March 14, 2019

Answering the “why” question: piecing together multiple pieces of causal information

((This is an abstract I just submitted to the 2019 DGeval conference in Bonn.))

We evaluators often ask questions about what causes what — because we want to know how things work, and because we want to know how stakeholders think things work, because this will influence their behaviour. As a species, our urgent task is to work out how to shift our civilisation onto a sustainable course, so we need good tools for understanding people’s behaviour and their reasons for it.

We human beings are good at learning and communicating causal connections. Of course our causal beliefs are just as vulnerable to bias and illusion as the rest of our beliefs. But we humans are still the best source of causal information available especially in those complex settings which involve other humans. We usually know how to calm a frightened child, or open a stuck door, and we each use this kind of information thousands of times a day. We even have opinions about what causes climate change, or what needs to be done to make more people better at recycling.

But as evaluators, we often prefer to ask stakeholders either quantitative questions (the opinions of the very same unreliable humans, but stripped of causal information and squeezed into a numerical format) or qualitative questions, which we hope to code into themes and categories and narrative summaries, with no special role for causal information.

This presentation looks at our options as evaluators to take a third way to gather information, namely to directly ask our stakeholders the why question”: what causes what?

We will look at the tools available to code, synthesise and analyse multiple fragments of causal information, each of which is likely to be unreliable and inaccurate in indifferent ways, and to present the synthesised findings, perhaps in the form of a network diagram or theory of change”.

In particular, this paper presents a new free and open-source online app designed specifically for this kind of task: to import multiple fragments of causal information, to code these fragments as links between different factors, to aggregate and analyse these multiple sets of links (do women mention this link more than men?) and to display the results.

Having tools like this which enable us to work more easily with these kinds of sets of multiple fragments of causal information will enable us as evaluators and researchers to quickly design and execute a new kind of research: one in which we simply ask respondents one or more why questions” from a relevant domain.

February 1, 2019

Better ways to present country-level data on a world map: equal-area cartograms

I’m just working on a report for the IFRC. There is lots of data from 190 different Red Cross Red Crescent National Societies around the world1.

Cutting to the chase:

.. that is our improved map-like solution, and I’ll show the code further down. What it is really missing is a nice stylised outline of continents in the background.

Why do I claim this is a better viz? Sometimes we’ve been using ordinary choropleths:

(In this case we are looking at the percentage of women on the governing board.) These maps look nice but they are really poor at presenting actual information IMHO. You can’t even tell which continent is doing better, and did you spot that Ghana has a pretty poor percentage? Did you spot the high scores for Georgia and Latvia and Moldova?

Equal-area aka equal-size cartograms are better, but I only found examples with ISO3 codes, which are really user-unfriendly. So I got my son to retype rbrath’s map into a spreadsheet and R does the rest.

You need this csv file which just gives you the iso and country names in terms of x and y coordinates.

Then you need a data.frame called say xx which has a column called iso3 and your data. Merge them together and there you go.

xx %>% dplyr::merge(maps,by = "iso3",all.y=F) %>% 
  ggplot(aes(x,y,fill=fem_GovBoard,label=name %>% str_wrap(8)))+geom_tile()+geom_text(hjust=.51,lineheight=.7,color="black")+
  scale_fill_gradient(low = "red",high="white")+
  theme(panel.background = element_rect(fill="#445566"),panel.grid = element_line(color="#445566"))

  1. FDRS, Federation-wide Databank and Reporting System

r nerdvana
January 11, 2019

A starter kit for reproducible research with R

This accompanies my draft short article for the UKES Bulletin: A reproducible workflow for evaluation reports.

Here are just two files which together get you started with a bare-bones reproducible research project.

  • Install and install R and RStudio.
  • Download this little example of a reproducible source file and an accompanying example Excel file describing the heights and ages of a bunch of boys and girls in a school.
  • Save them together in a new folder.
  • Double-click the source file to open it in RStudio. The source file has a couple of commands to read the Excel file, clean it and make a table and a chart, as well as plain text which becomes the headings and body of the document.
  • Press the blue Knit” button in RStudio, you will get a beautiful Word document.
  • Try editing the text and then press Knit” again to see what happens.
R reproducibleResearch

This blog by Steve Powell is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License, syndicated on r-bloggers and powered by Blot.
Privacy Policy