Longitudinal Scalogram Analysis

Description

Tests whether behaviours or abilities are acquired in a consistent cumulative order over time, revealing natural adoption sequences.

Scalogram analysis, introduced by Louis Guttman, tests whether a set of items forms a cumulative scale: if a person endorses or achieves a harder item, they also endorse or achieve all easier ones. Longitudinal scalogram analysis applies this logic over time to test whether people acquire behaviours or capabilities in a consistent order, such as developmental stages in children. In UX this suits feature adoption: do users typically adopt basic features before advanced ones, and in what order? A consistent sequence suggests a natural learning path that onboarding and progressive disclosure can follow. Rigour comes from clear item definitions, adequate observation windows, coefficients of reproducibility (traditionally 0.90 or more) and scalability, probabilistic extensions such as Mokken scaling to handle error, and checking whether sequences differ across segments.

Attributes

Loss type
Latent
Effort
Weeks

Steps

  1. Define the set of behaviours or features whose acquisition order is of interest.
  2. Define criteria for 'acquired' (for example used at least three times within 30 days).
  3. Extract per-user acquisition status at regular time points from logs or repeated surveys.
  4. Order items by difficulty (proportion acquired) and test cumulative patterns with coefficients of reproducibility and scalability.
  5. Apply probabilistic extensions such as Mokken scaling to tolerate error.
  6. Examine transitions over time and compare sequences across segments.
  7. Translate the sequence into onboarding and progressive disclosure design.

Sources

  • Longitudinal Scalogram Analysis
  • Measurement and Prediction (Stouffer et al.)

Also known as

  • Longitudinal Guttman scaling
  • Developmental scalogram analysis
  • Mokken scaling over time

Categories

  • Analytics & Behaviour
  • Survey & Psychometrics