/A-Monetary-Measure-of-the-Strength-and-Robustness-of-the-Attraction-Effect

replication files for this paper

Primary LanguageHTML

Replication package for: "A Monetary Measure of the Strength and Robustness of the Attraction Effect" by Crosetto and Gaudeul, Economics Letters 2016.

Here we will put all the replication files for this paper.

Contents

In particular, you can find

  • the paper, in the /Paper and Appendices folder
  • the experimental instructions and the appendices, in the /Paper and Appendices folder
  • a presentation given in Rome in 2016 in the /Presentation folder
  • the raw data in the /Data folder
  • the analysis scripts to generate the paper's results, in /Analysis
  • the information needed to recreate the stimuli for replication, in /Stimuli

Running the analysis: howto

To replicate the statistical analysis, head to the Analysis folder. Open the Analysis.R file. It has some dependencies (listed up front; install them if you need to) and it calls all the ancillary scripts to generate all the results. These are stored in the different /Figures and /Tables folders.

Recreating the stimuli: howto

The exact stimuli used in the original experiment hare provided in the /Stimuli folder. Subjects had to choose one item out of menus of 3 items and of 6 items (only the 3-menus are used in the Economics Letters paper). You get:

  • a Menus.csv that contains python lists in which each element is a menu option.

  • a Menu_Inspector pdf that shows visually the menus, the best option, and the profit of such best option.

  • a Shapes folder that contains all the images of all the shapes. Those were in turn generated with python

  • a RegularShapeCreator.py file that takes the Menus.csv and generates the shapes. This does not work, it has been written with python2.7 that is no more supported. But it gives you at least an idea of how we created the shapes. Other options are possible and actually potentially easier, using ggplot and R or seaborn/matplotlib and python3.

A note on the experimental software

The experimental software was written in python2.7 using wxPython. It does not run any more on python3. It never will run again, unless a lot of effort is devoted to that – and this is at the moment and in the foreseeable future not going to happen.

In the end it's a simple environment in which you choose among 3 options. We document all the options in the Stimuli folder. It should be straightforward to create a software to replicate our results, without using the original experimental software.

We are sorry about this – but to be completely honest, this was the first software that Paolo Crosetto ever wrote, and it was an amateurish effort. It did work and created reliable data; it was not easy to maintain, and did not follow any best software engineering practices whatsoever. Sorry.