEM

PSU

level 2 identifiers, is the group identifier for this data

STRAT

variable that identifies the strata (not needed in the command functions)

wave

time-dummies for level 1 units

idd

level 1 identifiers

wj

weights corresponding to level 2 units

w_ij

vector of weights corresponding to level 1 units, conditional to their respective level 2 unit (also longitudinal weights in the multivariate data)

wi_j

weights corresponding to level 1 and 2 units (not needed in the command functions)

Examples

data(datamultv)
summary(datamultv)

dataw1 data

Description

Crosssectional data set. A two-level data containing wave one from datamultv data. Generated from the first quarter of 2018 data of the Continuous PNAD (IBGE/Brazil).

Usage

data(dataw1)

Format

A data.frame with 337 observations and the following 13 variables.

Y

the logarithm of the monthly income in the main job for people aged 14 and over (only for people who received cash, products or goods in the main job) plus 1

X1

age of the resident in the reference date centered around 40

X2

indicator variable for male residents

X3

indicator variable for residents of white color or race

X4

the logarithm of hours actually worked in the reference week in all jobs for people aged 14 and over plus 1

X5

years of study (people aged 5 or over) standardized for elementary school 9 YEARS SYSTEM

PSU

level 2 identifiers, is the group identifier for this data

STRAT

variable that identifies the strata (not needed in the command functions)

wave

time-dummies for level 1 units

idd

level 1 identifiers

wj

weights corresponding to level 2 units

w_ij

vector of weights corresponding to level 1 units, conditional to their respective level 2 unit (also longitudinal weights in the multivariate data)

wi_j

weights corresponding to level 1 and 2 units (not needed in the command functions)

Examples

data(dataw1)
summary(dataw1)

Fit Weighted Linear Multilevel Model

Description

Fit a probability-weighted two-level linear model with unequal selection probabilities at each level, via IGLS algorithm.

Usage

pwigls2(formula, data = NULL, wj, wij)

Arguments

formula

a two-sided linear formula object describing both the fixed-effects and random-effects part of the model, with the response on the left of a ~ operator and the terms, separated by + operators, on the right. Random-effects terms are distinguished by vertical bars (|) separating expressions for design matrices from grouping factors.

data

an optional data frame containing the variables in formula. If not found in data, the variables are taken from the environment of formula (if specified as a formula) or from the parent frame (if specified as a character vector).

wj

a vector of sampling weights for level two units. Level two units are selected with inclusion probabilities. Then, sampling weights for the level two units are defined as the inverse of these probabilities.

wij

a vector of sampling weights for level one units. After selecting a level two unit, level one units belonging to them are selected with inclusion probabilities. Then, sampling weights for the level one units are defined as the inverse of these probabilities.

Details

Follows estimation process described in Pfeffermann et al. (1998). Uses probability-weighted IGLS with scaled weights.

Value

Estimated list of estimators

References

D. Pfeffermann; C. J. Skinner; D. J. Holmes; H. Goldstein; J. Rasbash, 2008, Weighting for Unequal Selection Probabilities in Multilevel Models Journal of the Royal Statistical Society. Series B (Statistical Methodology), Vol. 60, No. 1. (1998), pp. 23-40.

Examples

data(dataw1)
pwigls2( Y ~ X1 + X2 + (1 | PSU), data = dataw1, wj, wi_j)

Fit Weighted Multivariate Linear Multilevel Model to Longitudinal Data.

Description

Fit a two-level probability-weighted multivariate linear model with a linear error covariance matrix structure, via IGLS algorithm.

Usage

wmlmm(formula, data = NULL, ID3, ID2, ID1, wj, wij, type, rot = NULL)

Arguments

formula

a linear formula object with the response on the left of a ~ operator and the terms, separated by + operators, on the right.

data

an optional data frame containing the variables in formula. If not found in data, the variables are taken from the environment of formula (if specified as a formula) or from the parent frame (if specified as a character vector).

ID3

vector of indexes for level two units

ID2

vector of indexes for level one units.

ID1

vector of successive measurements within the same level one unit, for all units.

wj

a vector of sampling weights for level two units. Level two units are selected with inclusion probabilities. Then, sampling weights for the level two units are defined as the inverse of these probabilities.

wij

a vector of sampling weights for level one units. After selecting a level two unit, level one units belonging to them are selected with inclusion probabilities. Then, sampling weights for the level one units are defined as the inverse of these probabilities.

type

type of structure imposed in the error covariance matrix; "toep" refers to the toeplitz, "uns" refers to the unestructured and "genlin" refers to the general linear.

rot

vector of 0's and 1's related to the measurements in time when "genlin" is passed to the type argument. Use 1 if the data were collected in that specific time unit, and 0 otherwise.

Details

Follows estimation process described in Veiga et al. (2014). Uses probability-weighted IGLS with scaled weights.

Value

Estimated list of estimators

References

A. Veiga, P. W. F. Smith and J. J. Brown (2014), The use of sample weights in multivariate multilevel models with an application to income data collected by using a rotating panel survey Journal of the Royal Statistical Society. Series C (Applied Statistics) Vol. 63, No. 1 (JANUARY 2014), pp. 65-84 (20 pages)

Examples

data(datamultv)
 wmlmm ( Y ~ X1 + X2, data = datamultv, PSU, idd, wave, wj, wi_j, "toep")