- Email: [email protected]
Further insights into the relationship between the experience of life adversity and interrogative suggestibility
Personality and Individual Differences 51 (2011) 1056–1058
Contents lists available at SciVerse ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Short Communication
Further insights into the relationship between the experience of life adversity and interrogative suggestibility Kim E. Drake Department of Psychology, University of Derby, Kedleston Road, DE22 1GB, UK
a r t i c l e
i n f o
Article history: Received 18 June 2011 Received in revised form 11 August 2011 Accepted 17 August 2011 Available online 15 September 2011 Keywords: Interrogative suggestibility Gudjonsson Suggestibility Scale Life adversity
a b s t r a c t Research over the past few years has uncovered a consistent link between the reporting of intense negative life events and interrogative suggestibility on the Gudjonsson Suggestibility Scale (GSS). This study (N = 130) now investigates the functional form of this relationship. Findings show that a linear regression model provides the most acceptable fit to the observed values for yield 1, yield 2, shift and total suggestibility scores. However, results also show the presence of non linear, quadratic, aspects to the relationship between the experience of life adversity and yield 1 scores. Up until now GSS scores were thought to increase linearly with level of adversity experienced: this study provides some evidence that, for the yield 1 subscale of the GSS, this may not be the case – high levels of adversity may not necessarily lead to an increased risk of accepting misleading information during questioning as once presumed. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Interrogative suggestibility is a serious psychological vulnerability during police questioning that can manifest as the acceptance of misleading information, as a consequence of the use of leading questions, and/or answer-shifts in response to explicit pressure (in the form of critical feedback) during interview (see Drake, 2011; Gudjonsson, 2003). Research over the past few years has consistently uncovered a significant correlation between the reporting of intense negative life events (iNLEs) and interrogative suggestibility on the GSS (see Drake, 2010). The reported experience of intense negative life events appeared particularly related to answer shifts in response to negative feedback. A limitation within these prior studies however is that they assume a linear relationship without investigating the presence of non-linear effects, which can manifest as either acceleration or a levelling off of interrogative suggestibility levels at certain scores of reported life adversity experience. The aim of this brief report is to therefore explore the functional form of the relationship between the reported experience of iNLE and interrogative suggestibility. Unless these effects are explored, the work will remain of limited value to both researchers and practitioners (Flouri, Tzavidis, & Kallis, 2010). Theory would suggest a linear effect, whereby increments in scores of iNLE correspond with a steady additive effect on interrogative suggestibility scores (Gudjonsson & Clark, 1986). The greater the perceived intensity of the negative life events experienced (shown by such interviewees reporting their negative experiences E-mail address: [email protected] 0191-8869/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2011.08.017
are more intensely negative), the more negative/pessimistic an interviewee’s mindset (see Drake, 2010); during interview, when being questioned and especially when given negative feedback, such interviewees are more likely to interpret the interview situation itself as well as the explicit feedback as particularly negative. This may lead to a greater susceptibility to that feedback and any pressure associated with questioning and interrogative suggestibility scores on the GSS (see Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2007; Gudjonsson & Clark, 1986). It is therefore proposed that a linear regression model would provide the most acceptable fit to the data when predicting yield 1, yield 2, shift and total suggestibility scores based on scores of iNLEs. 2. Method 2.1. Participants The sample consisted of 130 participants, 92 females and 38 males (mean age = 19.35 years, standard deviation = 1.41, range = 18–26). Participants are an opportunity sample, recruited through the experimental participation scheme within the School of Psychology. All were undergraduates within the School. 2.2. Procedure 2.2.1. The Gudjonsson Suggestibility Scale 1 (GSS1; Gudjonsson, 1984; Gudjonsson, 1997) To measure interrogative suggestibility levels, participants were administered the Gudjonsson Suggestibility Scale 1 (GSS; Gudjonsson, 1984) individually by an interviewer, trained by a Chartered
1057
K.E. Drake / Personality and Individual Differences 51 (2011) 1056–1058
Forensic Psychologist in the administration of the GSS. The GSS is a semi structured interview procedure and is presented to each participant individually as a memory task. Participants are read a narrative describing a fictitious robbery, followed immediately by a free-recall phase and the distracter phase (where participants completed the Life Events Questionnaire [LEQ; Norbeck, 1984]). An interview on the content of the narrative follows. Within the GSS interview there are 20 questions. Responses to the first round of 15 leading questions (out of a total of 20 questions) provide the yield 1 score. This indicates the number of leading questions yielded to during the first round of questions. Immediately after this, negative feedback is given by the interviewer. The interviewee is told ‘‘you have made a number of errors, and it is therefore necessary to go through all of the questions once more and this time try to be more accurate’’. The 20 questions are then repeated to see how readily interviewees shift their initial answers as a result of the negative feedback received; a ‘‘shift’’ score is derived. A yield 2 score is also obtained (denoting the number of leading questions yielded to after negative feedback). 2.2.2. Life Events Questionnaire (LEQ) (Norbeck, 1984) The LEQ contains 82 items in total and is a modification of the instrument developed by Sarason, Johnson, and Siegel (1978) in that it has nine items of particular relevance to women. These include items such as ‘‘Major difficulties with birth control pills or devices’’. The nine additional items in the LEQ were introduced to reduce the gender bias in the Sarason et al. (1978) version. Participants were required to go through all the events listed, and if they had experienced them at any point of their life, to circle whether it had been a ‘‘good’’ experience or ‘‘bad’’ experience. They were then instructed to rate the extent to which those events had an effect on their lives at the time. The ratings went from 0 (‘‘no effect’’) to 3 (‘‘large effect’’). The LEQ has good test–retest reliability, with test–retest reliabilities of 0.78–0.83 and is a significant predictor of measures of (unfavourable) psychological and psychiatric symptoms. 3. Results and discussion 3.1. Descriptive statistics Table 1 presents the means and standard deviations for the observed variables. 3.2. Curve-fit estimations (Pedhazur, 1982) Linear models were fitted (and compared with quadratic regression models) to investigate the presence of linearity. Table 2 shows the R-square statistic of relationship strength (between the observed and model-predicted values of the
Table 1 Means (M) and standard deviations (SD) of observed variables (N = 130).
IR Y1 Y2 S TS iNLE
M
SD
14.1 4.45 5.55 3.74 8.19 17.9
5.69 2.32 2.81 2.56 3.38 9.74
Note: IR = immediate-recall; DR = delayed recall; Y1 = yield 1; Y2 = yield 2; S = shift; TS = total suggestibility; iNLE = intensity of negative life events.
Table 2 Linear (x) and quadratic (x2) regression model of the reported intensity rating of negative life events experienced on GSS scores (N = 130). GSS Y1 R2 x x2
Y2 F
Sig.
R2
S F
Sig.
R2
TS F
Sig.
R2
F
Sig.
.048 6.43 .012 .144 21.5 .000 .220 36.0 .000 .256 44.0 .000 .072 4.94 .009 .144 10.7 .000 .220 17.9 .000 .264 22.8 .000
Note: x = the reported intensity rating of negative life events experienced (see Drake, 2010). Y1 = yield 1; Y2 = yield 2; S = shift; TS = total suggestibility scores on the Gudjonsson Suggestibility Scale (GSS; Gudjonsson, 1984). For the linear model df = 1, 128; for the quadratic model df = 2, 127.
dependent variable) as well as the results of the F-test of model fit. When it comes to yield 2, shift and total suggestibility scores, adding the x2 parameter to the model does not increase the variance explained by iNLE scores. The linear model provides the most satisfactory account of the data. For yield 2, b = .109; for shift scores, b = .123 and for total suggestibility, b = .175. With yield 1 scores (see Table 2), iNLE only explains 4.8% (x) and 7.2% (x2) of the variance in yield 1 scores. The proportion of variance explained by the quadratic model though is greater than that explained by the linear model. For a quadratic regression model: y = c + bx + bx2 + e, where x = iNLE scores. Therefore: yield 1 scores (y) = 2.63 + .143 ⁄ iNLE .002 ⁄ iNLE2 (the negative sign indicates that after a score of 35.8, as iNLE increases yield 1 scores actually begin to decrease) (see Fig. 1a). When inspecting the residual plot no pattern is observed, suggesting the quadratic model is acceptable in the sense that the residuals are independent of the fit values. The Mahalanobis distance of the data point contributing to this pattern is 110.84 showing that this value is an outlier; however, when its Cooks distance is inspected (.042; less than 1), it appears that this value is having no undue influence upon the results as a whole. Whilst the x2 parameter only explains a very small proportion of the variance in yield 1, there is a theoretical explanation that may account for its presence suggesting that it may well be premature to exclude the data value: The curvilinear pattern to the relationship may stem from an interaction between the experience of life adversity and intrinsic stress-sensitivity levels within interviewees (see Belsky et al., 2007): during the first round of interview questions (giving rise to yield 1 scores), a proportion of individuals may not perceive the situation as stressful – such individuals score relatively low on trait stress-reactivity so are relatively unaffected by the mere presence of the interviewer and the questions (despite their experience of high levels of adversity; they may experience adversity but remain relatively unaffected by it). Once negative feedback is delivered though, at the end of the first round of questions, this triggers a reaction such that those interviewees’ yield 2 scores increase. Those scoring moderately highly on iNLE (below 35.8), yet relatively high on yield 1, would be expected to score high on stress-reactivity. At the start of the interview these individuals are already open to external influence – the mere presence of the interviewer and the prospect of being questioned may be perceived as relatively stressful; heightened uncertainty and expectations of success might be experienced even during the first round of questions (Gudjonsson & Clark, 1986). Finally, the proportion of interviewees scoring low on iNLE and low on yield 1 also do not perceive the interview situation as stressful owing to low levels of stress-sensitivity contributing to a lesser susceptibility to uncertainty and misleading information (Gudjonsson & Clark, 1986). An empirically based explanation for the data does exist that now needs to be tested, because high scores of iNLEs may not necessarily lead to an increased risk of accepting misleading information during questioning as once presumed (see Drake, 2011).
1058
K.E. Drake / Personality and Individual Differences 51 (2011) 1056–1058
Fig. 1. Curve fit graphs to show the fit of both the linear and quadratic regression models to the observed values: yield 1 (a), yield 2 (b), shift (c) and total suggestibility scores (d).
References Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2007). For better and for worse: Differential susceptibility to environmental influences. Current Directions in Psychological Science, 36, 300–304. Drake, K. E. (2010). The psychology of interrogative suggestibility: A vulnerability during interview. Personality and Individual Differences, 49, 683–688. doi: 10.1016/j.paid.2010.06.005. Drake, K. E. (2011). Why might innocents make false confessions? The Psychologist, 24, 2–5. Flouri, E., Tzavidis, N., & Kallis, C. (2010). Area and family effects on the psychopathology of the Millennium Cohort Study children and their older siblings. The Journal of Child Psychology and Psychiatry, 51, 152– 161.
Gudjonsson, G. H. (1984). A new scale of interrogative suggestibility. Personality and Individual Differences, 5, 303–314. Gudjonsson, G. H. (1997). The Gudjonsson Suggestibility Scales Manual. Hove: Psychology Press. Gudjonsson, G. H. (2003). The psychology of interrogations and confessions: A handbook. Chichester: Wiley. Gudjonsson, G. H., & Clark, N. K. (1986). Suggestibility in police interrogation: A social psychological model. Social Behaviour, 1, 83–104. Norbeck, J. S. (1984). Modification of life event questionnaires for use with female respondents. Research in Nursing Health, 7, 61–71. Pedhazur, E. J. (1982). Multiple regression in behavioral research: Explanation and prediction. New York: Holt, Rinehart, and Winston. Sarason, I., Johnson, J., & Siegel, J. (1978). Assessing the impact of life changes: Development of the life experiences survey. Journal of Consulting and Clinical Psychology, 46, 932–946.
Contents lists available at SciVerse ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Short Communication
Further insights into the relationship between the experience of life adversity and interrogative suggestibility Kim E. Drake Department of Psychology, University of Derby, Kedleston Road, DE22 1GB, UK
a r t i c l e
i n f o
Article history: Received 18 June 2011 Received in revised form 11 August 2011 Accepted 17 August 2011 Available online 15 September 2011 Keywords: Interrogative suggestibility Gudjonsson Suggestibility Scale Life adversity
a b s t r a c t Research over the past few years has uncovered a consistent link between the reporting of intense negative life events and interrogative suggestibility on the Gudjonsson Suggestibility Scale (GSS). This study (N = 130) now investigates the functional form of this relationship. Findings show that a linear regression model provides the most acceptable fit to the observed values for yield 1, yield 2, shift and total suggestibility scores. However, results also show the presence of non linear, quadratic, aspects to the relationship between the experience of life adversity and yield 1 scores. Up until now GSS scores were thought to increase linearly with level of adversity experienced: this study provides some evidence that, for the yield 1 subscale of the GSS, this may not be the case – high levels of adversity may not necessarily lead to an increased risk of accepting misleading information during questioning as once presumed. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Interrogative suggestibility is a serious psychological vulnerability during police questioning that can manifest as the acceptance of misleading information, as a consequence of the use of leading questions, and/or answer-shifts in response to explicit pressure (in the form of critical feedback) during interview (see Drake, 2011; Gudjonsson, 2003). Research over the past few years has consistently uncovered a significant correlation between the reporting of intense negative life events (iNLEs) and interrogative suggestibility on the GSS (see Drake, 2010). The reported experience of intense negative life events appeared particularly related to answer shifts in response to negative feedback. A limitation within these prior studies however is that they assume a linear relationship without investigating the presence of non-linear effects, which can manifest as either acceleration or a levelling off of interrogative suggestibility levels at certain scores of reported life adversity experience. The aim of this brief report is to therefore explore the functional form of the relationship between the reported experience of iNLE and interrogative suggestibility. Unless these effects are explored, the work will remain of limited value to both researchers and practitioners (Flouri, Tzavidis, & Kallis, 2010). Theory would suggest a linear effect, whereby increments in scores of iNLE correspond with a steady additive effect on interrogative suggestibility scores (Gudjonsson & Clark, 1986). The greater the perceived intensity of the negative life events experienced (shown by such interviewees reporting their negative experiences E-mail address: [email protected] 0191-8869/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.paid.2011.08.017
are more intensely negative), the more negative/pessimistic an interviewee’s mindset (see Drake, 2010); during interview, when being questioned and especially when given negative feedback, such interviewees are more likely to interpret the interview situation itself as well as the explicit feedback as particularly negative. This may lead to a greater susceptibility to that feedback and any pressure associated with questioning and interrogative suggestibility scores on the GSS (see Belsky, Bakermans-Kranenburg, & van Ijzendoorn, 2007; Gudjonsson & Clark, 1986). It is therefore proposed that a linear regression model would provide the most acceptable fit to the data when predicting yield 1, yield 2, shift and total suggestibility scores based on scores of iNLEs. 2. Method 2.1. Participants The sample consisted of 130 participants, 92 females and 38 males (mean age = 19.35 years, standard deviation = 1.41, range = 18–26). Participants are an opportunity sample, recruited through the experimental participation scheme within the School of Psychology. All were undergraduates within the School. 2.2. Procedure 2.2.1. The Gudjonsson Suggestibility Scale 1 (GSS1; Gudjonsson, 1984; Gudjonsson, 1997) To measure interrogative suggestibility levels, participants were administered the Gudjonsson Suggestibility Scale 1 (GSS; Gudjonsson, 1984) individually by an interviewer, trained by a Chartered
1057
K.E. Drake / Personality and Individual Differences 51 (2011) 1056–1058
Forensic Psychologist in the administration of the GSS. The GSS is a semi structured interview procedure and is presented to each participant individually as a memory task. Participants are read a narrative describing a fictitious robbery, followed immediately by a free-recall phase and the distracter phase (where participants completed the Life Events Questionnaire [LEQ; Norbeck, 1984]). An interview on the content of the narrative follows. Within the GSS interview there are 20 questions. Responses to the first round of 15 leading questions (out of a total of 20 questions) provide the yield 1 score. This indicates the number of leading questions yielded to during the first round of questions. Immediately after this, negative feedback is given by the interviewer. The interviewee is told ‘‘you have made a number of errors, and it is therefore necessary to go through all of the questions once more and this time try to be more accurate’’. The 20 questions are then repeated to see how readily interviewees shift their initial answers as a result of the negative feedback received; a ‘‘shift’’ score is derived. A yield 2 score is also obtained (denoting the number of leading questions yielded to after negative feedback). 2.2.2. Life Events Questionnaire (LEQ) (Norbeck, 1984) The LEQ contains 82 items in total and is a modification of the instrument developed by Sarason, Johnson, and Siegel (1978) in that it has nine items of particular relevance to women. These include items such as ‘‘Major difficulties with birth control pills or devices’’. The nine additional items in the LEQ were introduced to reduce the gender bias in the Sarason et al. (1978) version. Participants were required to go through all the events listed, and if they had experienced them at any point of their life, to circle whether it had been a ‘‘good’’ experience or ‘‘bad’’ experience. They were then instructed to rate the extent to which those events had an effect on their lives at the time. The ratings went from 0 (‘‘no effect’’) to 3 (‘‘large effect’’). The LEQ has good test–retest reliability, with test–retest reliabilities of 0.78–0.83 and is a significant predictor of measures of (unfavourable) psychological and psychiatric symptoms. 3. Results and discussion 3.1. Descriptive statistics Table 1 presents the means and standard deviations for the observed variables. 3.2. Curve-fit estimations (Pedhazur, 1982) Linear models were fitted (and compared with quadratic regression models) to investigate the presence of linearity. Table 2 shows the R-square statistic of relationship strength (between the observed and model-predicted values of the
Table 1 Means (M) and standard deviations (SD) of observed variables (N = 130).
IR Y1 Y2 S TS iNLE
M
SD
14.1 4.45 5.55 3.74 8.19 17.9
5.69 2.32 2.81 2.56 3.38 9.74
Note: IR = immediate-recall; DR = delayed recall; Y1 = yield 1; Y2 = yield 2; S = shift; TS = total suggestibility; iNLE = intensity of negative life events.
Table 2 Linear (x) and quadratic (x2) regression model of the reported intensity rating of negative life events experienced on GSS scores (N = 130). GSS Y1 R2 x x2
Y2 F
Sig.
R2
S F
Sig.
R2
TS F
Sig.
R2
F
Sig.
.048 6.43 .012 .144 21.5 .000 .220 36.0 .000 .256 44.0 .000 .072 4.94 .009 .144 10.7 .000 .220 17.9 .000 .264 22.8 .000
Note: x = the reported intensity rating of negative life events experienced (see Drake, 2010). Y1 = yield 1; Y2 = yield 2; S = shift; TS = total suggestibility scores on the Gudjonsson Suggestibility Scale (GSS; Gudjonsson, 1984). For the linear model df = 1, 128; for the quadratic model df = 2, 127.
dependent variable) as well as the results of the F-test of model fit. When it comes to yield 2, shift and total suggestibility scores, adding the x2 parameter to the model does not increase the variance explained by iNLE scores. The linear model provides the most satisfactory account of the data. For yield 2, b = .109; for shift scores, b = .123 and for total suggestibility, b = .175. With yield 1 scores (see Table 2), iNLE only explains 4.8% (x) and 7.2% (x2) of the variance in yield 1 scores. The proportion of variance explained by the quadratic model though is greater than that explained by the linear model. For a quadratic regression model: y = c + bx + bx2 + e, where x = iNLE scores. Therefore: yield 1 scores (y) = 2.63 + .143 ⁄ iNLE .002 ⁄ iNLE2 (the negative sign indicates that after a score of 35.8, as iNLE increases yield 1 scores actually begin to decrease) (see Fig. 1a). When inspecting the residual plot no pattern is observed, suggesting the quadratic model is acceptable in the sense that the residuals are independent of the fit values. The Mahalanobis distance of the data point contributing to this pattern is 110.84 showing that this value is an outlier; however, when its Cooks distance is inspected (.042; less than 1), it appears that this value is having no undue influence upon the results as a whole. Whilst the x2 parameter only explains a very small proportion of the variance in yield 1, there is a theoretical explanation that may account for its presence suggesting that it may well be premature to exclude the data value: The curvilinear pattern to the relationship may stem from an interaction between the experience of life adversity and intrinsic stress-sensitivity levels within interviewees (see Belsky et al., 2007): during the first round of interview questions (giving rise to yield 1 scores), a proportion of individuals may not perceive the situation as stressful – such individuals score relatively low on trait stress-reactivity so are relatively unaffected by the mere presence of the interviewer and the questions (despite their experience of high levels of adversity; they may experience adversity but remain relatively unaffected by it). Once negative feedback is delivered though, at the end of the first round of questions, this triggers a reaction such that those interviewees’ yield 2 scores increase. Those scoring moderately highly on iNLE (below 35.8), yet relatively high on yield 1, would be expected to score high on stress-reactivity. At the start of the interview these individuals are already open to external influence – the mere presence of the interviewer and the prospect of being questioned may be perceived as relatively stressful; heightened uncertainty and expectations of success might be experienced even during the first round of questions (Gudjonsson & Clark, 1986). Finally, the proportion of interviewees scoring low on iNLE and low on yield 1 also do not perceive the interview situation as stressful owing to low levels of stress-sensitivity contributing to a lesser susceptibility to uncertainty and misleading information (Gudjonsson & Clark, 1986). An empirically based explanation for the data does exist that now needs to be tested, because high scores of iNLEs may not necessarily lead to an increased risk of accepting misleading information during questioning as once presumed (see Drake, 2011).
1058
K.E. Drake / Personality and Individual Differences 51 (2011) 1056–1058
Fig. 1. Curve fit graphs to show the fit of both the linear and quadratic regression models to the observed values: yield 1 (a), yield 2 (b), shift (c) and total suggestibility scores (d).
References Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H. (2007). For better and for worse: Differential susceptibility to environmental influences. Current Directions in Psychological Science, 36, 300–304. Drake, K. E. (2010). The psychology of interrogative suggestibility: A vulnerability during interview. Personality and Individual Differences, 49, 683–688. doi: 10.1016/j.paid.2010.06.005. Drake, K. E. (2011). Why might innocents make false confessions? The Psychologist, 24, 2–5. Flouri, E., Tzavidis, N., & Kallis, C. (2010). Area and family effects on the psychopathology of the Millennium Cohort Study children and their older siblings. The Journal of Child Psychology and Psychiatry, 51, 152– 161.
Gudjonsson, G. H. (1984). A new scale of interrogative suggestibility. Personality and Individual Differences, 5, 303–314. Gudjonsson, G. H. (1997). The Gudjonsson Suggestibility Scales Manual. Hove: Psychology Press. Gudjonsson, G. H. (2003). The psychology of interrogations and confessions: A handbook. Chichester: Wiley. Gudjonsson, G. H., & Clark, N. K. (1986). Suggestibility in police interrogation: A social psychological model. Social Behaviour, 1, 83–104. Norbeck, J. S. (1984). Modification of life event questionnaires for use with female respondents. Research in Nursing Health, 7, 61–71. Pedhazur, E. J. (1982). Multiple regression in behavioral research: Explanation and prediction. New York: Holt, Rinehart, and Winston. Sarason, I., Johnson, J., & Siegel, J. (1978). Assessing the impact of life changes: Development of the life experiences survey. Journal of Consulting and Clinical Psychology, 46, 932–946.