Project Overview

January 1, 0001

Structure

MAPS is a crowd-sourced science experiment in the same vein as Many analysts,One Dataset but with an epidemiological dataset from the Avon Longitudinal Study of Parents and Children (ALSPAC or Children of the 90s study) concerning youth depression and computer use. The study is split up into two main phases, you are free to participate in either or both. You will gain authorship by participating in phase 1.

Key project information can be found in Get Involved page.

Phase 1: Data analysis

  1. Upon signing up (either individually or as a team) we will ask you to provide us information about yourself, your area of expertise and your prior beliefs on the topic.

  2. Once you’ve completed this form we will provide you with an annonymised dataset containing approximately 15’000 observations of over 80 variables and we will ask you to answer the following question:

    Is computer use during weekdays and weekends at 16 years old associated with depression at 18 years old?

  3. You are free to define depression and computer use however you like, as long as it is in the form of an odds ratio.

  4. We will ask you to submit your analysis as a RMarkdown or Jupyter notebook (meaning you must do it in R or Python) where you clearly show all the decision you made. We should be able to run it and get the same answer as you.

  5. Once we have all your answers we will conduct a multiverse analysis on ALL your models. That is, we will take all reasonable analytical choices (including type of model) and run them all.

Phase 2: Visualisation challenge

  1. At this point we will have a project database of different types of models, model specifications, effect sizes (as well as some model diagnostics) and information on the analysts.

  2. The visualisation challenge is to make sense of all this data! In particular we’d really like visualisations that allow us to understand the following questions:

    1. How do the teams results compare to the multiverse of results and to each other?
    2. Does particular types of expertise or prior beliefs affect the teams results?

  3. You are free to interpret “compare” in any way you wish. You are also free to extract other features of the project database and visualise those (this will be a great dataset for some unsupervised learning!).

  4. The visualisations will be judged and a prize will be given for the best one (to receive the prize there may be eligibility restrictions).

Timeline

Comments

comments powered by Disqus