Statement Validity Assessment - Northwestern University: Psychology

advertisement

Statement Validity Assessment

Vrij: Chapter 8

What is Statement Validity Assesment?

A “verbal veracity assessment tool”

 Originated in Sweden (1963) as a method to determine the credibility of child witnesses in sexual abuse cases

 Credibility of children in sexual abuse cases is critical, especially when there are no corroborating witnesses or physical evidence

So…

 Unlike non-verbal deception detection techniques, you are not looking for “tells” as to when a person is lying

Problems with child witness testimonies

 Vrij cites Craig, 1995, stating estimates range between 6% to 60% that child witness statements about sexual abuse are inaccurate

– Due to parental influence, outside pressure, simple misidentification, or complete lies

 Adults tend to mistrust statements made by children

History of SVA

 Udo Undeutsch and the West German

Supreme Court

Presented case of a 14-year-old alleged victim of rape using a method called statement analysis

Court ruled that outside psychologists had more and better resources to determine truthfulness than court “fact finders”

1955 – court requires use of psychological interviews and credibility assessments in disputed cases

History of SVA continued…

Undeutsch was the first to create a comprehensive list of criteria to assess credibility

In 1988, K ӧhnken and Steller refined the criteria and standardized it in to a formal assessment procedure

– Called it Statement Validity Analysis (SVA)

History of SVA continued…

So…

The current SVA method wasn’t created until the

1980s, more than 30 years after the German courts looked in to statement analysis

– Until this point, no studies had been done analyzing the validity of SA or SVA

Four Stages of SVA

 1. Case-file analysis

 2. Semi-structured interview

 3. Criteria-Based Content Analysis (CBCA)

 4. Evaluation of CBCA with the Validity

Checklist

Stage 1: Case-File Analysis

 Analysis of facts in a case

 Expert forms hypotheses about what happened. Details from the analysis will help the expert focus on critical details later in the interview.

Stage 2: Semi-Structured Interview

 What the Criteria-Based Content Analysis

(Stage 3) will analyze

 Child gives his/her account of the allegation

 Can be very difficult do to lack of verbal or cognitive skills in young children

– Also highly influenced by personality factors such as anxiety or simple embarrassment

 Skill and knowledge of interviewer is critical

Stage 2: Semi-Structured Interview continued…

 Interviewer must have a strategy for eliciting as much detailed information as possible

 Has to ask the right questions in the right way

– Must avoid leading, yes or no, questions

– Must get child (or adult for that matter) to tell story without interviewer influence

Stage 2: Semi-Structured Interview, continued…

Proper kinds of questions/techniques:

– Openended (e.g. “Tell me what happened.”)

– Facilitative responses

“OK”, “mmhm”, head nods, etc

– Focused questions

 Focus on specific details or aspects of the event

Problematic questions:

Leading (e.g. “Was it your dad?”

Optionposing (e.g. “Was the man white or black?”)

Stage 3: Criteria-Based Content

Analysis (CBCA)

 Used on transcripts of the interviews

 Consists of 19 criteria judged on a three point scale.

“0” if criteria is absent, “1” if criteria is present, “2” if criteria is strongly present

 Consists of four categories

Stage 3: CBCA – The Four Categories

 1. General Characteristics

 2. Specific Contents

 3. Motivation-Related Contents

 4. Offence-Specific Elements

Stage 3: CBCA – General

Characteristics (1-3)

 1. Logical Structure

– Statement is coherent and logically consistent

 2. Unstructured Production

– Information is presented in non-chronological order

 3. Quality of Details

– Statement is rich in details

Stage 3: CBCA – Specific Contents (4-

13)

4. Contextual Embedding

– Events are placed in time and location

5. Descriptions of Interactions

– Statements contain information that interlinks the alleged perpetrator and witness

6. Reproduction of Conversation

– Specific dialogue, not summaries of what people said

7. Unexpected Complications During the Incident

Stage 3: CBCA – Specific Contents (4-

13) Continued…

8. Unusual Details

– Tattoos, stutters, individual quirks

9. Superfluous Details

– Details that are non-essential to the allegation

10. Accurately Reported Details Misunderstood

Mentioning of details outside a person’s scope of understanding

11. Related External Associations

Stage 3: CBCA – Specific Contents (4-

13) Continued…

12. Accounts of Subjective Mental State

Description of a change in a subject’s feelings during the incident

13. Attribution of Perpetrator’s Mental State

Witness describes perpetrator’s feelings

Stage 3: CBCA – Motivated-Related

Contents (14-18)

 14. Spontaneous Corrections

15. Admitting Lack of Memory

16. Raising Doubts About One’s Own

Testimony

 17. Self-Deprecation

 18. Pardoning the Perpetrator

Stage 3: CBCA – Details Characteristic of the Offence (19)

 19. Offence-Specific Elements

– Descriptions of elements that are known by professionals to be typical of a crime

Stage 4: Evaluation of CBCA with the

Validity Checklist

 The CBCA score alone is not enough to determine if a person is being truthful

 The examiner must also take into account other factors that could have affected the outcome

– Leading by the interviewer, outside influences, witness’s cognitive abilities, etc…

 The CBCA is NOT a standardized test

Stage 4: Evaluation of CBCA with the

Validity Checklist, continued…

 Attempts to standardize the CBCA results through an 11 point checklist

 Allows the examiner to consider alternative reasons for CBCA outcomes

 As these alternative reasons are rejected, the CBCA results become stronger (in the assumption that the score represents the veracity of the statement)

Stage 4: Evaluation of CBCA with the

Validity Checklist, continued…

 The Four Stages:

– 1. Psychological Characteristics

2. Interview Characteristics

3. Motivation

– 4. Investigative Questions

Stage 4: Evaluation of CBCA with the

Validity Checklist – Psych

Characteristics

 1. Inappropriateness of Language and

Knowledge

 2. Inappropriateness of Affect

 3. Susceptibility to Suggestion

Stage 4: Evaluation of CBCA with the

Validity Checklist – Interview

Characteristics

 4. Suggestive, Leading, or Coercive

Interviewing

 5. Overall Inadequacy of the Interview

Stage 4: Evaluation of CBCA with the

Validity Checklist – Motivation

 6. Questionable Motives to Report

– Both for witness and other parties involved

 7. Questionable Context of the Original

Disclosure or Report

 8. Pressures to Report Falsely

Stage 4: Evaluation of CBCA with the

Validity Checklist – Investigative

Questions

 9. Inconsistency with the Laws of Nature

 10. Inconsistency with Other Statements

 11. Inconsistency with Other Evidence

SVA – Issues

 Effectiveness of individual criteria in CBCA

 Effectiveness of Validity Checklist

 Differences between laboratory and field studies

 Detection rates and false-positives

 Countermeasures

 Applicability to adults?

 The Daubert Standard

CBCA – Issues

 Not all statements are equally effective

– A claim by a young child with less detail will be scored lower on the CBCA scale than that of an older child or adult

 Not all criteria are created equal

– Generally, the criteria in groups 1 and 2 are the most effective at distinguishing truth-tellers from liars

CBCA – Inter-Rater Reliability

 Are CBCA scores found by one rater close to those of a second, independent rater?

– Good for most criteria, except unstructured production and spontaneous corrections

– Overall score agreement is higher than on individual criteria

Vrij’s Literature Review

 Laboratory vs. Field studies

Deficiencies for one type are the other’s strengths

 Lab – Not realistic, often based off observation of a video

 Field – “Ground truth” cannot always be established, methods of finding it are not always consistent

In field studies, low quality statements are less likely to obtain a truthful diagnosis or a conviction/confession, even if true

High CBCA scores on false claims can lead to falseconfessions or convictions

Therefore, relationship between CBCA scores and convictions or confessions may not be accurate

Esplin et al., (1988)

 Field study

 CBCA scored on 0-2 scale (range of scores could be 0-38)

– Confirmed statement average = 24.8

– Doubtful statement average = 3.6

Differences between groups found in 16/19 criteria

However, there are criticisms…

CBCA results from other studies

Boychuck (1991) – 13/19

Lamb et al. (1997b)* – 5/14

– Plausible average = 6.74

– Implausible average = 4.85

Parker & Brown (2000) – 6/18

Rassin & van der Sleen (2005) – 2/5

Craig et al. (1999)*

Confirmed average = 7.2

Doubtful average = 5.7

 * used a 0-1 pt scale on CBCA

Critical Difference to Non-verbal

Studies:

 All results found were in the expected direction, supporting the Undeutsch

Hypothesis

Results in non-verbal studies are highly erratic

You may find non-verbal cues within individuals, but between groups these do not exist

CBCA – Lab Studies

 Difficult to create realistic situations

 Accuracy rates ranged from 54% to 90%

– Average rates for truths = 70.81%

– Average rates for lies = 71.12%

– Rates did not differ between children, adults, witnesses, victims, or suspects

CBCA – Lab Studies, continued…

 Serious methodological problems:

– Different situations used

Different analysis methods used

Different amounts of training for raters

– Some studies do not use the Validity Checklist and base diagnoses purely upon the CBCA

CBCA – Lab Studies, continued…

 But some important results remain

– For the most part, all differences found were in the correct direction, once again supporting

Undeutsch

– Some individual criteria are more effective than others

 Support percentages (differences found / studies investigated)

– Range from 76% (Criteria 3) to 0% (Criteria 17)

CBCA – Lab Studies, continued…

 Other effective criteria:

– 4. Contextual embeddings

6. Reproductions of conversations

8. Unusual details

 Least effective:

– 14-18 – Motivational Criteria

– 17. Self deprecation actually occurred less in truth tellers in two studies

CBCA – Classifications

 1. Discriminate (statistical) analysis is the most common method

 2. Rater makes own truth/lie classification

Computer analysis better at detecting lies

 80% vs. 60% for human raters

People better at detecting truths

 80% vs. 53% for computers

 3. General decision rules

– E.G. Criteria 1-5, plus two others

Reviewing the Validity Checklist

 Focuses on three things:

1. Age of interviewee

 Highly affects cognitive abilities

 Older age correlates with higher CBCA scores

2. Interviewer’s style

Open-ended questions are most effective

The “Cognitive Interview”

3. Coaching of interviewee

 Countermeasures

Training of subject to include CBCA criteria in their statement

Easily defeat the CBCA analysis (only 27% of coached liars caught)

What the layperson believes…

Generally correct about number of details (Criterion

3) and descriptions of interactions (5)

Generally believe liars include more contextual embeddings (Criterion 4), unusual details (8), and superfluous details (9) in stories

Overall, the layperson’s view differs somewhat from the experts’ view

– This, potentially, is a good thing

Problems with the Validity Checklist

 Difficulty in identifying issues

– Coaching by an adult is hard to discover

 Difficulty in measuring issues

– E.g. susceptibility to suggestion

 Difficulty in determining impact of issues

The validity checklist is much more subjective and less formalized than the CBCA

It is therefore harder to study

Vrij’s specific problems with VC

Issue 2 – Inappropriateness of Affect

– Cites research that suggests there are two main psychological reactions to a rape

 1. Expressed style

 2. Numbed style

Issue 10 – Inconsistencies between statements

– Human memory is not perfect, details can be unintentional

– A practiced lie will not contain as many inconsistencies

Issue 9 and 11 (Consistency with laws of nature, consistency with other evidence)

Children’s scope of understanding often include fantasies and other things not in agreement with natural laws

– Sometimes, even in a true allegation, no other evidence can be found

Vrij’s specific problems with VC, continued…

 Embedded false statements are difficult to detect

 False memories

The Daubert Standard

 Daubert vs. Merrel Dow Pharmaceuticals,

Inc. (1993)

– Set standards for the inclusion of expert witness testimony in court cases in the United states

– Consists of 5 criteria that must be met for evidence to be admissible in court

The Daubert Standard, continued…

 1. Is the scientific hypothesis testable?

 2. Has the proposition been tested?

 3. Is there a known error rate?

 4. Has the hypothesis and/or technique been subjected to peer review and publication?

 5. Is the theory upon which the hypothesis and/or technique based generally accepted in the appropriate scientific community?

So, what about SVA?

CBCA

Lab

Yes 1. Is the scientific hypothesis testable?

2. Has the proposition been tested?

3. Is there a known error rate?

Yes No

Yes, too high No

CBCA

Field

Problematic

4. Has the hypothesis been subjected to peer review/publication?

5. Is the theory based on generally accepted principles?

Yes

Unknown

Yes

Unknown Unknown

No

No

No

Validity

Checklist

Problematic

SVA

Problematic

No

No

No

Unknown

Error rates

 Refer to subjects that are classified incorrectly

– Truth tellers classified as liars, and vice-versa

 Error rate for CBCA judgments made in laboratory research is nearly 30% for both truths and lies

– This is EXTREMELY high

Overall evaluation of SVA

While results from research on SVA strongly support the Undeutsch Hypothesis, SVA does not meet the requirements of the five criteria established by the

Daubert Standard

70% correct classification is OK

30% error rate is much too high for a valid test

Certain criteria in the CBCA appear to be highly effective at discriminating truth tellers from liars

Other criteria are wholly ineffective

In the end…

 CBCA and SVA would be an effective tool for use in the initial stages of investigations

 Results from these tests can guide police throughout investigations

 CBCA and SVA appears to be effective on adults also, not just useful in situations of child sexual abuse

Download