A common assumption in data used for popular behavioral analysis modalities like typing, gaits, and swipes is that features extracted from the data follow the normal distribution. The assumption of normality impacts key facets of research, such as decisions of sampling techniques and classification models and performance and results from the resulting systems. Through the analysis of eight open-access datasets collected on tablets and phones (gait, swipes, and typing), and desktops (typing), we question the assumption of normality in the extracted features. Using nonparametric normality tests (Lilliefors test and Shapiro–Wilk test), we test the null hypothesis “the test sample comes from a normal distribution” and examine features that have been popularly used in the literature from these activities. In most cases, less than 25% of the tested samples have p-values \(>\) 0.05, which asserts that a majority of features do not follow a normal distribution. Although nonnormality in keystroke latencies on the desktop has been shown in the literature, no previous work examined a large umbrella of biometric data, such as keystroke latencies, gait, and swipe data on desktops, phones, and tablets. We also provide alternate solutions to address the nonnormality in mobile or wearable device-based behavioral biometric data. Our work raises the questions, “Should the assumption of normality be the norm in behavioral biometric modalities?” We posit that our results will change how behavioral data analysis is approached by emphasizing the validity of assumptions about the underlying data. This study can potentially impact a large body of work in desktop, mobile, and wearable device-based behavior analysis.