Validity, Reliability, IOA, Procedural Integrity, and Dosage

Key Takeaways

  • Validity asks whether the data measure the intended behavior and dimension.
  • Reliability concerns consistency of measurement; IOA estimates agreement between observers.
  • High IOA does not prove data are valid if observers share the same flawed definition.
  • Procedural integrity and dosage data help interpret whether implementation matched the plan.
Last updated: May 2026

Data Quality and Implementation Data

Validity asks whether measurement captures the intended behavior and dimension. Reliability asks whether measurement is consistent. IOA is a formal check of observer agreement, but it is not proof that observers measured the right thing.

Quality Decision Chain

Problem patternFirst check
Observers disagreeDefinition, training, scoring rules, IOA method
Data do not match the clinical questionValidity of the measure and dimension
Intervention data are flatProcedural integrity and dosage before rejecting the plan
Change appears only on some daysSetting events, schedule, observer coverage, representativeness

Procedural integrity measures whether the plan was implemented as written. Dosage measures the amount of exposure, such as minutes, sessions, opportunities, or trials.

If outcome data show no improvement but integrity is low, the most defensible decision is usually to improve implementation and collect more data. If integrity and dosage are adequate, then modification may be warranted.

Exam Tie-In

Do not let IOA distract from validity. Observers can agree perfectly on the wrong response class or wrong dimension. On scenario items, check whether the data answer the referral question, whether the plan was implemented, and whether the client had enough exposure before changing treatment.

Test Your Knowledge

Two observers have 96% IOA for disruption, but the definition includes crying, refusal, aggression, and leaving the area as one broad category. The team cannot tell which response changed. What is the main concern?

A
B
C
D
Test Your Knowledge

A graph shows no improvement after an intervention begins. Procedural integrity data show staff implemented only 52% of required steps. What should the BCBA do first?

A
B
C
D
Test Your Knowledge

A BCBA tracks the number of teaching opportunities delivered per session in addition to percent correct. What implementation variable is being measured?

A
B
C
D