Keywords

1 Introduction

Cybersickness may be described generally as a bodily discomfort induced by exposure to VR content. It may be one of the enduring hindrances in order to use the full potential of VR technology. The symptoms of cybersickness are similar to the ones which people suffer from when they get motion sick. There is a number of different definitions of cybersickness. In general, cybersickness is a set of symptoms produced by VR exposure that temporarily affects the well-being of the VR user and impedes the use of VR [4]. The most common symptoms are nausea, vertigo, disorientation and pallor [10]. It is most often categorized as a form of simulator sickness [16], although there is a distinction between the experience of cybersickness and simulator sickness. The dominant symptoms are different, i.e., in cybersickness, the disorientation symptoms tend to occur more often.

1.1 Measuring Cybersickness

There are two types of cybersickness measures, objective and subjective. Objective measures include physiological markers, such as bradygastric activity, respiration rate [3, 8], heart rate [15], galvanic skin response [5] and behavioral measures, such as early termination of VR experience [9], delivery of incomplete task [15], impeding the further VR use [2]. Subjective measurements, on the other hand, are different multi-item questionnaires, such as Simulator Sickness Questionnaire (SSQ; [6]). The subscale measuring the experienced symptoms level consists of 16 items. It gives a possibility to generate a total sickness score and the levels of oculomotor discomfort, disorientation, and nausea [18]. It is commonly appreciated as a measurement of cybersickness in VR.

Other measurements that are offered for evaluating cybersickness are Cybersickness Questionnaire (CSQ; [17]) and Virtual Reality Sickness Questionnaire (VRSQ; [7]). CSQ was developed to capture the occurrence of cybersickness in VR context, and because it is based on SSQ, uses the SSQ answers to estimate dizziness and troubles with focusing. The VRSQ is a questionnaire which is aimed to measure motion sickness in a virtual reality environment [7]. It is a modified version of SSQ, and consists of 9 items grouped into two components, namely oculomotor and disorientation.

1.2 Predicting Cybersickness

Cybersickness Susceptibility Questionnaire predicts the tolerance to VR exposure. The questionnaire consists of three parts and it is aimed to be used before the VR environment exposure. It was created to demonstrate the effect of biological, chemical and psychological occurrence of cybersickness [4]. The first part of CSSQ collects demographic questions, the second part evaluates the physical health and fitness levels. There are questions aimed at investigating the headaches, stomach complaints, alcohol and substance consumption within 24 h prior to the VR exposure as well as acute complaints, as all of these are the indicators of a higher probability of cybersickness occurrence [4]. The last part of the questionnaire evaluates the sensibility to motion sickness caused by various types of motion, such as driving as a passenger, traveling by train, or sitting on a roller coaster ride.

Most of the questionnaires which measure cybersickness focus mainly on the outcomes, not on what triggers the problem to occur. It is important to combine both the physiological and psychological factors of cybersickness to better understand this phenomenon, as the occurrence of cybersickness is one of the reasons impeding people from the use of VR [14]. Moreover, there is some evidence that cybersickness may be negatively related to presence in VR. The sense of presence presumably suppresses cybersickness because the attention is directed to the unpleasant motions from the body [18]. Thus, in some studies, screening potential participants for their susceptibility to cybersickness may help reduce the risk of unpleasant or poor VR experiences for participants and the dropout rate for researchers simultaneously.

The main objective of the presented study is to prepare the Polish version of CSSQ and test its psychometric characteristics to evaluate its usefulness in research and other VR applications, e.g., physiotherapy or education.

2 Method

The adaptation of the scale was a part of a larger study on emotional reactions to immersive art. The study design was similar to that used by the Authors of the original scale [4]. The Nightsss experience has been chosen as a material to be shown to participants, as it contains two parallel versions, i.e., an interactive and a cinematic one [11, 12]. The content of both was identical, and the difference was in the interactivity level. In the interactive version, the participants were allowed to walk freely in the 3D environment and have some interactions with objects, while in the Cinematic VR (CVR) version, they were only able to watch the experience in the form of a 360 3D movie. The procedure was conducted on the HTC Vive Pro Eye HMD with the wireless addon and the Intel I9 2.8 GHz PC with Gigabyte Geforce RTX 3070. The study obtained ethics approval from the Ethics Committee of the Institute of Psychology Polish Academy of Sciences.

2.1 Participants and Apparatus

Eighty five Caucasian (Polish) participants (26 male, 58 female, 1 nonbinary or other) took part in the experiment. The mean age was respectively 31, 31 (SD = 9 in both groups), and 22. Levene’s test for homoscedasticity of variance detected no differences (p = .296).

2.2 Measures

Polish adaptation of SSQ [1] was used. The original scale is divided into two parts: one with the questions about the current health state, alcohol, and medicine consumption and the other measuring the symptoms of simulator sickness, on 4 points scale, from none to severe. The results range was 0–179.5. We omitted non-diagnostic questions as the whole procedure was aimed to be not very bothersome for participants.

CSSQ [4] consists of three parts. The first collects demographic data, second consists of questions considering the health and fitness of participants, including two questions on a 5-point Likert scale and 11 binary ones. The last part measures vulnerability to motion sickness with thirteen questions on a 5-point Likert scale from 0 - very rarely to 4 - very often (Table 1).

2.3 Stimuli and Procedure

After entering the vrLab, participants were asked to fill the first part of the Polish adaptation of the SSQ (demographic data, questions about health condition, past illness, alcohol, and drug intake, and sleep quality) [6] together with CSSQ-PL. Next, they were informed about the equipment, procedure, the possible effects of using the VR headset, and the collected data. Then the GSR and HRV electrodes were placed. In the first part of the VR experiment, the participants stood in an area limited to 1,5 sq m and viewed short video clips of neutral, positive, and negative valence from the Public Database Of 360 Videos With Corresponding Ratings Of Arousal And Valence [13]. Next, the participants saw either CVR standing in one place or interactive “Nightsss” where they were able to move around an area limited to about 7 sq m. The experience lasted around 8 min. After removing the VR goggles, the participants were asked to fill out the SSQ and evaluate their feelings and opinion with open-ended questions.

2.4 Validation

The language adaptation to Polish was performed as follows: (1) the scale was translated by three independent researchers from English to Polish, (2) all three versions were compared to select the final version, (3) the subscale’s reliability and validity were tested. The questions with a binary answer scale in the “Health and fitness” subscale were coded as yes = 1 and no = 0. Reliability of the “Health & Fitness” subscale was tested by point biserial correlation between the item score and the total subscale score. The “Motion Sickness” subscale’s reliability was evaluated by analyzing the alpha Cronbach’s scores if an item would be removed (Table 1). Next, we calculated the sum of the “Health and fitness” and “Motion sickness” subscales as a sum score of CSSQ items excluding the demographic variables. We also calculated the total score of the SSQ-post and its subscales (Nausea, Oculomotor, Disorientation), following Biernacki et al. [1] to test the validity of the adapted scale. Next, we visually analyzed the distributions of the variables and normalized the variables to a 0–1 scale, and calculated the correlation matrix.

Table 1. Cybersickness Susceptibility Questionnaire - PL.
Full size table

3 Results

3.1 Language Adaptation

The final version of the Polish scale and the original scale published by Freiwald et al. [4] can be found in Table 1. Individual translations used to build the final scale can be obtained on request from the Authors.

3.2 Reliability

Due to the content of the “Health and Fitness” subscale, which refers mainly to the medical records and is constructed mainly on binary variables, the reliability analysis was carried out only for the “Motion sickness” subscale. The overall Cronbach’s \(\alpha \) = .87, which indicated very high reliability. Cronbach’s \(\alpha \) and items correlations per item can be found in Table 1. Based on the analysis, we also prepared a short version of the “Motion sickness” subscale, which included six items (selected on the basis of itemwise analysis to maintain high internal consistency - Cronbach’s \(\alpha \) above .80). The subscale is characterized by high reliability (\(\alpha \) = 0.82) and can be used for a quick screening. Items selected for the short scale are marked with an asterisk in Table 1.

3.3 Distributions

Overall, most participants reported very low to low susceptibility to cybersickness (M = 7.36, SD = 6.86) and very low to low cybersickness after the VR exposure (M = 16.37, SD = 15.93). The distributions were strongly left-skewed for all subscales (see Fig. 1).

Fig. 1.
figure 1

Distribution of subscales

Full size image
Fig. 2.
figure 2

Difference between the prediction (CSSQ) and the obtained cybersickness levels (SSQ)

Full size image

3.4 Validity

Due to the skewness, we decided to test the predictive power of the CSSQ scale by applying the authors’ original approach. First, we normalized both CSSQ and SSQ total scores to 0–1 scale, and extracted the difference between the estimated susceptibility (CSSQ) and the obtained cybersickness levels (SSQ) to evaluate whether these scores matched. The variable had close-to-normal distribution with a mean of −.011 and standard deviation of .212. The mean was very close to zero and 80% of the participants had scores within one standard deviation (see Fig. 2).

Moreover, we tested the strength of the linear association between the subscales and demographic variables (Fig. 3). We found a strong positive correlation of SSQ with CSSQ (r = .58), a moderate correlation of gender with both CSSQ and SSQ. Analyzing the correlations between the CSSQ and SSQ subscales, we can observe that “Health & Fitness” subscale was weakly correlated only with SSQ total score (r = .32), while the “Motion Sickness” subscale correlated weakly to moderately with all SSQ measures. All subscales within the same construct correlated in the expected direction. Age did not correlate with any of the variables. Moreover, female gender and lower height were related to higher levels of motion sickness sensitivity and overall cybersickness, although the association was relatively weak (r = .35\(\times \).03). Since the sample consisted of more women than men, and there was a strong negative correlation between gender and height, it is possible that the correlation with height was an artifact.

Fig. 3.
figure 3

Pearson’s correlation matrix. Scales were normalized to 0–1. Visible correlations are significant at \(p<.005\)

Full size image

4 Discussion and Future Directions

We have demonstrated the reliability and validity of the Polish version of the CSSQ. Both long and short versions are characterized by high internal consistency. At the same time, there are no items to be removed to significantly improve consistency. Unfortunately, we also observed a strong skewness of the results. This may be due to the fact that potential participants were previously informed about possible contraindications for VR studies and resigned from participation at the recruitment stage. Unfortunately, no satisfactory solution to this limitation is available since a different approach would be against the rules of ethics. CSSQ-PL scores match the levels of actual symptoms of cybersickness. This makes the questionnaire a good tool for screening participants in VR research and applications, including gaming, physiotherapy, or education, and protects potential participants from unpleasant experiences. In future research, it is necessary to 1) recruit a much larger sample in order to at least partially reduce the impact of skewness of the distribution on the results and 2) select a longer and more diverse VR material in order to better verify the accuracy of the predictions.