The Role of Multiple Mini Interviews (MMI) in selection Dr Jon Dowell, Admissions Convenor, Dundee Medical School 1 Overview Rationale behind MMI for selection A brief history What do MMIs assess How do MMIs work Performance in UK context Feedback from applicants and assessors The future? Primus Medicine has ‘previous’ Alexander Monros 1720 - 1846 Secundus "I dislike him & his lectures so much that I cannot speak with decency about them. He is so dirty in person & actions." Tertious Rationale Personal qualities considered ‘important’ but not reliably assessed by interview and without much predictive validity. (Goho + Blackman 2006. cognitive r 0.06, clinical r 0.17) Introduced for medicine at McMaster, Canada (Reiter, Eva et al). Piloted 2002 – 12 x 8 min. Generalisability since +/- 0.7 Predicted OSCE performance ß 0.44 (interview ns) Clerkship ratings ß 0.57 (interview ns) MCC Part 1 (selected components) ß 0.3-0.4 (interview ns) Increasing detail emerging A cost efficiency comparison between the multiple mini-interview and traditional admissions interviews. Set up costs, running equates Rosenfeld JM, Reiter HI, Trinh K, Eva KW. Adv Health Sci Educ Theory Pract. 2008 Mar;13(1):43-58. The multiple mini-interview: how long is long enough? 10 – 8 – 5 mins Dodson M, et al. Med Educ. 2009 Feb;43(2):168-74 Factors affecting the utility of the multiple mini-interview in selecting candidates for graduate-entry medical school. 8Q reliability 0.7 Roberts et al. Med Educ. 2008 Apr;42(4):396-404 applicant 22% variance Should candidate scores be adjusted for interviewer stringency or leniency in the multiple mini-interview? Rasch Modeling Roberts C, Rothnie I, Zoanetti N, Crossley J. Med Educ. 2010 Jul;44(7):690-8 Increasingly widespread Countries • Canada, • Australia • UK • US Disciplines • UG Medicine • UG Veterinary • UG Dentistry MMI in Dundee 7 minute x 10 stations Reliability – 2008/9 n 450 – 2009/10 n 500 – 2010/11 n 600 0.7 0.69 0.69/.67 (0.88) – 2011/12 Dental school collaboration MMI in St Georges 5 minute stations x 8 Stations Muriel Shannon +Aileen O’Brien Reliability – 2009/10 – 2010/11 n 1078 n 1333 0.69 MMI in Belfast 5 minute x 9 stations Keith Steele Reliability – 2010/11 n 500 0.56 MMIs typically aim to assess – Interpersonal and communication skills (inc empathy) – Teamwork and leadership – Preparation and motivation – Critical thinking, problem solving – Ethical/moral reasoning – ‘Integrity’ How? Practicalities Capacity (n = 600) 20 applicants per ‘Run’ per session Two teams in parallel (80/day) 8 days total Staffing (2 academic leaders + 2 admin) For each run 5 faculty (interview) 5 student or Simulated Patients (interactive) 2 ‘actors’ (employed junior students) 3 ‘role players’ (students or SP) Preparation = big job Stations - Drafted and piloted with students • • • • Assessor instructions Candidate instructions Resources (script, shapes, materials, UCAS forms etc) Actor training Training at start of every session 30 min general 20 min station specific Station/Assessor/ Candidate Information Domain Scores Overall Judgement Red Flag & Narrative Performance of Dundee MMI Score Distribution Station Reliability 1 ItemTotal Correl 0.34 Alpha if Item Deleted 0.67 2 0.31 0.67 3 0.17 0.70 4 0.37 0.66 5 0.43 0.65 6 0.47 0.64 7 0.34 0.67 8 0.28 0.67 9 0.38 0.66 10 0.39 0.66 Overall Alpha 0.69 Domain Reliability Number of Stations Number of Scores Cronbach’s alpha Communication 9 12 .767 Critical Thinking 6 6 .453 Moral Reasoning 4 4 .170 Prep +Motivation 3 4 .515 Teamwork 2 3 .395 Integrity 1 1 - Gender difference - Yes Interactive One - one Females performed significantly better than males. 2009 Female average 109 /150 vs 105 p = < .01 Effect size: 0.14 Applicant Group difference? 2009 – 2010 2010 - 2011 Apologies but we are intending to eventually publish from this data so have removed it from circulated slides. Assessors Survey 2008-9 116 /201 (58%) assessors completed online survey. • 91% thought the process fair. • 88% thought applicant stress moderate or less. • All stations rated highly relevant, except for the ‘Question of Ethics’ where opinions divided. Candidate Survey 2008-9 • 324 /433 (75%) completed online survey. • 94% felt MMI was ‘fair’. • 90% felt it is a ‘valid way to assess candidates’. • 71% preferred MMI to traditional interview. • ‘Student Counselor’ (role play) most enjoyable We are happy but where next? Hawk-Dove Effect STAT I O N S A 1 2 3 4 5 6 B C D E P. H. Harasym, Ph.D. 7 8 9 10 Dundee 2011 Rasch vertical ruler MFRM creates a common scale for candidate ability, examiner stringency, item difficulty and ...... Comparing raw and ‘fair’ scores for 350 offers 34 (9.7%) different candidates selected. Conclusions and Next Steps MMI appear very promising: • Set up ‘costs’ considerable • Running costs comparable (if students used) • Student involvement positive • Applicants are positive • Predictive validity emerging and encouraging. Outstanding issues: • Rotating content. • Application of IRT (DIFF, Fair scores etc) • Oh yes – and the bloopers are great. Thank you Discussion?