Effects of Eye-Protection Devices on Eye Strain During Prolonged Gaming

Author

Callum Barnsley

Background

With the recent growth in computer use, the potential risks associated with prolonged screen exposure have also increased. These risks include eye strain, discomfort, and visual fatigue caused by screen light, particularly blue light (Sheppard & Wolffsohn, 2018). Eye strain, referred to as computer vision syndrome, has been documented to occur among people who engage in activities with extended screen time use (Rosenfield, 2016).

In response to these concerns, gaming eyewear is a product currently marketed to mitigate these effects, typically using blue-light filters to reduce visual strain. Standard gaming glasses generally provide a fixed tint on the lenses. While these do reduce the exposure to blue light, they do not dynamically adjust the variations in screen brightness nor the chromatic intensity. In contrast, adaptive blue-light filtering devices dynamically adjust the filtering based on on-screen chroma and brightness. This has the potential to provide more protection during extended gaming sessions. 

Despite the growing commercial adoption, understanding how adaptive devices compare to both standard gaming glasses and no protection, and whether any benefits persist over multiple gaming sessions, it is essential to do a statistical analysis to evaluate their effectiveness.

This study aims to test three hypotheses:

1: Device Effect Within Session

To examine whether the type of device influences eye strain score within a session:

H₀₁ (Null Hypothesis):

There is no effect of the device on the mean eye strain score. The mean eye strain does not change between the adaptive blue filter, standard gaming glasses and no protection within a session.

H0:μadaptive = μstandard = μnone

H₁₁ (Alternative Hypothesis):

There is an effect of the device on the mean eye strain score; the adaptive blue filter results in a lower mean eye strain score than both standard gaming glasses and no protection within a session.H11:μadaptive < μstandard and μadaptive < μnone

2: Persistence (Session Effect for Adaptive Device)

To examine whether the adaptive device maintains or improves its benefit across sessions:

H₀₂ (Null Hypothesis):
Participants used the adaptive blue filter; the mean eye score in session 2 is greater than or equal to the mean eye strain score in session 1.

H0:μadaptive,session2 ≥ μadaptive,session1

H₁₂ (Alternative Hypothesis):
Participants using the adaptive blue filter, the mean eye strain score in session 2 is less than the mean eye strain score in session 1.

H1:μadaptive,session2 < μadaptive,session1

3: Baseline Benefit of Standard Gaming Glasses

To examine whether standard gaming glasses reduce eye strain compared to no protection:

H₀₃ (Null Hypothesis):

There is no difference in the mean for the eye strain score between standard gaming glasses and no protection.H0:μstandard = μnone

H₁₃ (Alternative Hypothesis):
The mean for the eye strain score for standard gaming glasses is less than the mean eye strain score for no protection.The mean for the eye strain score for standard gaming glasses is less than the mean eye strain score for no protection.

H1:μstandard < μnone

Methods and Results

Method

This study uses a repeated-measure approach to examine the effects of gaming eyewear during extended gaming sessions. The devices that participants used are: adaptive blue-filter, standard gaming glasses and no protection, and sessions were 5-hour gaming sessions. Each device was used over two sessions, where the eye strain of the corresponding participant was documented and reported for each session.

Participants were randomly assigned to a device condition at the first session, and this assignment was held constant for the second session. Because each participant was measured twice under the same treatment condition, observations were not independent. Therefore, gamer_id was included as a blocking factor using an Error term in the repeated-measures ANOVA to account for within-subject correlation.

A post-hoc analysis had to be performed as the third hypothesis was unable to be supported or disproved. To do this, a package called emmeans was used to do the post-hoc comparisons, using estimated marginal means with Bonferroni correction to control multiple comparisons.

Data Description

The dataset that I was provided with consisted of repeated measures of eye strain collected from the participant after they had completed one 5-hour gaming session using a randomly selected device and again after another 5-hour gaming session.

Each participant was randomly assigned one of three devices:

  1. Adaptive blue filter

  2. Standard gaming glasses

  3. No protection

The dataset was structured in tidy format with one row per observation.

Because each participant was measured twice under the same conditions, observations were not independent. Therefore, analysis required a repeated- measure approach that accounts for within-subject correlation.

Statistical Approach

All analyses were done using the base version of R in R Studio. A two-factor mixed repeated-measures ANOVA was done. The device was considered a between- subject factor, and the session was a within-subject factor. The ANOVA model includes an Error term to account for repeated measurements within participants: 

eye_strain_score ~ device * session + Error(gamer_id/session).

I chose this approach over a standard two-way ANOVA because a simple ANOVA would violate the independence assumption. Including  gamer_id as a blocking (grouping) factor.

The Statistical Significance was evaluated at 0.05.

#install.packages("emmeans")
library(emmeans)
Welcome to emmeans.
Caution: You lose important information if you filter this package's results.
See '? untidy'
## Read dataset
eye_data <- read.csv("eye_strain_data.csv")

## Inspect structure
head(eye_data)
  X session gamer_id               device eye_strain_score
1 1    2016        1 adaptive_blue_filter             79.2
2 2    2016        2 adaptive_blue_filter             83.7
3 3    2016        3 adaptive_blue_filter             83.8
4 4    2016        4 adaptive_blue_filter             90.8
5 5    2016        5 adaptive_blue_filter             83.5
6 6    2016        1 adaptive_blue_filter             94.5
str(eye_data)
'data.frame':   600 obs. of  5 variables:
 $ X               : int  1 2 3 4 5 6 7 8 9 10 ...
 $ session         : int  2016 2016 2016 2016 2016 2016 2016 2016 2016 2016 ...
 $ gamer_id        : int  1 2 3 4 5 1 2 3 4 5 ...
 $ device          : chr  "adaptive_blue_filter" "adaptive_blue_filter" "adaptive_blue_filter" "adaptive_blue_filter" ...
 $ eye_strain_score: num  79.2 83.7 83.8 90.8 83.5 94.5 95 96.8 95.7 96.3 ...
## Convert categorical variables to factors
eye_data$device <- factor(eye_data$device)
eye_data$session <- factor(eye_data$session)
eye_data$gamer_id <- factor(eye_data$gamer_id)

## Fit repeated-measures ANOVA
anova_model <- aov(
  eye_strain_score ~ device * session + Error(gamer_id/session),
  data = eye_data
)

## Display ANOVA summary
summary(anova_model)

Error: gamer_id
          Df Sum Sq Mean Sq F value Pr(>F)
device     1   76.4   76.42   0.273  0.637
Residuals  3  838.4  279.48               

Error: gamer_id:session
          Df Sum Sq Mean Sq F value Pr(>F)   
device     1    913     913   7.712 0.0692 . 
session    1   4774    4774  40.336 0.0079 **
Residuals  3    355     118                  
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Error: Within
                Df Sum Sq Mean Sq F value   Pr(>F)    
device           2   1708   853.8  26.364 1.09e-11 ***
device:session   2     63    31.3   0.965    0.382    
Residuals      585  18946    32.4                     
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Create summary table
summary_table <- aggregate(
  eye_strain_score ~ device + session,
  data = eye_data,
  FUN = function(x) c(mean = mean(x), sd = sd(x))
)

## Expand results
summary_table <- do.call(data.frame, summary_table)

## Rename columns
colnames(summary_table) <- c("Device", "Session", "Mean", "SD")

## Print
summary_table
                   Device Session     Mean       SD
1    adaptive_blue_filter    2016 82.29400 8.411726
2           no_protection    2016 85.00100 7.244650
3 standard_gaming_glasses    2016 85.87475 5.789440
4    adaptive_blue_filter    2017 87.86600 4.489724
5           no_protection    2017 92.05100 4.030853
6 standard_gaming_glasses    2017 91.69400 3.397766
##post-hoc 
model <- aov(eye_strain_score ~ device * session, data = eye_data)

## Get estimated marginal means by device within each session (post-hoc)
emm_device_session <- emmeans(model, pairwise ~ device | session, adjust = "bonferroni")

## Extract only the emmeans (ignores contrasts for now)
emmeans_table <- as.data.frame(emm_device_session$emmeans)

## Round numeric columns for reporting
emmeans_table$emmean <- round(emmeans_table$emmean, 1)
emmeans_table$SE <- round(emmeans_table$SE, 2)
emmeans_table$lower.CL <- round(emmeans_table$lower.CL, 2)
emmeans_table$upper.CL <- round(emmeans_table$upper.CL, 2)

# Rename columns to match your example
colnames(emmeans_table) <- c("Device", "Session", "Mean", "SE", "df", "LowerCI", "UpperCI")

# Keep only the columns you want for your table (Device, Session, Mean, SE)
emmeans_table <- emmeans_table[, c("Device", "Session", "Mean", "SE")]

# Print
print(emmeans_table)
                   Device Session Mean   SE
1    adaptive_blue_filter    2016 82.3 0.58
2           no_protection    2016 85.0 0.58
3 standard_gaming_glasses    2016 85.9 0.59
4    adaptive_blue_filter    2017 87.9 0.58
5           no_protection    2017 92.1 0.58
6 standard_gaming_glasses    2017 91.7 0.58
Device Session Mean SE
adaptive_blue_filter 2016 82.3 0.58
no_protection 2016 85 0.58
standard_gaming_glasses 2016 85.9 0.59
adaptive_blue_filter 2017 87.9 0.58
no_protection 2017 92.1 0.58
standard_gaming_glasses 2017 91.7 0.58

Results

A two-factor mixed repeated-measures ANOVA was conducted to examine the effects of the eye protection device and session on eye strain. The analysis revealed a significant main effect of device, F(2, 585) = 26.36, p < .001, indicating that mean eye strain differed across the three types of eye protection. There was also a significant main effect of session, F(1, 3) = 40.34, p = .0079, suggesting that eye strain increased from Session 1 to Session 2. The device × session interaction was not significant, F(2, 585) = 0.97, p = .382, indicating that the increase in eye strain over time was similar across all devices.

Post-hoc comparisons using estimated marginal means (emmeans) with Bonferroni correction indicated that, in both sessions, the adaptive blue filter resulted in significantly lower eye strain than both no protection (2016: M = 82.3, SE = 0.58 vs 85.0, SE = 0.58; 2017: M = 87.9, SE = 0.58 vs 92.1, SE = 0.58) and standard gaming glasses (2016: M = 82.3, SE = 0.58 vs 85.9, SE = 0.59; 2017: M = 87.9, SE = 0.58 vs 91.7, SE = 0.58). There was no significant difference between standard gaming glasses and no protection in either session. Comparisons across sessions for each device indicated that eye strain increased significantly from 2016 to 2017 for all devices: adaptive blue filter (M difference = 5.57, SE = 0.58), standard gaming glasses (M difference = 5.82, SE = 0.59), and no protection (M difference = 7.05, SE = 0.58).

Overall, the adaptive blue filter consistently produces results showing lower eye strain; however, all types of eye protection have similar increases in eye strain over time.

Interpretation Relative to Hypotheses

  • H₁₁: Supported. Device type significantly influenced eye strain.

  • H₁₂: Not supported. No evidence of a persistence or carry-over advantage for the adaptive device.

  • H₁₃: Not supported. Standard gaming glasses did not reduce eye strain by a significant amount when compared to no protection.

Conclusion

In conclusion, the adaptive blue light filtering device significantly lowers participants’ eye strain during extended gaming sessions compared to both standard gaming glasses and no protection. The eye strain increased over the duration of session 2 for all participants, and the adaptive blue light filtering device did not maintain a stronger protection over time. Overall, the adaptive blue light filtering device provides an immediate benefit to the user, but there is no evidence to suggest that the effectiveness improves or persists over multiple sessions.

References

I used AI to help me understand the analysis and to help me clarify my understanding.

Pardhan, S. et al. (2022) Risks of Digital Screen Time and Recommendations for Mitigating Adverse Outcomes in Children and Adolescents. The Journal of school health. [Online] 92 (8), 765–773.

(Accessed: 24/02/2026)

Rosenfield, M., 2011. Computer vision syndrome: a review of ocular causes and potential treatments. Ophthalmic and Physiological Optics, 31(5), pp.502-515.

(Accessed : 15/02/2026)

Sheppard, A.L. and Wolffsohn, J.S., 2018. Digital eye strain: prevalence, measurement and amelioration. BMJ open ophthalmology, 3(1).

(Accessed: 13/02/2026)