The "t" time: Investigating handedness through strokes and slopes

Area of Science:

• Forensic Science
• Biometrics
• Graphology

Background:

• Handwriting analysis is crucial in forensic investigations.
• Differentiating between left-handed and right-handed writing can aid authorship attribution.
• Existing methods for handedness determination from handwriting are limited.

Purpose of the Study:

• To investigate and compare stroke and slope characteristics in left-handed and right-handed handwriting.
• To develop and validate a model for classifying handedness based on handwriting features.
• To assess the model's accuracy in identifying the author's handedness in disputed documents.

Main Methods:

• Analysis of stroke and slope directions in specific letters (t, f, đ, H, l, d, g) from in-house and public handwriting samples.
• Development of a logistic regression model using averaged strokes of the letter 't' to classify handedness.
• Validation of the model on independent handwriting datasets.

Main Results:

• Statistically significant differences in stroke directions were observed between left-handed and right-handed writers (p ≤ 0.05).
• Right-handers predominantly showed left-to-right strokes (98%-100%), while left-handers exhibited more variability.
• The developed model achieved 100% accuracy in identifying left-handed writing and 73%-97% accuracy for right-handed writing.

Conclusions:

• Handwriting stroke characteristics, particularly of the letter 't', can reliably differentiate between left-handed and right-handed authors.
• The developed logistic regression model shows high potential for forensic applications in authorship determination.
• This method can effectively exclude or narrow down potential authors in disputed handwriting cases.