Dataset about actions performed by users in a supervised virtual reality environment focused on emergency training

Overall description

This dataset contains the actions of a group of users who participated in an experiment conducted by the Faculty of Computer Science of the Complutense University of Madrid (Madrid, Spain) during the periods November 18–19, 2024, and February 18–21, 2025. This study aimed to understand how the modality in which information is presented affects an interactive tutoring process in a virtual reality environment. To this end, comparisons were made between four types of modalities into which the participants were divided (12 users per group):

• No information (BaseGroup): No information is provided to the user. This is the baseline case, intended to observe user behavior when no information is given.

• Visual information (VideoGroup): Information is provided through videos that show, from a first-person perspective, how actions are performed within the virtual environment.

• Textual information (TextGroup): Information is provided through texts that explain how to perform actions and the implications of those actions.

• Visual and textual information (VideoTextGroup): Information is presented using both videos and texts as described above.

This dataset was used in the article Effectiveness of information modality in virtual reality tutorials, published in the journal Virtual Reality (DOI:https://doi.org/10.1007/s10055-025-01240-y). This article also details the virtual reality tool used, the procedure followed in the experiment, and the final results. The conclusions of the study were as follows:

1. Regardless of the modality in which information is presented, an interactive tutoring process improves user behavior. This conclusion is drawn from observing that the group without information (BaseGroup) obtained the worst results in most of the variables analyzed.

2. The effectiveness of the modality depends on the type of explanation being taught. Textual information yields better results for explanations that introduce a new, previously unknown functionality requiring a new type of action. On the other hand, visual explanations are effective for showing variations of previously introduced functionalities. For example, text is effective for explaining which button must be pressed to pick up or drop an object, whereas video is effective for showing how to place objects in specific locations.

These conclusions and the results presented in the aforementioned article were obtained using the provided dataset. This dataset includes the following files:

• raw/: Events related to users’ direct actions in the virtual environment and their questionnaire responses:

• user.csv: List of users.

• gameplay.csv: Events related to the objectives/tasks proposed to and completed by the user.

• user_inputs.csv: Controller button presses performed by the user.

• item_interaction.csv: Interactions with virtual objects.

• tutorial_taks.csv: Similar to gameplay.csv, but focused exclusively on tasks completed in the tutoring scenarios.

• move.csv: Events related to user movement both in the real world and in the virtual world.

• questionnaire.csv: Users’ responses to the proposed questionnaires. The questionnaires used are as follows:

• Personal user information such as age, gender, level of education, or familiarity with virtual reality technology, among others.

• Questions related to the virtual environment. For example, which buttons must be pressed to perform certain actions or which objects must be used at specific moments.

• Questionnaire to assess user experience. The questionnaire used was the User Experience Questionnaire (UEQ: https://www.ueq-online.org/).

• Questionnaire to assess system usability. The questionnaire used was the System Usability Scale (SUS).

• Questionnaire to assess the cognitive load experienced by the user. The questionnaire used was the NASA-TLX (Task Load Index).

• processed/: Files containing users’ data processed to enable the analysis presented in the aforementioned article.

• docs/: Documents providing a detailed description of the variables stored in each dataset, the data collection methodology, and a description to reproduce the results shown in the aforementioned research article.

Keywords

Virtual reality, tutorials, learning processes, information modality, training, human-computer interaction

Methodology

The data presented here were collected in a controlled environment. These data were collected in two ways:

1. Data related to users’ actions within the virtual reality environment were stored in real time in .csv files.

2. Data related to questionnaires were collected through Google Forms.

The document methodology.md details how the data were collected.

Funding

This publication is part of the R&D &I projects DARK NITE, PID2023-146308OB-I00, funded by MICIU/AEI/10.13039/501100011033/ and ERDF, EU “A way of making Europe”; ADARVE (SUBV20/2021), funded by the Spanish Council of Nuclear Security; CANTOR (PID2019-108927RB-I00), funded by the Spanish Ministry of Science and Innovation; and EA-DIGIFOLK: An European and Ibero-American approach for the digital collection, analysis and dissemination of folk music project (grant agreement no. 101086338), funded by the European Commission.

Related works

2025: Effectiveness of information modality in virtual reality tutorials. Alejandro Villar y Carlos León. Virtual Reality 30, 2 (2026). DOI: https://doi.org/10.1007/s10055-025-01240-y.

License

Creative Commons Attribution 4.0 International