Improving Nutrition Status for under 5 children in Zambezia and Nampula project baseline and endline survey data analysis and reporting workflow

This repository is a docker-containerised, {targets}-based, {renv}-enabled R workflow developed for the data management, data analysis, and reporting of the Improving Nutrition Status for under 5 Children in Zambezia and Nampula project baseline and endline survey.

Repository Structure

• data/ contains intermediate data outputs produced by the workflow including an endline survey codebook describing all variables of the endline survey;
• R/ contains functions created for use in this workflow;
• reports/ contains literate code for R Markdown reports generated in the workflow;
• outputs/ contains compiled reports and figures;
• auth/ contains credentials for Google service account (see below for more information);
• docs/ contains archived high frequency data checks reports produced during the implementation of the survey;
• did/ contains R script for performing the difference-in-difference analysis using the results tables produced by the analysis pipeline/workflow;
• _targets.R file defines the steps in this workflow’s data management and analysis pipeline.

Reproducibility

R package dependencies

This project requires R 4.2.2 (2022-11-10 r83330). This project uses the {renv} framework to record R package dependencies and versions. Packages and versions used are recorded in renv.lock and code used to manage dependencies is in renv/ and other files in the root project directory. On starting an R session in the working directory, run renv::restore() to install R package dependencies.

System requirements

This project’s data management and analysis pipeline was built to utilise multi-threaded parallel computing to make calculations faster and more efficient. This project’s pipeline uses 4 computing threads. To run without issues, ensure that you use a machine with at least 5 computing threads when reproducing this pipeline (4 threads for the pipeline and the remaininig threads to run the machines regular processes).

With multi-threaded computation, the entire pipeline completes between 30 minutes to an hour (depending on CPU speed and RAM size). Without multi-threading, the entire pipeline completes between 3-5 hours.

Data management and analysis

This project uses the {targets} package to create its data management and analysis pipeline as defined in the _targets.R file.

To execute the data management and processing workflow for baseline and endline survey data, run:

targets::tar_make(baseline_data_weighted)
targets::tar_make(endline_data_weighted)

The schematic figure below summarises the steps in the data management and processing workflow:

✔ Successfully auto-authenticated via auth/mozambique-s3m-e9da207bc2a3.json

graph LR
  subgraph Graph
    xaa8b198e46c12205(["baseline_data_downloads"]):::uptodate --> x34303849e9b0482a(["baseline_raw_data_spss"]):::uptodate
    x3292f12efa6fc224(["endline_raw_data"]):::uptodate --> x9fee25520ed83a1a(["endline_sample_weight"]):::uptodate
    xa54f854256062899(["survey_sampling_list_endline"]):::uptodate --> x9fee25520ed83a1a(["endline_sample_weight"]):::uptodate
    x8c16f11d9b877f97(["endline_data_processed"]):::uptodate --> x1a4d1d913b004605(["endline_data_weighted"]):::uptodate
    x9fee25520ed83a1a(["endline_sample_weight"]):::uptodate --> x1a4d1d913b004605(["endline_data_weighted"]):::uptodate
    xcbf2ec7505f9fcf3(["baseline_raw_data_stata"]):::uptodate --> xaa47d78bda713fd1(["baseline_sample_weight"]):::uptodate
    x500c8b308ead6406(["survey_sampling_list"]):::uptodate --> xaa47d78bda713fd1(["baseline_sample_weight"]):::uptodate
    x500c8b308ead6406(["survey_sampling_list"]):::uptodate --> xa54f854256062899(["survey_sampling_list_endline"]):::uptodate
    xaa8b198e46c12205(["baseline_data_downloads"]):::uptodate --> xcbf2ec7505f9fcf3(["baseline_raw_data_stata"]):::uptodate
    x3292f12efa6fc224(["endline_raw_data"]):::uptodate --> x8c16f11d9b877f97(["endline_data_processed"]):::uptodate
    xb093acd1bf068c99(["survey_endline_choices"]):::uptodate --> x8c16f11d9b877f97(["endline_data_processed"]):::uptodate
    x34303849e9b0482a(["baseline_raw_data_spss"]):::uptodate --> x3fb2eec55e3093c2(["baseline_data_processed"]):::uptodate
    xcbf2ec7505f9fcf3(["baseline_raw_data_stata"]):::uptodate --> x3fb2eec55e3093c2(["baseline_data_processed"]):::uptodate
    x13d5233859395ad9(["survey_endline_form_id"]):::uptodate --> xb093acd1bf068c99(["survey_endline_choices"]):::uptodate
    x3fb2eec55e3093c2(["baseline_data_processed"]):::uptodate --> xcbc9dbd4b57c3d95(["baseline_data_weighted"]):::uptodate
    xaa47d78bda713fd1(["baseline_sample_weight"]):::uptodate --> xcbc9dbd4b57c3d95(["baseline_data_weighted"]):::uptodate
    x2ada552bbfadb16c(["survey_sampling_list_download"]):::uptodate --> x500c8b308ead6406(["survey_sampling_list"]):::uptodate
    x500c8b308ead6406(["survey_sampling_list"]):::uptodate --> x3292f12efa6fc224(["endline_raw_data"]):::uptodate
    xa54f854256062899(["survey_sampling_list_endline"]):::uptodate --> x3292f12efa6fc224(["endline_raw_data"]):::uptodate
  end
  classDef uptodate stroke:#000000,color:#ffffff,fill:#354823;
  classDef none stroke:#000000,color:#000000,fill:#94a4ac;

Loading

This produces the following outputs:

• R object for processed baseline data that includes probability weights for use in weighted analysis;
• R object for processed endline data that includes probability weights for use in weighted analysis.

To execute the data analysis workflow for the baseline and endline survey data, run:

targets::tar_make(dplyr::starts_with("baseline"))
targets::tar_make(dplyr::starts_with("endline"))

The schematic figure below summarises the steps in the data analysis workflow:

✔ Successfully auto-authenticated via auth/mozambique-s3m-e9da207bc2a3.json

graph LR
  subgraph Graph
    x2ae00276a0a5b418(["overall_results_all"]):::outdated --> x0bbe9f288492eeec(["overall_results_xlsx"]):::outdated
    xb8de1d18ad2cc4f6(["baseline_results_all"]):::outdated --> xaa3f6c085cefc626(["baseline_results_csv"]):::outdated
    xa8885ae938ad1519(["overall_results_subset"]):::outdated --> x14c4c0c6cf2ca972(["baseline_results_subset"]):::outdated
    x2ae00276a0a5b418(["overall_results_all"]):::outdated --> x02fbd9452a16eea6(["baseline_survey_results_report"]):::outdated
    xcf2fba5ab534b33c(["endline_results_subset"]):::outdated --> x6ddc0eabdf53e510(["endline_results_subset_csv"]):::outdated
    x14c4c0c6cf2ca972(["baseline_results_subset"]):::outdated --> x969d7dd50bd8a411(["baseline_results_subset_csv"]):::outdated
    xa8885ae938ad1519(["overall_results_subset"]):::outdated --> xcf2fba5ab534b33c(["endline_results_subset"]):::outdated
    x2ae00276a0a5b418(["overall_results_all"]):::outdated --> xcd1b6eeb6bb99aba(["endline_survey_results_report"]):::outdated
    x2ae00276a0a5b418(["overall_results_all"]):::outdated --> xdf25b6550ed0d67d(["endline_results_all"]):::outdated
    xb8de1d18ad2cc4f6(["baseline_results_all"]):::outdated --> xe30152fe1d53bcbe(["baseline_results_xlsx"]):::outdated
    xdf25b6550ed0d67d(["endline_results_all"]):::outdated --> xa5c7be8027626ce4(["endline_results_csv"]):::outdated
    xdf25b6550ed0d67d(["endline_results_all"]):::outdated --> x87d5a544c4ad6d2e(["endline_results_xlsx"]):::outdated
    x2ae00276a0a5b418(["overall_results_all"]):::outdated --> xb8de1d18ad2cc4f6(["baseline_results_all"]):::outdated
    x26387a2c59ebbcd2(["endline_survey_food_security_results_report"]):::outdated --> x26387a2c59ebbcd2(["endline_survey_food_security_results_report"]):::outdated
  end
  classDef outdated stroke:#000000,color:#000000,fill:#78B7C5;
  classDef none stroke:#000000,color:#000000,fill:#94a4ac;

Loading

This pipeline produces the following outputs:

• Baseline results tables in CSV and in XLSX format;
• Baseline results tables subset in CSV format which is used for the difference-in-difference analysis;
• Endline results tables in CSV and in XLSX format;
• Endline results tables subset in CSV format which is used for the difference-in-difference analysis.

These outputs are saved by the pipeline in the outputs directory with the following filenames:

• baseline_results.csv;
• baseline_results.xlsx;
• baseline_results_subset.csv;
• endline_results.csv;
• endline_results.xlsx;
• endline_results_subset.csv.

Difference-in-difference analysis

The difference-in-difference analysis was performed using a single R script found in the did/ directory.

To run the difference-in-difference analysis, one should first run the data analysis pipeline describe above using the following command in R:

targets::tar_make(dplyr::starts_with("baseline"))
targets::tar_make(dplyr::starts_with("endline"))

which will produce the necessary results outputs needed for the difference-in-difference analysis.

Then, one should run the difference-in-difference R script using the following R command:

source("did/did.R")

The outputs of the difference-in-difference R script are:

• an HTML file called analysisDiD.html;
• all the plots (in .png format) showing the difference-in-difference results for each indicator on which the analysis was performed.

These outputs are saved in the did/ directory after the difference-in-difference R script is run.

This repository doesn’t contain the outputs of data management and analysis workflows described above. The user is expected to use this code along with the data from UNICEF Mozambique (restricted; request access only; see below) to run the workflows and produce the outputs themselves. All outputs are available from UNICEF Mozambique on request.

Encryption

This repository uses git-crypt to enable transparent encryption and decryption of the .env file and the auth/ directory.

The .env file contains:

• variables for authenticating with Google Cloud services account setup for this project for accessing the baseline survey data;

• variables for accessing the endline survey data direct from ONA;

• variables for SMTP_PASSWORD and EMAIL_RECIPIENTS which are used for sending the email updates for the high frequency data checks.

The auth/ directory contains credentials for Google service account.

Those who would like to reproduce the results of this project will require ability to decrypt the .env file and the auth/ directory.

To be able to work on this repository, a user/collaborator on this project will need to:

• Create their own PGP (Pretty Good Privacy) public and private keys; and,

• Share their public key to the authors and request for it to be added to the repository.

Once added, a collaborator can now decrypt the .env file and the auth/ directory after pulling/cloning the repository by running:

git-crypt unlock

on the terminal.

Authors

• Mark Myatt
• Ernest Guevarra

License

The datasets for both baseline and endline survey (not included in this repository) are owned by UNICEF Mozambique. Access to these datasets is restricted. Communicate with UNICEF Mozambique to request access to the survey datasets.

The code included in this repository is owned by the authors and is licensed under a GNU General Public License 3 (GPL-3).

Feedback

Feedback, bug reports and feature requests are welcome; file issues or seek support here.