Analysis of Scrabble scores following each Official Word List (OWL) update.

Scope

By 2017, there were three (3) updates to Scrabble's Official Word List (OWL), with the first being released in 2006. I hypothesized that a more expanded list would reap better performance for players, given they had more words to play. To test this hypothesis, I analysed data from Scrabble games held from 1973 to 2017.

Analysis

For this analysis, I created a time-series graph of average weekly scores by player rating - noting that players with ratings over 500 are more experienced than those with ratings under 500 - as determined by the official rules for calculating player rating and ranking by state. The lowest rating at the time was 535, with higher ratings up to 2014 points (NASPA, 2022). I later decided to split the data further into winning and losing scores - to get a sense of overall change without losing information on how different players performed. I additionally marked the first word list update as well as two others up to the year of the last data point 2017.

Results

Average Winning Scores per Week by Player Rating

Average Losing Scores per Week by Player Rating

Conclusion

The results show a general increase in scores across levels following each OWL release. Players with higher ratings had a uniform increase, while less experienced players performed more sporadically, indicating that more experienced players probably benefited more from an expanded word list than those with less experience.

It would be interesting to see an analysis of the impact of word difficulty on player scores, but it appears that there isn't a standardized way to assess this. Since current methods of evaluating word difficulty are beyond the scope of this analysis, I will attempt this at a later time.

References

1. Curto, P., Mamede, N., & Baptista, J. (2015). Automatic Text Difficulty Classifier—Assisting the Selection Of Adequate Reading Materials For European Portuguese Teaching: Proceedings of the 7th International Conference on Computer Supported Education, 36–44. https://doi.org/10.5220/0005428300360044
2. Data/scrabble_games.csv at master · fivethirtyeight/data. (2017, April 18). GitHub. Retrieved February 11, 2022, from https://github.com/fivethirtyeight/data
3. NASPA: Top Ratings by State and Province. (n.d.). Retrieved February 27, 2022, from http://scrabbleplayers.org/ratings/bystate.html
4. NASPA Zyzzyva Download—NASPAWiki. (n.d.). Retrieved February 11, 2022, from http://www.scrabbleplayers.org/w/NASPA_Zyzzyva_Download
5. Official Tournament and Club Word List—NASPAWiki. (n.d.). Retrieved February 11, 2022, from http://scrabbleplayers.org/w/Official_Tournament_and_Club_Word_List
6. Rating system overview—NASPAWiki. (n.d.). Retrieved February 27, 2022, from http://scrabbleplayers.org/w/Rating_system_overview
7. Roeder, O. (2017, April 19). How ‘Qi’ And ‘Za’ Changed Scrabble. FiveThirtyEight. https://fivethirtyeight.com/features/how-qi-and-za-changed-scrabble/
8. Zhang, S., Jia, Q., Shen, L., & Zhao, Y. (2020). Automatic Classification and Comparison of Words by Difficulty. In H. Yang, K. Pasupa, A. C.-S. Leung, J. T. Kwok, J. H. Chan, & I. King (Eds.), Neural Information Processing (pp. 635–642). Springer International Publishing. https://doi.org/10.1007/978-3-030-63820-7_72