Longitudinal evaluation of video assistant referee implementation on goal timing distributions and card statistics in Turkish Super League

The Video Assistant Referee (VAR) system has become an integral component of modern football, yet its long-term influence on disciplinary outcomes and goal-scoring patterns remains debated. Therefore, this study examined the association between VAR implementation and changes in disciplinary rates and goal-scoring dynamics in the Turkish Super League across twelve seasons (2012–2013 to 2023–2024). Match information from all Turkish Super League games played during this period was aggregated at the season level, yielding six pre-VAR and six post-VAR observations.

Data were analyzed using generalized linear models with a negative binomial distribution. To account for unequal seasonal match volumes, all count-based models incorporated log-transformed seasonal match totals as exposure offsets. A quasi-binomial regression model was additionally employed to examine the occurrence of goals during added-time periods.

Sensitivity analyses were conducted to evaluate the consistency of the findings under alternative historical specifications. The results revealed significant reductions in yellow cards (IRR = 0.87, p < .001), total cards (IRR = 0.88, p < .001), and card points (IRR = 0.88, p < .001) following VAR implementation. In contrast, direct red cards and second yellow-to-red cards showed no statistically significant differences between periods.

While overall goal productivity remained unchanged (IRR = 0.97, p = .555), a greater proportion of goals occurred during added-time periods in the post-VAR era. Significant increases were observed in first-half added-time goals (IRR = 2.92, p < .001), second-half added-time goals (IRR = 1.41, p = .003), and total added-time goals (IRR = 1.71, p < .001). Quasi-binomial regression analyses further indicated that the association between VAR period and the conditional probability of an observed goal occurring during added time differed by match half, with higher odds in the post-VAR period during both the first half (OR = 3.15, 95% CI [2.03, 4.91], p < .001) and second half (OR = 1.56, 95% CI [1.21, 2.02], p = .001).

Sensitivity analyses demonstrated consistent directional associations across multiple model specifications. Overall, VAR implementation was associated with lower rates of caution-related disciplinary actions and with changes in the temporal distribution of goals, while overall goal productivity remained stable. The findings indicate that a greater proportion of goals occurred during added-time periods following VAR implementation.

However, because actual added-time durations were unavailable, these results should be interpreted as a redistribution of goal occurrence rather than definitive evidence of increased scoring intensity during stoppage time. We would like to express our gratitude to the transfermarkt.com for providing the permission to use the data. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.