Statistics Glossary → Item difficulty, discrimination and guessing (1PL, 2PL, 3PL)
Item difficulty, discrimination and guessing (1PL, 2PL, 3PL)concept
In item response theory, the parameters that determine the probability that people of different ability levels answer an item correctly.
In the three-parameter logistic model, the probability that a person of ability θ answers an item correctly is P(θ) = c + (1 - c) / (1 + exp(-a (θ - b))). b is the difficulty parameter: with c = 0 it is the ability level at which the probability of a correct answer is 50%, and the item gets harder as b grows. a is the discrimination parameter: it gives the steepness of the item characteristic curve at b; a high a means the item sharply separates ability levels around b. c is the guessing parameter: the probability that even a person of very low ability answers correctly, which matters for multiple-choice items.
The Rasch / 1PL model treats all items as equally discriminating and estimates only their difficulties; this makes the total score a sufficient statistic for ability. The 2PL lets discrimination vary from item to item, and the 3PL adds the guessing parameter. More parameters require larger samples, and c in particular is estimated unstably. Whereas item difficulty in classical test theory (the proportion correct) depends on the sample, IRT parameters are regarded as sample-independent if the model holds.
An everyday example
Consider a hard item in a four-option exam. Half of the students slightly above average answer it correctly (high b); success changes sharply between students just below and just above that level (high a). A student who knows nothing can still answer correctly about one time in four by guessing (c ≈ 0.25).
In MerQur
The 'Model' option of Item Response Theory: Rasch/1PL, 2PL and 3PL for binary items; PCM, RSM, GRM and GPCM for polytomous items.
Analyses where this concept appears
Item Response Theory (IRT / Rasch)
Related concepts
Person ability (θ) estimation (EAP, MAP, ML, WLE)Item fit (infit, outfit, S-X²)