Statsplorer

Abstract

The difference between mouse and trackpad is significant Welch's t(20) = 2.39, p = .27… time technique The user selects variables of interest The user selects levels to compare Statsplorer shows suitable graphs Statsplorer automatically tests statistical assumptions and executes appropriate tests The user interprets the results The user learns about statistical analysis process Statsplorer visualizes the results Statsplorer generates report text 0 1 s m a ll m e d iu m la rg e 0.51 d Figure 1. Basic process of statistical analysis with Statsplorer ABSTRACTEach step of statistical analysis requires researchers to make decisions based on both statistical knowledge and the knowledge of their own data. For novice analysts, this is cognitively demanding and can lead to mistakes and misinterpretations of the results. We present Statsplorer, a software that helps novices learn and perform inferential statistical tests. It lets the user kick-start data analysis from their research questions. Statsplorer automatically tests necessary statistical assumptions and uses visualizations to guide the user in both selecting statistical tests and interpreting the results. We compared Statsplorer with a statistics lecture and investigated how Statsplorer prepares novices for learning statistics in an AB/BA crossover experiment. The results indicates that using Statsplorer prior to the lecture leads to significantly better test scores in understanding statistical assumptions and choosing appropriate statistical tests. Statsplorer is open-source and is available online at: http://hci.rwth-aachen.de/statsplorer.

Assessing the Graphical Perception of Time and Speed on 2D+Time Trajectories

“…This approach has been recommended for reporting statistical results in HCI over the traditional null hypothesis significance testing (with p-values only), which leads to dichotomous thinking [31]. It has seen increased use recently in HCI and visualization (e.g., see [17,30,42,69,74,79]).…”

Designing for Preregistration in Practice: Multiple Norms and Purposes

“…There is an opportunity to better integrate preregistration into existing tools that aid the experimental design and analysis process in order to improve report quality. Interactive tools for experimental design and/or analysis (for example, TouchStone Mackay et al (2007), which aids in experimental design, or StatSplorer Wacharamanotham, Subramanian, Völkel, and Borchers (2015), which aids in analysis-even suggesting text to report a result in a paper), could have support for preregistration explicitly designed into them. Such a tool could aid a researcher in designing an experiment, preregistering that design, submitting that preregistration to an online registry, analyzing the results (while keeping track of any protocol deviations), and even give suggested text for writing up results with clear indications of what is exploratory/confirmatory and where protocol deviations occurred.…”

In-depth examination of spatiotemporal figures in open reproducible research

“…Allowing users to control the content of a figure can improve their understanding (Wacharamanotham et al, 2015). Moreover, interactive figures might tackle issues of static illustrations such as hidden details (Weissgerber et al, 2016).…”

“…Similar to Jupyter, RMarkdown enables readers to change the code and to incorporate interactive figures (R Studio, 2017). Interactive figures are also automatically created by Statsplorer (Wacharamanotham et al, 2015) which supports students in understanding statistical tests. Adnan et al (2016) observed that even basic interactions, e.g.…”

“…sliders), readers can change the model used for the interpolation map. Such controls might improve a reader's understanding of the analysis and results (Wacharamanotham, Subramanian, Völkel, & Borchers, 2015).…”