IMC 2003 statistics — difficulty & discrimination

Studolymp / IMC / 2003

IMC 2003

contestants

185

problems

12 (6+6)

score scale

0–20

per-problem results

yes

Notes on this edition’s data

Scored on a 20-point scale this era (8/16 = solved).

Per-problem difficulty

Day 1

Day 2

Difficulty vs discrimination

Each dot is one problem. X — how many solved it (scored ≥ 8 of 10); Y — discrimination, i.e. how well the problem told strong contestants from weak ones (score-vs-total correlation).

A good problem sits high and away from the edges. Bottom-left dots separated nobody — almost everyone failed them the same way.

Problem statistics

Same data as the Day 1 / Day 2 charts above.

	#	difficulty	avg	solved	near-0	disc	topics
Day 1
	P1	easy	8.8	83%	5%	0.51	sequences & series
	P2	medium	6.1	38%	13%	0.54	abstract algebra
	P3	medium	5.8	45%	25%	0.57	linear algebra
	P4	easy	7.6	69%	14%	0.52	number theory
	P5	hard	3.7	29%	46%	0.54	real analysis integration
	P6	very hard	1.6	8%	74%	0.44	polynomials inequalities
Day 2
	P7	easy	7.9	76%	12%	0.44	linear algebra
	P8	easy	6.7	56%	18%	0.57	integration real analysis
	P9	easy	6.0	56%	29%	0.54	topology & metric spaces
	P10	very hard	2.2	12%	67%	0.46	combinatorics
	P11	hard	2.9	18%	42%	0.50	set theory & logic real analysis
	P12	very hard	1.5	12%	84%	0.38	sequences & series

Field cohort = total ≥ 6 points; difficulty is computed live by the API. Discrimination is the corrected item–total correlation.