tmalsburg

A potential pitfall when running Ibex experiments on Amazon Mechanical Turk

Ibex does Latin squares in a way that can potentially have serious unintended consequences in the form of spurious effects.

The problem: When you submit an experiment to Amazon Mechanical Turk, a lot of workers will immediately jump at it but the rate of participation quickly decays (the distribution over time often looks like an exponential decay). For every participant, Ibex selects a stimulus list based on an internal counter and this counter is incremented when a participant submits their results. Unfortunately, this means that the initial wave of participants all work on the same list of the Latin square and this list will therefore be strongly overrepresented. This can lead to strong spurious effects that are not due to the experimental manipulation but due to between-item differences. This is an easy-to-miss problem and I would not be surprised if some published results obtained with Ibex were false because of this prob

tmalsburg

\usepackage{xpatch}% or use http://tex.stackexchange.com/a/40705

\def\makenamesetup{%
  \def\bibnamedelima{~}%
  \def\bibnamedelimb{ }%
  \def\bibnamedelimc{ }%
  \def\bibnamedelimd{ }%
  \def\bibnamedelimi{ }%
  \def\bibinitperiod{.}%
  \def\bibinitdelim{~}%

tmalsburg

Predict vs simulate in lme4

For this investigation we are going to use the sleepdata data set from the lme4 package. Here is the head of the data frame:

tmalsburg

library(lme4)

head(sleepstudy)
summary(sleepstudy)

# Relabel subjects:
d <- sleepstudy
d$Subject <- factor(rep(1:18, each=10))

# Fit model:

tmalsburg

#!/usr/bin/env python
# -*- coding: utf-8 -*-

# Written by Titus von der Malsburg <[email protected]>, March 2014.

# This is a simple script for converting vCard files to
# org-contacts.  There is one mandatory argument: the name of the
# vCard file.  The result is printed to standard out.

# Usage: