Code & Datasets for a Large-n systematic review of flash flood literature

We reviewed abstracts of peer-reviewed work published between January 2000 and July 2020 in scientific journals, technical conferences and multi-author books. A total of 3748 abstracts from Web of Science Core collection were screened for relevance.

We screened all abstracts for relevance to flash floods and then further sorted relevant abstracts into three categories: core understanding, geophysical, and social.

We then did a secondary screening that further categorized and tagged the papers in the social category. This refined category will be turned into a searchable dataset for general use. Future work will hopefully automate this process and create tools for large reviews of social impact hazards.

Understanding Our File Structure

This directories in this GitHub are organized to follow the timeline of use and creation.

within /data

01_Web-Of-Science contines the original .bib files pulled from the Web of Science Database. This service only allows downloading in 500 item groupings.

02_Covidence contains the output from our initial Phase 1 screening using the software tool Covidence. The folder contans csv and .ris files of the abstracts that were deemed Yes or Maybe and the abstracts that were deemed NO. These were used in the creation of the dataframe used in the Phase 2 screening.

03_clean-screening-data includes two files: UNSCREENED-OLD-screeningDATA and UNSCREENED-screeningData. The OLD file contains the initial screening test file that was used during the development of Step 1, outlined above. The other file contains all abstracts used in the initial Step 1 screening after the process had been determined.

04_backup-screening-data containes backup copies of the dataframe as we screened.

05_screened-data holds the data for the finished full abstract screening. One file BACKUP is not to be touched and serves as a failsafe if the ACTIVE data develops issues during analysis.

06_backup-social-screening-data After the initial screening, we expanded our total included papers. The two internal directories hold backups of the first and second round of our second screening phase.

07_data_org The directory holding our outputted screening data. This is primarily csvs and xlsx files. The file, FINAL_ScreenedData_20210511_Tidy contains our cleaned data.

within /code

01_screening-code contains all R files used in the intial full and secondary screenings.

02_analysis-code contains all R files used in data analysis. This folder is currently in development.

within /exports

exports will contain plots created during analysis. it also contains exports from software VOSViewer.

within of interest

screenshots and papers that are relevant to this review

within zz-misc

will contain misclenaous, testing, or other items we are not yet ready to delete

Systematic Review Software

An R-shiny App to make systematically reviewing easier in the Natural and Social Sciences.

This app is still very much in development, but feel free to share and use! Let us know if you have any suggestions.

How to use

Working on a systematic review? Here's our current workflow.

Step 1

Create your screening data using Set-up code.R While creating this file, take some time to think through what information and questions you'll be asking for this analysis. We recommend constructing a series of Yes/No questions you can ask of every abstract. Here are ours:

1. Is the paper relevant to flash flooding or hydrology in general? (If NO - Not Relevant)
2. If Yes, is the paper about more than the underlying hydrology behind flooding (If NO - Hydrology)
3. If Yes the paper will likely be included in our analysis. What kind of paper is it?
    1. Is the paper primarily geophysically focused ()? Or socio-politically/impact focused?  (Select appropriate toggle, it can be both)
    2. Is the paper about an event? (If YES - Event)
4. Does the paper disaggregate by flood type? (if NO - Not Disaggregated)
5. Is the paper extremely relevant to our analysis and questions about the impact and vulnerability associated with flash floods? (If YES - Review Database) 

These questions directly inform the columns we created during the screening data set-up.

Step 2

With questions and screening data in hand, using MainSystReviewCode.R you can begin the screening process. Adjust the text in the shiny app to align with your questions and columns. And go to town reviewing papers.

• Step 2A - How we keep organized

• After finishing a screening session, I recommend copying the screeningData.rData file into the 03_backup-screening-data directory. Should I make this happen automatically? Yes. Have I done that yet. No. Maybe once I've finished actually doing the systematic review.

Support

Something not working? Have any questions?

Tweet at me @zentouro

Whodunit ✨

This project exists because of these excellent folks:

miriam 💻

Helen Greatrex 💻

This project follows the all-contributors specification. Contributions of any kind welcome! (emoji key)

Feel free to 🍴 this repo or clone it to your own machine and review to your hearts content.

Pull requests are also welcome. For major changes, please open an issue first to discuss what you would like to change.

This work grew out of the NASA GEO Forecasting Flash Flood Impact Project.