Letter and Number Series: The ICAR Approach
Letter and Number Series items present a sequence of symbols obeying a hidden rule and ask the test-taker to identify the next term. They are the most common item type in numerical reasoning subtests and one of the four core families in the International Cognitive Ability Resource (ICAR) catalog.
Letter and Number Series items present a sequence of symbols obeying a hidden rule and ask the test-taker to identify the next term. They are the most common item type in numerical reasoning subtests and one of the four core families in the International Cognitive Ability Resource (ICAR) catalog, alongside matrix reasoning, verbal reasoning, and three-dimensional rotation.
The cognitive demand of series items decomposes into three sub-skills: recognizing the units (numbers, letters, or mixed symbols), generating candidate rules that link consecutive terms, and verifying each candidate rule against the entire sequence. Strong performers run these sub-skills in parallel: they read the sequence once while mentally testing several common rule families (constant difference, constant ratio, polynomial growth, alternating sub-sequences, primes, squares, Fibonacci-like sums) and converge on the answer when one candidate matches.
Difficulty in series items scales with rule complexity. Easy items use a single arithmetic operation (constant difference, constant ratio). Medium items use a polynomial or exponential growth pattern (squares, cubes, factorials). Hard items use compound rules (two interleaved sub-sequences, a non-monotonic pattern, or a rule based on a property of each term such as digit count or prime-factorization).
Cultural confounds are minimal in number-series items but present in letter-series items, where the test-taker must know the alphabet and its ordering. ICAR specifically uses the Roman alphabet, which is a cultural assumption that limits cross-population validity for some users. Number-series items are less culturally loaded but still depend on numeracy education.
Series items show among the highest g-loadings of any verbal-numerical item type — typically 0.7 to 0.8 in factor-analytic studies. They are also among the most coachable item types: a few hours of practice with a wide variety of series families produces measurable score gains on subsequent administration. This coachability is one reason responsible test designers rotate the underlying rule families across alternate forms.