Journal of the Royal Statistical Society: Series A (Statistics in Society)

Cover image for Vol. 178 Issue 2

Edited By: H. Goldstein and L. Sharples

Impact Factor: 1.573

ISI Journal Citation Reports © Ranking: 2013: 11/45 (Social Sciences Mathematical Methods); 23/119 (Statistics & Probability)

Online ISSN: 1467-985X

Associated Title(s): Journal of the Royal Statistical Society: Series B (Statistical Methodology), Journal of the Royal Statistical Society: Series C (Applied Statistics), Significance

177:2


Missing ordinal covariate with informative selection, by A. Miranda and S. Rabe-Hesketh, Journal of the Royal Statistical Society, Series A, Statistics in Society, Volume 177, part 2 (2014), pages 419 - 438

The Stata 'do' file shows how to install "lcmc" from the SSC archive and how to run the model on the provided simulated data.

The code is already public and deposited in the Statistical Software Components (SSC) archive of Boston College and can be downloaded directly from within Stata. We have also made available a data set that contains simulated data that can be used to learn how to run "lcmc", this data set is also already public and can be downloaded from the SSC.

The conditions of use of the "lcmc" command is that the code and the main paper should be cited together at all times.

Alfonso Miranda
Economics Division Center for Research and Teaching in Economics
Carretera México-Toluca 3655 Col.
Lomas de Santa Fe 01210 México DF
México

E-mail: alfonso.miranda@cide.edu

Dataset

On modelling early life weight trajectories, by C. Pizzi, T. J. Cole, C. Corvalan, I. dos Santos Silva, L. Richirdi and B. L. De Stavola, Journal of the Royal Statistical Society, Series A, Statistics in Society, Volume 177, part 2 (2014), pages 371 - 396

RELEVANT PROGRAMS:

Three programs, produced with R version 2.11.1, are included as supplements to this paper. These include example codes for fitting the three growth models compared in the paper. Note that in each program, wt indicates weight in kg, month indicates age since birth (in months), while data is the name of the data frame that holds these variables.

1. jb.R

This includes the example code for fitting the Jenss-Bayley random effects model.

2. reed.R

This includes the example code for fitting the Reed random effects model.

3. sitar.R

This includes the example code for fitting the SITAR shape invariant random effects model. In this example the SITAR model is fitted on the log-kg and month scales, including fixed effects for the size and velocity parameters, and with a spline function with 4 knots. NOTE: Please note that the following code, based on that published by Beath (Beath K. J. 2007, Stat Med 26(12),2547-64), is a simplified version of the actual function used in the paper. The latter is available on request from Professor Tim Cole, who is currently developing a dedicated R library (tim.cole@ucl.ac.uk).

NOTE ABOUT THE DATA: None of the three datasets analyzed for this paper can be made directly accessible to the Journal's readership because of confidentiality restrictions. Data from each cohort study are available on direct request to the study coordinators. Interested readers should use these auxiliary programs to understand how the analyses in the paper were conducted.

Costanza Pizzi
Cancer Epidemiology Unit
Department of Medical Sciences
University of Turin
Via Santena 7
10126
Turin
Italy

E-mail: costanza.pizzi@lshtm.ac.uk

Dataset

Optimal Bayesian sequential sampling rules for the economic evaluation of health technologies, by P. Pertile, M. Forster and D. La Torre, Journal of the Royal Statistical Society, Series A, Statistics in Society, Volume 177, part 2 (2014), pages 419 - 438

The Matlab files baseline.m, boundary.m, oneshot.m, reward.m and wiener.m contain the code used to derive Fig. 4 of the paper and the values in Table 2 (output rows 1, 3, 4 and 5). The file supporting.pdf contains a short guide to the code.

Martin Forster
Department of Economics and Related Studies
University of York
Heslington
York
YO10 5DD

E-mail: mf8@york.ac.uk

Dataset

Re-engaging with survey non-respondents: evidence from three household panels, by N. Watson and M. Wooden, Journal of the Royal Statistical Society, Series A, Statistics in Society, Volume 177, part 2 (2014), pages 499 - 522

This paper uses data from three household panel studies and the data are only available under licence. The datasets are:

i) Household, Income and Labour Dynamics in Australia (HILDA) Survey, waves 1 to 9 (Release 9) - for details of how to order the data see http://www.melbourneinstitute.com/hilda/data/

ii) British Household Panel Survey (BHPS), waves 1 to 18 - for details of how to order the data see https://www.iser.essex.ac.uk/bhps/about/latest-release-of-bhps-data

iii) German Socio-Economic Panle (SOEP), waves 1 to 25 - for details of how to order the data see http://www.diw.de/en/diw_02.c.221180.en/research_data_center_soep.html

There are a series of programs to set up the data in a consistent format and analyse them. The files are:

i) construct resp patterns in BHPS - a Stata program that constructs a master file similar to the HILDA master file

ii) construct resp patterns in GSOEP - a Stata program that constructs a master file similar to the HILDA master file

iii) attrition patterns - a SAS program that sets up the HILDA data for analysis and combines the master files for HILDA, BHPS and SOEP

iv) create BHPS dataset for analysis - a Stata program that sets up the BHPS data for analysis

v) create SOEP dataset for analysis - a Stata program that sets up the SOEP data for analysis

vi) nonmono attrition - a Stata program that undertakes the analysis of re-engaging non-respondents and continuation of respondents

vii) nonmono attrition marginal effects - a Stata program that calculate mean marginal effects with estimated random effects

viii) nonmono attrition test interactions - a Stata program that tests interactions

ix) nonmono attrition test interaction continuity and nr reason - a Stata program that tests interactions between interviewer continuity and reason for non-response in prior wave

x) master batch program - a batch program to run the above series of programs and convert datasets from/to SAS/Stata as necessary

Nicole Watson
Melbourne Institute of Applied Economics and Social Research
Level 5, FBE Building
111 Barry St
University of Melbourne

E-mail: n.watson@unimelb.edu.au

Dataset

Evaluating nationwide health interventions: Malawi's insecticide-treated-net distribution programme, by E. Deuchert and C. Wunsch, Journal of the Royal Statistical Society, Series A, Statistics in Society, Volume 177, part 2 (2014), pages 523 - 552

Data cleaning in STATA
***************************************************************************************************

Explanation for Stata code "Data cleaning.do"

*** Data provided by Measure DHS (Stata files): http://www.measuredhs.com.
Description of the data is provided by DHS Macro.

** for the year 2000
      Household Recode (rename to Malawi HH 2000.DTA)
      Wealth Index (rename to Malawi wealth 2000.dta)
      Children's Recode (rename to Malawi children 2000.DTA)

** for the year 2004
      Household Recode (rename to Malawi HH 2004.dta)
      Children's Recode (rename to Malawi children 2004.dta)

*** GIS data provided by authors:
      Malawi GIS joint.dta provides distances to the next city and lake for each cluster in each wave

*** The following file structure is needed:
      ** raw data
            save the following data files
                  Malawi HH 2000.DTA
                  Malawi wealth 2000.dta
                  Malawi children 2000.DTA
                  rename to Malawi HH 2004.dta
                  Malawi children 2004.dta
                  Malawi GIS joint.dta

** clean data files
      all new files will be saved here
      all estimation is carried out on "estimation_file.dta"

** do adjust the path and files names in the do file "Data cleaning.do"

In case there are any question on the data cleaning file, please contact eva.deuchert@unisg.ch.

Use Statransfer to convert STATA output "estimation_file.dta" to GAUSS input "estimation_file.dat". ****************************************************************************************************

Estimation in GAUSS ****************************************************************************************************

1. Create working directory.

2. In this directory create the following subfolders:
      dat
      prg
      out
      temp

3. Copy all *.inc and *.prg in subfolder "prg".

4. Change name of working directory and, if desired, names of output files in all prg-files where indicated in these files.

5. Run programs:
      main_estimations.prg (main estimation as well as lines 2-15 in Table 3)
      cases_with_missings_excluded.prg (line 16 in Table 3)
      missing_indicator_in_probit.prg (line 16 in Table 3)
      vaccination_vitamina_diarrhoea.prg (lines 6-8 in Table 1)
      sensitivity_main.prg (all other sensitivity checks)
Output is written to subfolder "out".

Conny Wunsch
Department of Economics
VU University
Amsterdam De Boelelaan 1105
1081 HV Amsterdam
The Nethelands

E-mail: c.wunsch@vu.nl

Conny Wunsch
Department of Economics
VU University Amsterdam De Boelelaan 1105 1081 HV Amsterdam The Nethelands E-mail: c.wunsch@vu.nl.

SEARCH

SEARCH BY CITATION