Abstract / Description of output
Introduction
Non-cognitive traits should be considered when selecting candidates to study medicine. However, evaluating these traits remains difficult. We explored whether measuring undesirable non-cognitive behaviour (‘Red Flags’) added value to a medical school admissions system. Red Flags included rudeness, ignoring the contributions of others, disrespectful behaviour, or poor communication.
Methods
Following an admissions interview testing non-cognitive attributes in 648 applicants to a UK medical school, we measured the association between interview score and Red Flag frequency. We tested linear and polynomial regression models to evaluate whether the association was linear or non-linear.
Results
In total, 1126 Red Flags were observed. While Red Flags were concentrated among low-scorers, candidates in the highest- and second-highest deciles for interview score still received Red Flags (six and twenty-two, respectively). The polynomial regression model indicated candidates with higher scores received fewer Red Flags, but the association was not linear (F(3644) = 159.8, p = .001, adjusted R2 = 0.42).
Conclusions
The non-linear association between interview score and Red Flag frequency shows some candidates with desirable non-cognitive attributes will still display undesirable–or even exclusionary–non-cognitive attributes. Recording Red Flag behaviour reduces the likelihood such candidates will be offered a place at medical school.
Non-cognitive traits should be considered when selecting candidates to study medicine. However, evaluating these traits remains difficult. We explored whether measuring undesirable non-cognitive behaviour (‘Red Flags’) added value to a medical school admissions system. Red Flags included rudeness, ignoring the contributions of others, disrespectful behaviour, or poor communication.
Methods
Following an admissions interview testing non-cognitive attributes in 648 applicants to a UK medical school, we measured the association between interview score and Red Flag frequency. We tested linear and polynomial regression models to evaluate whether the association was linear or non-linear.
Results
In total, 1126 Red Flags were observed. While Red Flags were concentrated among low-scorers, candidates in the highest- and second-highest deciles for interview score still received Red Flags (six and twenty-two, respectively). The polynomial regression model indicated candidates with higher scores received fewer Red Flags, but the association was not linear (F(3644) = 159.8, p = .001, adjusted R2 = 0.42).
Conclusions
The non-linear association between interview score and Red Flag frequency shows some candidates with desirable non-cognitive attributes will still display undesirable–or even exclusionary–non-cognitive attributes. Recording Red Flag behaviour reduces the likelihood such candidates will be offered a place at medical school.
Original language | English |
---|---|
Journal | Medical Teacher |
Early online date | 7 Jul 2023 |
DOIs | |
Publication status | E-pub ahead of print - 7 Jul 2023 |