Which keystroke features correlate with writing grade changes across stages of the writing process: pause-variance and focus-shift features matter early, sentence-transition and event features later¶

Claim · All claims
Evidence · 1 study · associational r1 · q2 · i2 medium
Currency · newest study 2022 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i2 SD of IKI (r = 0.24) and focus shifts to translation (r = 0.19) and task (r = 0.22) related to final grade only in the beginning of writing, while sentence-transition, pause, and cut/paste/jump features related only in the middle or end. → Rianne Conijn 2022

Evidence¶

Rianne Conijn 2022¶

Rianne Conijn, Christine Cook, Menno van Zaanen, Luuk Van Waes. (2022). Early prediction of writing quality using keystroke logging. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-021-00268-w

q2 · i2 · associational · r1

Stage-wise correlational analyses (beginning 0–10, middle 10–20, end 20–30 min) on the same 126-student summary-writing dataset. Early-stage and later-stage features form disjoint significant sets, showing "the relation of the features with final grade differs over time".

"The SD of the IKI ( r = 0.24) and the number of focus shifts to translation ( r = 0.19) and task ( r = 0.22) were significantly related to final grade only in the beginning of the writing process. In contrast, the SD of the time between sentences ( r = -0.20), percentage of long pauses between words ( r = 0.20 and r = 0.23), number of production cycles ( r = 0.18), percentage of linear transitions between sentences ( r = -0.18 and r = -0.24), and the SD of the time in cut, paste, and jump events ( r = 0.18) were only significantly related to final grade in the middle or end of the writing process."

Discussion¶

Related Claims¶

• Important keystroke features for classifying at-risk writers differ by stage: early models emphasize focus shifts and initial pause, later models emphasize inter-keystroke-interval variance — related
• Over the full writing process, only one of 54 keystroke features (percentage of long pauses between words) correlates significantly with final writing grade in ESL academic summaries — related
• Keystroke-feature regression models fail to predict final writing grade at any point during the writing process — reports the opposite
• The relationship between keystroke data and writing quality may be less clear than previously posited — related
• The proposed model's effectiveness was demonstrated by applying it to actual keystroke log datasets — related