A suggestibility scale for children

https://doi.org/10.1016/S0191-8869(00)00077-5Get rights and content

Abstract

This article describes the development and properties of the Video Suggestibility Scale for Children (Video SSC), a scale designed to measure individual differences in suggestibility in preschool children. The scale was administered to 195 3- to 5-year-old children, and the results suggest that they tend to respond affirmatively to suggestive questions (‘Yield’) and change their answers in response to negative feedback (‘Shift’) in a manner similar to adults, though at an elevated rate. Older children were able to remember the video better than younger children in memory recall, but were also more likely to shift their answers in response to negative feedback. The scale items had satisfactory internal consistency and were factor analyzed using the Varimax procedure. Yield and Shift were found to load on different factors, supporting Gudjonsson’s (1984) view that there are at least two basic types of interrogative suggestibility [Gudjonsson, G. H. (1984). A new scale of interrogative suggestibility. Personality and Individual Differences, 5, 303–314]. Finally, preliminary validity analyses with an early version of the Video SSC reveals that it is predictive of suggestibility errors made in an independent study, using different procedures.

Introduction

The suggestibility of children in interrogative situations is an issue of major importance, from both developmental and legal standpoints. In America alone each year there are hundreds of thousands of very young children involved in juvenile and criminal justice proceedings1. Although only a fraction of these children end up testifying in criminal court, all of them are interviewed by law enforcement officials and/or child protective service workers, and many will give depositions and unsworn testimony. The most recent national incidence (Sedlak & Broadhurst, 1996) makes clear that young children are the most likely age group to be involved in serious matters, with an incidence rate of 5.1 per 1000. Summing across all reports of sexual abuse, children aged 7 and younger comprise 40.65% of all cases. To this evidence, we can add the data from the most recent Department of Health and Human Services (1999): ‘When victims are grouped in 4-year age categories, those 4–7 years old were the highest proportion of victims (26.2%)’.

Thus, the sheer numbers of children who are interviewed each year is enormous, and in many of these cases the child’s answers to the interviewer’s questions emerge as centrally important. And if one adds to the above high incidence of abuse and neglect proceedings involving very young children, the number of young children embroiled in acrimonious custody disputes, PINS (Persons In Need of Supervision) actions, neglect proceedings, and termination of parental rights suits, then the absolute numbers of young children enmeshed in the justice system is astronomical. Hence, it behooves society to better understand individual differences in the way young children respond to investigative interviewers’ questions.

Recent appeals court decisions in America have overturned a number of convictions of people convicted of crimes against children because the evidence in the cases was largely built upon the testimony of children whose reliability may have been compromised by suggestive interviewing techniques (e.g. see Commonwealth of Massachusetts vs LeFave, 1998). In the highly publicized Kelly Michaels case, the appellate court went further than merely overturning the conviction of preschool teacher Kelly Michaels. They held that if the interviews of the children are so suggestive as to render them tainted, and if a defendant can demonstrate that a child’s testimony was influenced by pretrial suggestive interviews, the child will be prohibited from testifying unless the state can establish the testimony’s reliability by clear and convincing evidence (Lyon, 1999). In the court’s words:

In assessing the trustworthiness of a child’s statements, the Michaels Court recognized that ‘Among the factors that can undermine neutrality and create undue suggestiveness is … the pursuit by an interviewer of a preconceived notion of what has happened to the child’ (State vs Michaels, at 309, 642 A.2d 1372).

It is against this backdrop of large numbers of children being interviewed and courts rejecting their claims if obtained in suggestive interviews that the present research was situated. Below we briefly review some of the seminal research on both children’s suggestibility and on the development of adolescent and adult scales for assessing suggestibility. Following this, we describe our own attempts to develop and validate a suggestibility scale for the preschool age group.

Throughout the 20th century, researchers have conducted studies that have sought to identify both individual and situational variables that affect the reliability of children’s testimony (see Ceci and Bruck, 1993, Ceci and Bruck, 1995, for reviews). Starting with Alfred Binet’s pioneering research on this topic 100 years ago (Cunningham, 1988), researchers have studied situational variables that induce children to comply with adults’ wishes. For example, in Binet’s early studies the most potent variable was whether the children heard others previously report incorrectly, as this seemed to encourage them to conform. Hence, Binet lamented: ‘Woe betide him who is alone’ (Malheur a celui qui est seul) (Binet, 1905).

Despite important progress made on identifying situational variables that are associated with children’s suggestibility, much less is known about the characteristics within the children themselves that cause some to be extremely accurate while others are easily misled (for recent reviews see Bruck et al., 1997, Quas et al., 1997). According to these reviews, there seems to be an emerging consensus that children’s suggestibility entails an interplay of both individual characteristics and situational factors. It is only fairly recently that individual-difference factors that contribute to children’s suggestibility have begun to be examined systematically. In the present study, we will combine data from some recent rounds of data collection that examined individual differences in children’s suggestibility using the Video Suggestibility Scale for Children developed by Scullin and his colleagues (Video SSC; Scullin & Hembrooke, 1998) in an attempt to validate this measure and examine its relationship to some presumed measures of interrogative suggestibility.

In psychological research, the term suggestibility has an extremely broad array of connotations. For example, in Gudjonsson and Clark’s (1986) model of interrogative suggestibility, suggestibility was defined as ‘the extent to which, within a closed social interaction, people come to accept messages communicated during formal questioning, as the result of which their subsequent behavioural response is affected’ (p. 84). Ceci and Bruck’s (1993) definition of suggestibility in children stated that it ‘concerns the degree to which children’s encoding, storage, retrieval, and reporting of events can be influenced by a range of social and psychological factors’ (p. 404). The Gudjonsson and Clark model placed an emphasis on coping strategies a witness develops to deal with the uncertainty and expectations of an interrogation and the incorporation of post-event information into memory. Ceci and Bruck additionally emphasized the possibility that acquiescence, lying, and information the witness receives preceding an event may affect the encoding, storage, and retrieval of memory of an event.

The seminal research on interrogative suggestibility in adults was conducted in England by Gudjonsson, 1984, Gudjonsson, 1987, Gudjonsson, 1997), who developed two forms of the Gudjonsson Suggestibility Scales (the GSS1 and GSS2). These scales have been used extensively in forensic research. In Gudjonsson’s scales, a brief story with 40 salient points is played on audiotape. Immediately after the tape is played, the participant is asked to tell as much of the story as he or she can remember and the number of salient points recalled is scored. The participant is then asked a series of 20 questions, 15 of which consist of varieties of leading questions and 5 of which are ‘true,’ calling for an affirmative response to something that actually happened in the story. Gudjonsson devised the scale to measure suggestibility along two independent dimensions. The first is called Yield, which is the degree to which a person initially responds to the 15 suggestive questions that have a degree of expectation in them. After a participant initially is read all of the questions, he or she is given some negative feedback about his or her performance (e.g. ‘You made some errors so I’m going to ask the questions again.’) and read all the questions again. This leads to the second measure, Shift, which assesses the number of all 20 answers changed in response to the negative feedback. Total suggestibility is measured by adding Yield and Shift.

Gudjonsson, 1984, Gudjonsson, 1992a) factor analyzed data from both forms of the scale and found that when Yield and Shift scores were analyzed together, Yield and Shift items loaded on two different factors after Varimax rotation. This supported Gudjonsson’s view that there are two types of interrogative suggestibility which correspond to the degree people give into misleading questions and how they respond to negative feedback.

A number of studies have addressed whether the types of suggestibility identified by Gudjonsson are developmentally invariant. Gudjonsson and his colleagues have used their scale to do some developmental analyses, finding that there is a decline in total suggestibility from adolescence to adulthood (reviewed in Gudjonsson, 1992b). Using Gudjonsson’s scale, Danielsdottir, Sigurgeirsdottir, Einarsdottir and Haraldsson (1993) also found a decline in Total Suggestibility in children between the ages of 6 and 12, and that boys were significantly more likely to Shift their answers than girls. Warren, Hulse-Trotter and Tubbs (1991) also found a marked decrease in suggestibility with age in a study using the GSS1 and comparing 7-year-olds, 12-year-olds, and adults. Importantly, Warren et al. (1991) played the audiotaped story twice for the 7-year-old children in order to ensure that they understood and followed it. However, reading a story twice to children has been found to greatly reduce suggestibility by increasing memory trace strength (Endres, Poggenpohl & Erben, 1999). Richardson, Gudjonsson and Kelly (1995) have found that in both normal and adolescent delinquent populations, adolescents tend to Shift (but not Yield) more than adults. Finally, Henry and Gudjonsson (1999) compared performance of a group of mentally retarded 12-year-olds, a comparable mental age comparison group of 7-year-olds, and a group of normal 12-year-olds on both the GSS2 and a forensic-style interview about a staged event that occurred a day earlier. They found that mentally retarded 12-year-olds were most similar to the 7-year-olds on Total Suggestibility, Yield, and Free Recall, but more similar to normal 12-year-olds on Shift. Additionally, they found that Yield was the best predictor of eyewitness performance in the mentally retarded group and that Free Recall was the best predictor of eyewitness performance among the 7-year-olds, but there was no relationship between the GSS2 measures and eyewitness performance among the normal 12-year-olds.

Thus, it appears that between the ages of six and young adulthood, there is a marked decline in how often children shift their answers in response to negative feedback from an interviewer, as well as in how often they yield in response to suggestive questions. The finding of an overall decline in suggestibility with age in these studies with the full scale GSS is consistent with Ceci and Bruck’s (1993) review of a wide variety of research that has consistently found that there is a negative correlation between age and suggestibility.

In order to develop a fuller picture of children’s suggestibility and investigate individual differences in suggestibility in preschool children, researchers have begun to develop measures specifically for very young children. In this research, an effort has been made to separate different types of suggestibility (Yield vs Shift) to make developmental comparisons possible over the entire age range between early childhood and adulthood.

While the GSS has been used with children as young as 6-years-old, the use of an audiotaped story makes its use with younger children somewhat problematic. Scullin and his colleagues (Scullin & Hembrooke, 1998) used Gudjonsson’s procedure as a starting point for the development of a Video SSC, retaining the measures of Yield, Shift, and Total Suggestibility. In an attempt to provide a story that young children could easily follow, a video (rather than an audiotape) was used to tell a story that would be encoded in a manner similar to that of an event that a child might actually witness or experience (i.e. involving both auditory and visual encoding). Scale questions were administered at intervals of one day to one week after the story (rather than immediately after the story) in order to better reflect a demand for retrieval of memories of an event some time after it occurs, thus allowing more time for forgetting story details as might occur in an actual forensic investigation. Additionally, children were given mild negative feedback (e.g. ‘You missed a few of the questions. Let’s go through them again and see if you can do better this time.’) halfway through the scale procedure and again at the end in order to keep the feedback salient, with half of the questions repeated each time. Two of the three types of suggestive questions identified by Gudjonsson (1984) were retained. These included leading questions, whose phrasing contained premises that provided expectations that a certain answer would be provided (e.g. ‘Did the kids break a balloon while they were hitting them around?’ contained the premise that the kids were hitting a balloon around). Also included were affirmative questions, which contained no premises but seemingly called for an affirmative answer (e.g. ‘Was there a little white doggie at the party?’ when there was no dog at the party). Due to the difficulty young children have answering Gudjonsson’s third type of questions, false alternative questions (e.g. ‘Was the ball red or blue?’ when it was actually yellow), this type of question was dropped from the scale2.

In the initial study with the Video SSC (Scullin & Hembrooke, 1998), the correlation between Total Suggestibility scores and how quickly the child assented to an imagined event about which the child was interviewed over the course of several weeks (in an earlier, completely independent experiment conducted by another research team) was, although reliable, fairly low (r=0.38). A Cronbach’s alpha analysis of the Shift part of the scoring revealed that it had poor internal consistency. By deleting items with poor item-total correlations on Shift, the scale was reduced to seven questions with satisfactory internal consistency on Shift. When this was done the correlation between the total suggestibility score on the reduced scale and how readily the child assented to the imagined event in repeated interviews of the earlier experiment increased (r=0.62)3. Because many of the questions in the initial version of the scale had poor internal consistency, they were replaced with new questions. This article reports data obtained with the revised scale.

The present study was intended to examine the following hypotheses: (1) Yield and Shift will load on separate factors when factor analyzed; (2) Yield and Shift have satisfactory internal consistency; (3) the child’s gender and the interval between viewing the video and the administration of the scale questions will not significantly affect the suggestibility scores; and (4) scores on the suggestibility measures will significantly decrease with age.

Section snippets

Participants

Ninety-seven boys and 98 girls who were enrolled in daycare centers in central New York State served as participants. The ages of these 195 children ranged from 33 to 76 months, with a mean age of 52.54 months and a standard deviation of 8.71 months.

Design and procedure

Individually or in groups of 2 or 3, the children were shown a 5-min video about a birthday party. The video featured two children, Tammy and Suzie, who arrived at a house just as a birthday party was getting underway. The children were greeted by

Frequency of variables

Table 1 presents the mean, standard deviation, and number of observations for each of the variables in the study. Of note is that the total number of items recalled in Memory Recall was quite small, with a mean of only 2.43 (SD=2.08) pieces of correct information. This low amount of Memory Recall is consistent with the literature showing that what preschool-aged children produce in response to free recall instructions is very low but accurate (Ceci & Bruck, 1995). Children’s Yield scores

Discussion

This paper discussed the development of a video suggestibility scale for children to extend the study of individual differences in suggestibility to children as young as 3 years of age. The Video SSC shows a similar factor pattern to the GSS1 and the GSS2, with comparably high levels of internal consistency. There were no differences between boys and girls on the measure, and varying the length of time between 1 day and 7 days did not affect scale scores. These results suggest that the study of

References (31)

  • Commonwealth of Massachusetts vs LeFave. (1998). Findings of fact, rulings of law and order on defendant’s motion for a...
  • J.L. Cunningham

    Contributions to the history of psychology: XLVL. The Pioneer work of Alfred Binet on children as eyewitnesses

    Psychological Reports

    (1988)
  • Department of Health and Human Services. (1999). Child Maltreatment 1997: Reports from the States to the National Child...
  • J. Endres et al.

    Repetitions, warnings, and video: Cognitive and motivational components in preschool children’s suggestibility

    Legal and Criminological Psychology

    (1999)
  • G.H. Gudjonsson

    A parallel form of the Gudjonsson Suggestibility Scale

    British Journal of Clinical Psychology

    (1987)
  • Cited by (84)

    • Attentional difficulty is a risk factor for interrogative suggestibility in preschoolers

      2020, Children and Youth Services Review
      Citation Excerpt :

      According to the Department of Human Services information about general income, occupation, and education of residents in the area, just over half of the sample (n = 74, 32 boys) was recruited from schools with predominantly low SES families, while the remaining children (n = 66, 31 boys) attended schools with predominantly high SES families. Video Suggestibility Scale for Children (VSSC; Scullin & Ceci, 2001). The VSSC is a standardised tool to test memory and suggestibility in preschoolers.

    • Valence, implicated actor, and children's acquiescence to false suggestions

      2016, Journal of Applied Developmental Psychology
      Citation Excerpt :

      We employed a delay of approximately one week between the to-be-remembered event and the memory interview. This delay is common to that in prior studies that have examined highly suggestive reinforcement tactics on children's suggestibility and found robust effects (Garven et al., 1998, 2000; Kulkofsky & Klemfuss, 2008; Scullin & Ceci, 2001). We also analyzed children's responses to questions about true details and found a fairly high level of response accuracy (76%–82%).

    View all citing articles on Scopus
    View full text