Fig. 2

Comparisons between 450K data normalization methods: percentages of comparisons identifying the newer method as better than the older. The x-axis shows which two methods are being compared (see Additional files 1 and 2 for the references of the methods, abbreviations, and any aliases used in the paper). “Introducing” (i.e., “non-neutral”) comparisons (red dots) are those from publications introducing the new method. “Neutral” comparisons (blue dots) are from subsequent studies written by authors who developed neither of the methods being compared. Two analyses are presented. In the first analysis (light dots), the percentage is calculated over papers, i.e., where the overall rank from a paper is taken as the comparison. The percentage is thus either 0% or 100% for introducing comparisons, as for each pair there is only one introducing study (that in which the newer method is introduced). Papers were excluded from this analysis if individual components of the evaluation used only a subset of the methods examined in the paper (“partial substudies”). In the second analysis (dark dots), the percentage is calculated over substudies, i.e., individual comparisons within a paper. The “number of comparisons” refers to the number of papers (for the first analysis, which takes papers as units) or to the number of substudies (for the second analysis, which takes substudies as units)