Udo Undeutsch was a German psychologist asked, in the 1950s, to help courts assess the testimony of children in sexual abuse cases — a setting where there is frequently no physical evidence, no witness, and nothing to evaluate but the account itself.
His hypothesis was modest and testable: a statement based on a real experience differs in content and quality from one that is invented. Not in the speaker's demeanor. In the text.
Max Steller and Günter Köhnken operationalized this as Criteria-Based Content Analysis, which sits inside a wider procedure called Statement Validity Assessment. CBCA scores nineteen criteria, and the reasoning behind them is what makes the method interesting.
They include logical structure and quantity of detail; unstructured production, meaning the account is not delivered in tidy chronological order; contextual embedding, placing the event in a specific time and place within the routine of a life; descriptions of interactions; reproduction of speech; unexpected complications; unusual and superfluous details; accurately reported details the witness did not understand at the time; spontaneous corrections; admitted lack of memory; raising doubts about one's own testimony; self-deprecation; and pardoning the accused.
The last several are the diagnostic core, and they follow from a single insight: a fabricator is managing an impression, and certain features are contrary to impression management. Someone constructing an accusation does not spontaneously say I might be wrong about that part, or I did not see what happened next, or he was usually kind to me. Those features damage the account, which is precisely why they are evidence.
Similarly, unstructured production and superfluous detail run against how invention works. Invented narratives are ordered and economical, because the inventor supplies what is needed. Real memories are cluttered and arrive out of sequence.
Crucially, CBCA is not scored in isolation. Statement Validity Assessment requires the CBCA analysis to be evaluated against a validity checklist covering the witness's cognitive capacity, the interview's suggestiveness, the possibility of contamination or coaching, and any motive to fabricate. A high CBCA score from an interview full of leading questions means nothing.
Meta-analytic reviews find CBCA discriminates above chance, commonly with accuracy in the high sixties to low seventies. That is meaningfully better than human judgment and meaningfully short of forensic reliability. Performance degrades with younger children, with coaching, and with interview contamination.
CBCA is used in German and several other European courts as expert evidence. It is not admitted in most Anglo-American jurisdictions.
The honest summary is that this is the most empirically grounded content-based method available, which is a statement about the alternatives as much as about CBCA. Compared with the previous chapter's method it is a serious instrument; compared with a fingerprint it is not evidence, and the courts that use it treat it as one input among several, which is the only defensible way to use anything in this part of the guide.
Cross-ref: SCAN critique chapter above