Mathematics

Experiments on Visual Inference

Several experiments were conducted to validate and apply visual statistical inferential procedure. Human subjects were recruited through Amazon Mechanical Turk (MTurk) and redirected to these experiment sites for providing evaluation on lineups. The results were published in numerous papers as linked in the description. These experiments were funded in parts by NSF grant DMS 1007697.

Experiment Description
Experiment #1 Boxplots of residuals by group. Experiments #1-#3 validate the lineup protocol in the setting of linear models; results are published in "Validation of Visual Statistical Inference, Applied to Linear Models", JASA 2013.
Experiment #2 Scatterplots with trend line superimposed
Experiment #3 Scatterplots with added noise but no trend line
Experiment #4 Polar and Euclidean barcharts. Experiments #4-#5 form the experimental basis of "Graphical Tests for Power Comparison of Competing Designs", IEEE Transactions on Visualization and Computer Graphics, 2013.
Experiment #5 Assessing vertical shifts between groups in dotplots, boxplots, histograms and density plots.
Experiment #6 Boxplot variations. This experiment forms the basis for the results published in Why old is good for boxplots
Experiment #7 Scatterplots and dotplots of coloured groups. This experiment forms the basis for the results published in "Visual Statistical Inference for Large p, Small n Data", Computational Statistics, to appear.
Experiment #8 Sine illusion charts. This experiment forms the basis for the results published in "The curse of three dimensions: Why your brain is lying to you" best student paper award from ASA Section on Statistical Graphics, 2014.
Experiment #9 Interaction between two groups. This experiment forms a part of the results published in "Visual Mining Methods for RNA-Seq data: Examining Data structure, Understanding Dispersion estimation and Significance Testing" Journal of Data Mining in Genomics & Proteomics, 2014.
Experiment #10 Variations of Q-Q plots. Manuscript of results is in preparation.
Experiment #11 Diagnostics for hierarchical linear models. Manuscript of results is in preparation.
Experiment #12 Testing for expression difference in RNA-Seq data - a pilot study. Manuscript of results is in preparation.
Experiment #13 Variations of Q-Q plots; this is another replication of the same experiment #10 in a different time with different subject. Manuscript of results is in preparation.