THE COGNITIVE INTERVIEW AND ITS EFFECT ON WITNESSES’ CONFIDENCE

Description
THE COGNITIVE INTERVIEW AND ITS EFFECT ON WITNESSES’ CONFIDENCE

Please download to get full document.

View again

of 16
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Screenplays & Play

Publish on:

Views: 7 | Pages: 16

Extension: PDF | Download: 0

Share
Tags
Transcript
  THE COGNITIVE INTERVIEW AND ITS EFFECT ONWITNESSES’ CONFIDENCE PA¨ R ANDERS GRANHAG a *, ANNA-CARIN JONSSON b and CARLMARTIN ALLWOOD ba Department of Psychology, Go¨ teborg Uni  v ersity, PO Box 500, SE-40530 Go¨ teborg, Sweden; b Department of Psychology, Lund Uni  v ersity, PO Box 213, SE-22100 Lund, Sweden Today there is ample evidence that the Cognitive Interview (CI) enhances witnesses’ memory. However, less isknown about how the CI affects eyewitnesses’ confidence. To address this shortcoming we conducted a studyanalyzing how realism in confidence was affected by the CI. All participants ( n  / 79) were first shown a filmedkidnapping. After 2 weeks we interviewed one-third of the participants according to the guidelines of the CI,one-third according to a Standard Interview (SI), and one-third were not interviewed at all (Controlcondition). Participants in all three conditions were then asked to answer 45 forced-choice questions, and togive a confidence judgment after each choice. For the 45 questions, no differences in accuracy were foundbetween the three conditions. Confidence was higher in the CI and SI conditions, compared with the Controlcondition. CI and SI did not differ in metacognitive realism but both showed lower realism compared with theControl condition, although only CI significantly so. The results indicate that the inflation in confidence ismore likely to be explained in terms of a  reiteration effect , than as a consequence of the particular mnemonicscharacterizing the CI (e.g. ‘‘mental reinstatement of context’’). In sum, CI does not seem to impair (orimprove) the realism in witnesses’ confidence, and does not inflate confidence in erroneous recall, compared toa SI. Keywords:  Realism in Confidence; Calibration; The Cognitive Interview; Reiteration Effect INTRODUCTION Many criminal investigations are dependent on information from witnesses. However, policeofficers commonly express that they find witnesses’ statements to be very general andincomplete (Kebbel and Milne, 1998). This view is supported by archival studies focusing onthe amount and type of information reported by witnesses (van Koppen and Lochun, 1997).Hence, it is of paramount importance that interview techniques that help witnesses andvictims to elicit both correct and complete statements are developed.In the last 10 years much research has focused on one such technique, the  Cogniti  v eInter v iew  (Fisher and Geiselman, 1992; Milne and Bull, 1999). Although several differentaspects of the Cognitive Interview (CI) have been addressed in previous research, littleattention has been paid to how the technique affect the interviewees’ metacognitive judgments. In order to improve our understanding of this, the main focus of the presentstudy is to examine the extent to which the use of the CI influences the realism of witnesses’confidence judgments. We will first briefly review research on the CI, and then we will turn toresearch investigating witnesses’ confidence. ISSN 1068-316X print/ISSN 1477-2744 online # 2004 Taylor & Francis LtdDOI: 10.1080/1068316021000030577 *Corresponding author. E-mail: pag@psy.gu.se Psychology, Crime & Law,  March 2004, Vol. 10(1), pp. 37    / 52 37  The Cognitive Interview The CI rests upon two well-known memory principles. First, according to the multi-component view of memory (e.g. Bower, 1967), memory traces are multi-featured, and if onefeature is not accessible using a particular type of retrieval probe, another type of probemight help to unlock the memory. The second principle is the ‘‘encoding-specificityprinciple’’, which suggests that the most effective retrieval probes are those that create aretrieval environment most similar to the srcinal encoding environment (e.g. Tulving andThomson, 1973).Using these principles as the point of departure, the founders of the CI, Geiselman andFisher, suggested four mnemonics (Geiselman  et al. , 1986) that would enhance recall: (1) the‘‘mental reinstatement of context’’ instruction, which asks the interviewee to reinstatementally both the external context (physical surroundings) and internal context (subjectivestate-of-mind), of the event in question; (2) the ‘‘report everything’’ instruction, whichinstructs the interviewee to report all details that he or she can remember; (3) the ‘‘reverse-order recall’’ instruction, which encourages the interviewee to recall the event in analternative temporal order; and (4) the ‘‘change perspective’’ instruction, which asks theinterviewee to recall the event from the perspective of another witness.Since the mid-1980s, a large number of studies examining the effectiveness of the CI havebeen conducted. The resulting pattern is very clear; the CI elicits more correct informationthan a comparison interview (for a meta-analysis, see Ko¨hnken  et al. , 1999). However, insome studies it has also been found that the CI also generates a slight increase in the numberof incorrect details (Ko¨hnken  et al  ., 1999).In recent years, the effectiveness of the CI has been investigated in relation to a number of different populations, including adults (Geiselman  et al  ., 1986), adults with learningdisabilities (Milne  et al. , 1999), elderly people (Mello and Fisher, 1996) and children(Granhag and Spjut, 2001; McCauley and Fisher, 1995). Furthermore, several differentaspects of the CI have been addressed, such as whether the positive effects of the CI hold forreal-life witnesses (Fisher  et al. , 1989) or over longer retention intervals (Larsson  et al. ,2003). However, many important aspects of the CI still remain to be examined, and one suchaspect relates to how the use of the technique affects the realism of witnesses’ metacognitive judgments with respect to the correctness of their recall. Eyewitness Confidence To date, several systematic reviews have reported that the relationship between eyewitnesses’accuracy and confidence is weak (e.g. Wells and Murray, 1984; Bothwell  et al. , 1987; Sporer et al. , 1995). In sharp contrast to this finding, several studies have shown that there is astrong tendency among, for example, jurors to rely on eyewitnesses’ confidence wheninferring the accuracy in a testimony (Wells  et al. , 1979; Cutler  et al. , 1988). Consequently,an important research objective is to map factors that increase, as well as factors thatdecrease, the realism of witnesses’ confidence.Previous research has very clearly demonstrated that eyewitnesses are overconfident intheir judgments, and that confidence, as such, is malleable in its nature. Specifically, researchhas demonstrated that factors such as hindsight bias (Granhag  et al. , 2000), the use of arepeated recall schedule (Shaw, 1996; Granhag, 1997) and feedback, for example in the terms of different types of co-witness information (Luus and Wells, 1994), moderates witnesses’ 38 P. A. GRANHAG  et al  .  confidence. In short, using different sorts of manipulations, eyewitnesses’ confidence caneither be deflated or inflated. The Cognitive Interview and Eyewitnesses’ Confidence Several arguments can be given as to why the type of interview technique employed canmoderate witnesses’ confidence (see also Gwyer and Clifford, 1997). First, the vividness of arecalled memory may, at least in part, be dependent upon the particular interview techniqueused. For example, the use of the ‘‘mental reinstatement’’ mnemonic might help to facilitatevivid memories. In turn, the vividness of a memory is likely to act as a cue for confidence(Garry  et al. , 1996). Second, the amount of information recalled may affect witnesses’confidence (Gwyer and Clifford, 1997). For example, the ability to recall a large number of details may, independently of the actual correctness of the statement, induce a sense of excellent memory performance. Third, it is reasonable to assume that witnesses’ confidence isinfluenced by social aspects of the interview situation, aspects that are dependent upon theinterview technique used. The CI draws both on vividness (the ‘‘mental reinstatement’’instruction), completeness (the ‘‘recall all details’’ instruction) and social aspects of theinterview situation (e.g. ‘‘building rapport’’). Finally, Roberts and Higham (2002) presentedevidence that metacognitive realism was better for forensically central information comparedwith peripheral information. However, it is unclear if this conclusion also pertains to otherinterview techniques since no such comparisons were made.McCauley and Fisher (1995) first let second graders play a ‘‘Simon says game’’ and theninterviewed them according to either the CI or a Standard Interview (basically, buildingreport, asking for free narrative and then asking specific questions about unclear parts of thenarrative). Following each interview, the participants indicated the confidence of theirrecollections on a scale from 1 (‘‘not at all sure’’) to 5 (‘‘really, really sure’’). The resultshowed that the level of confidence was unaffected by type of interview technique used.Gwyer and Clifford (1997) showed adults a live staged event (an interaction between ateacher and a confederate) and then conducted interviews according to either the CI or aStructured Interview (used as a control interview). Confidence measures were collected bothin relation to participants’ recall memory and recognition memory (line-up identifications).As the current paper focuses on witnesses’ descriptions, only the results for recall memorywill be discussed. Each participant assessed their confidence before, throughout and after theinterview. For pre- and post-interview confidence the participants had to estimate how surethey were that they would be able to answer the forthcoming questions (pre), and how surethey were that they actually had answered the questions correctly (post). In addition,confidence ratings for three different event categories were collected throughout theinterview (i.e. person descriptions, actions and objects).It was found that participants who were interviewed according to the CI had asignificantly higher pre-post confidence difference (higher level in the post rating) comparedwith the participants interviewed according to the Structured Interview. Gwyer and Cliffordargued that the increase in confidence for the CI condition is intrinsically related to the CI asan overall technique. Of the four mnemonics characterizing the technique, they speculatedthat particularly the ‘‘context reinstatement’’ and the ‘‘report all detail’’ instruction mayboost the witness’s feeling of confidence. In brief, as the witness has recalled much and vividinformation about the event, the witness may come to believe that he or she must have averygood memory. COGNITIVE INTERVIEWS AND WITNESSES’ CONFIDENCE 39  The average confidence level for the three event categories showed a mixed pattern; higherconfidence scores for information pertaining to person descriptions for the CI condition,higher confidence scores for information pertaining to object information for the SIcondition, and no differences in confidence for information pertaining to actions. For the CIcondition a significant confidence    / accuracy relationship was found for person descriptionrecall ( r  / 0.51), and for the SI condition a significant confidence    / accuracy relationship wasfound for object recall ( r  / 0.36). None of the other correlations calculated was significant.Unfortunately, from the way datawere analyzed in the two studies reviewed above there is noway of telling how the use of the CI affected the realism in the participants’ confidence judgments.Furthermore, Gwyer and Clifford used the traditional method of measuring theaccuracy    / confidence relation, by calculating point-biserial correlation coefficients. Asargued by, for example, Juslin  et al.  (1996), the correlation coefficient primarily speaks tothe issue of discrimination rather than calibration, and a so-called calibration approachis recommended. For extensive discussions on the advantages of using the calibrationapproach over the traditional correlation measure when analyzing the relationship betweenaccuracy and confidence in eyewitness contexts see Cutler and Penrod (1989) and, especially,Juslin  et al.  (1996). Frequency Judgments In order to obtain a measure of how a participant views his or her overall memoryperformance, one can ask for a so-called  frequency judgment . Previous research investigatingsemantic knowledge shows that participants who are asked ‘‘how many questions do youthink you have answered correctly?’’ will, under some conditions, give highly realisticestimates (Gigerenzer  et al. , 1991), and under other conditions underestimate their ownperformance (Allwood and Granhag, 1996). Both Granhag (1997) and Granhag  et al  .(2000), using episodic material, found that witnesses tended to overestimate theirperformance when giving item-specific confidence judgments and, for the same material,underestimate their performance when giving frequency judgments. These findings allow forsituations where a witness, for example during a police interview, may state high confidencein the details constituting the report, and after the interview, express doubts when it comes tothe overall reliability of his or her memory report. In the current study we investigated howthe use of different interview techniques affected the frequency judgments. The Present Study In order to examine the extent to which the use of the CI affects the realism in witnesses’confidence, we showed our participants a video-clip of a staged crime and then used threeconditions. In the CI condition the participants were interviewed according to the CI, and inthe SI condition according to a Standard Interview (SI). The participants in the Controlcondition were not interviewed at all. Then all participants answered a set of two-alternative-forced-choice (2AFC) questions, and gave confidence ratings for each answer. Hence,participants’ confidence judgments were collected in a subsequent memory test, and notduring the actual interview. In order to analyze the realism in witnesses’ confidence judgments we used calibration measures (further described in the Methods section). 40 P. A. GRANHAG  et al  .  It should also be noted that research on eyewitness’s confidence has focused heavily onperson identification (Sporer, 1996). Lately, scholars within the field have highlighted theneed for more research on witnesses’ descriptions of persons and events (Wells  et al. , 2000).The present study contributes to filling this gap in the literature.With reference to previous research (Ko¨hnken  et al  ., 1999), we predicted that participantswould report more correct information in their interviews than in the Standard Interview. Inour second prediction we expected higher confidence in the CI and SI conditions comparedwith the Control condition. This was partially based on the assumption that the participantsin the CI condition would show higher confidence judgments than would the participants inthe SI and the Control condition, due to the specific mnemonics constituting the CI (i.e. the‘‘mental reinstatement of context’’ and the ‘‘report everything’’ instruction). Furthermore,due to the fact that the participants in both the SI and CI conditions first went through aninterview, and then answered a set of questions, we assumed that there would be a so-called reiteration effect  for both these conditions (Hertwig  et al. , 1997). The reiteration effect, (i.e.that repetition of an assertion increases the confidence in its truth), has found strong supportin previous research (e.g. Shaw, 1996; Shaw and McClure, 1996). Hence, we expected that the reiteration effect would add to the difference in confidence between the CI and the Controlcondition (i.e. higher confidence for the CI condition), and separate the SI and Controlcondition (i.e. higher confidence for the SI condition).Third, drawing on results from previous research (Ko¨hnken  et al  ., 1999), we predicted thatthe CI condition would result in better memory performance when answering our 2AFCquestions compared with the SI condition, and especially so compared with the Controlcondition. Fourth, following the above predictions, we expected no differences in terms of realism in confidence between the CI and the SI condition (the effect of increase inconfidence in CI caused by the CI mnemonics would be offset by the expected increase inaccuracy due to the same mnemonics), but that both of these conditions would show worserealism than the Control condition. Fifth, drawing on previous research, we predicted thatthe participants would underestimate their actual performance when making frequency judgments. METHODSParticipants The participants were 79 undergraduate students, 59 women and 20 men, attending variouscourses in psychology and sociology at Lund University, Sweden. All were paidapproximately US $7 for their participation. None of the participants had any previousexperience of this type of task. Design The participants were randomized into three conditions, the CI condition ( n  / 26), the SIcondition ( n  / 27), and the Control condition ( n  / 26). COGNITIVE INTERVIEWS AND WITNESSES’ CONFIDENCE 41
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks