Norm Development and Standard-Score Transformation

Description

Standardise an instrument on a representative sample and convert raw scores to z, T, stanine or percentile scores so individual results can be interpreted against a norm group.

Raw scores cannot be compared across subtests, instruments or people without a reference. Norm development administers the final instrument to a large, representative standardisation sample, computes its mean and SD, and provides derived scores: z = (X - M)/SD; T = 10z + 50 (McCall's T scale); C scores and stanines (nine bands); and percentile, age or grade norm tables. Individual or product scores are then read against the norm group, for example as 'at the 70th percentile'. As Mehrens and Lehmann stress, norms describe actual performance, not standards of what is good, and they are only as meaningful as the norm group is relevant: a score's percentile changes with the reference population. Norms need periodic renewal as populations and products change.

Attributes

Loss type
Siloed
Effort
Weeks

Steps

  1. Define the population the norms should represent and draw a large representative standardisation sample.
  2. Administer the final instrument under standard conditions.
  3. Compute the mean and SD, and check the score distribution.
  4. Convert raw scores to z and T scores (or stanines and C scores).
  5. Build percentile tables, and age, segment or product-category norms where relevant.
  6. Publish the norms with a description of the norm group, and interpret individual scores against them, refreshing norms periodically.

Sources

  • Research Methodology in Behavioural Sciences (Mangal & Mangal, 2013)

Also known as

  • Test norms
  • Standard scores
  • z scores
  • T scores
  • Stanines
  • C scores
  • Percentile norms
  • Age and grade norms

Categories

  • Analytics & Behaviour
  • Survey & Psychometrics