Certainty Based Assessment - CBA
US-CBA for Browsers
Certainty-Based assessment

London Approved Protocol for Teaching - LAPT

Demo Modules |  US-CBA Home |  UK-CBA Home |  TCFIR Home |  Publications |   Review |  History

LAPT Confidence-Based, or Certainty-Based Marking (CBM) 

Certainty-based marking (in various forms) is many decades old. It has repeatedly been shown to stimulate
more careful thinking and learning than simple (right/wrong) marking, and to provide more reliable
assessments. LAPT aims to provide the best possible implementation of CBM as a learning tool, for open use.


What is Certainty-Based Marking (CBM)? 

  • After each answer, you indicate your degree of certainty that your answer will be marked as correct.
  • In LAPT, this is on a 3-point scale: 1 (low), 2 (mid) or 3 (high)
  • We deliberately do not use words like 'sure' or 'very sure' because these mean different things to different people
  • The mark scheme and the risk of penalties will guide your choice of C level: see below.
Why use CBM? 
  • To make you think about how reliable your answer is
  • To encourage you to try to understand the issues, not just to react immediately to a question
  • To encourage you to think laterally: other pieces of knowledge may help to validate or question your answer
  • To challenge you - if you won't risk losing marks if wrong, then you don't really know the answer.
  • If you are a careful thinker, but not very confident, you will gain in confidence
  • It is more fair - a thoughtful and confident correct answer deserves more marks than a lucky hunch
  • You need to pay attention if you make confident wrong answers - to think, reflect and learn!
  • Efficient study requires that you constantly question how your ideas arise, and how reliable they are
CBM: teachers How Does It Work

CBM marks a student according to confidence (or 'degree of certainty') in each answer.

Confidence level :

C=1
(low)

C=2
(mid)

C=3
(high)

No Reply

Mark if correct :

1

2

3

(0)

Penalty if wrong :

0

- 2

- 6

(0)

 Such a scheme rewards students who reflect to the point that they can either :-

   (a) justify high confidence so that they are prepared to risk a penalty if wrong, or

   (b) identify reasons for reservation so that they lower their confidence and eliminate risk.

Both ways they gain by thinking more deeply and by correctly judging reliability. A student who distinguishes reliable from uncertain areas of knowledge does better than one with the same number of correct answers who cannot judge this correctly.

CBM discriminates more significant levels of knowledge than does mere correctness. Roughly :

   3 = knowledge    2 = uncertainty    1,0 = ignorance    -2 = misconception    -6 = delusion! or

   3,2 = usable knowledge    0,1 = unusable knowledge    -2,-6 = dangerous knowledge

In assessment, CBM weights uncertain answers less than confident ones. This reduces the variance due to 'guessing' and increases the statistical reliability of exam results, as shown in several research studies and borne out in analysis of exam data at UCL. CBM scores have been found to be the best predictors even of the % correct on a separate set of questions.

Practical Issues

·         CBM can be used wherever objective testing (right/wrong answers) is used (e.g.  T/F ,  MCQ ,  EMQs ,  numerical ,  word/phrase answers)

·         It does not require a special style of question setting.

·         Objective Testing is not just about factual knowledge.

·         Exercises can be set up with WebCT, with authenticated student grades returned to the VLE.

·         Question files are easily adapted from any systematic text/graphic/HTML format using the exercise specification manual, and can be drafted in WORD or (for easy application of flexible arrangements for different formats, conditional explanations, etc.) with a dedicated authoring tool on the website.

·         Exercises (for formative use without recording of data) can be run from a CD-ROM or local drive.

·         For summative exams, till your experience may inform new strategies, provisional passmarks can always be set by reference to the number passing on conventional criteria (% correct answers).

·         Help, advice, troubleshooting and adaptations are all freely available from UCL (cusplap@ucl.ac.uk).

Tony Gardner-Medwin  --  Physiology, UCL  --   (ucgbarg@ucl.ac.uk) or Reid Cornwell, Director TCFIR wrc@tcfir.org
Exercises on US-CBA