• 2D input like the mouse favors 3D visualization tasks where a high accuracy is demanded whereas 3D input devices like a tangible tablet allows faster adjustment if accuracy is not a key requirement. • Compared with 2D screen, 3D output largely helps spatial understanding of positions, but has limited effect for understanding rotation. • We found no evidence to suggest that the parity of dimensionality between input and output devices plays an important role on users’ performance. Thus, for 3D visualization tasks, users are not constrained to choose an input device that matches the dimensionality of the output space. They are free to use any devices that is good for a specific task. • Mouse remains an efficient tool accomplishing 3D visualization tasks both with the screen and with the augmented reality (AR) output space. It thus remains a good choice controlling a hybrid PC and AR setting. Focusing on interaction needs for scientific data exploration, we evaluated people’s performance using a 2D mouse, 3D SpaceMouse, or 3D-tangible tablet as input devices to interact with visualizations on 2D screens or stereoscopic augmented reality (AR) head-mounted displays. The increasing availability and power of immersive displays drives us to try to understand how to choose input devices, interaction techniques and output displays for the visualization of scientific data, thus to finally help us guide the interaction design for hybrid AR and PC visualization systems. With a docking task and a clipping plane placement/orientation task, we measure our participants’ performance (completion time and accuracy) with each of the different combinations of input and output. We also report on their perceived workload, their preference, and on other qualitative feedback. Results show that the mouse remains good with any display, especially for tasks that require a high accuracy. Our results highlight the potential to retain the mouse as a primary input device, and to complement it with other 3D interaction devices for specific uses.