What question did this study set out to answer?

The paper aims to investigate how response group size and features of predictor variables influence model performance in accounting choice studies.

March 18, 2026The Accounting Review

Tradeoffs in the choice between logit and OLS for accounting choice studies.

View Full Paper

Ask AI

Bookmark

Authors

MSMary StoneUniversity of North Alabama JRJohn RaspStetson University

Discussion

Member takes

Implication

This analysis examines error rates and model sensitivity in accounting studies using logit and OLS for dichotomous choices, indicating model preferences.

Key Points

The paper aims to investigate how response group size and features of predictor variables influence model performance in accounting choice studies.
Conducted simulations and analyzed real accounting data.
Compared logit and OLS models in terms of error rates and sensitivity to predictor variables.
Assessed classification ability and calibration of both models under varying data conditions.
Logit test statistics show bias in small samples, primarily due to skewed predictor variables.
OLS may yield slightly better-calibrated test statistics but could present higher Type I error rates when classifying.
Logit is suggested as the preferable model for accounting choice studies, even with sample sizes around 50.

Cite This Study

Stone et al. (1991) studied this question.

synapsesocial.com/papers/69ba43884e9516ffd37a4d7d https://doi.org/10.2308/tar-9605080011

View Full Paper

Ask AI

Bookmark

Implication

This analysis examines error rates and model sensitivity in accounting studies using logit and OLS for dichotomous choices, indicating model preferences.

Tradeoffs in the choice between logit and OLS for accounting choice studies.

Discussion

Member takes

Implication

Key Points

Cite This Study

Tradeoffs in the choice between logit and OLS for accounting choice studies.

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider