Pitfalls in using t -statistic based breaks. Three possible problems when deriving a color scale based on t-statistics. The veritcal lines mark (-2, +2), outside of which we will consider the t-statistics significant. In (a), marginally significant t-statistics are shown in extremely saturated colors, because the range of t-statistics does not extend far past the significance threshold. In (b), a small number of t-statistics extends far past the significance threshold, resulting in less significant t-statistics to appear extremely light (e.g. the colors assigned to +5 and -5). In (c), the range of t-statistics is not symmetric around 0, resulting in slightly positive t-statistics (approximately 0 - 1) to have a light blue color instead of white or light red.