Faking good on the MCMI-III and MCMI-IV
Date of Award
Doctor of Philosophy
School of Arts and Humanities
Dr Greg Dear
Professor Alfred Allan
Dr Ricks Allan
In certain situations respondents completing self-report psychological tests are motivated to present themselves in the best possible light and attempt to alter their responses accordingly (termed faking good). Although the Millon Clinical Multiaxial Inventory - third edition (MCMI-III; Millon, 1994) possesses a scale (Y-Desirability) that can alert the clinician to the probability that a respondent has made an attempt to fake good, there remains controversy surrounding the use of this test, especially in high stakes contexts. When respondents fake good, the Y-Desirability scale elevates but there is a tendency for three Clinical Personality Pattern scales (4-Histrionic, 5-Narcissistic, and 7-Compulsive scales) to also elevate for reasons other than that the test taker has a high level of histrionic, narcissistic, or compulsive traits. An elevation on those four scales has been dubbed the normal quartet (Halon, 2001). It is left to the clinician to estimate the degree to which the elevations on these scales are actually indicative of psychopathology, despite the elevation on the Y-Desirability scale suggesting otherwise. The risk of incorrect conclusions being drawn is unacceptable, particularly given the stakes (e.g. child custody). This research project was conducted to help researchers and clinicians understand why elevations are found on clinical scales designed to measure psychopathology, despite a respondent‟s attempt to fake good.
Four studies were conducted. The central purpose of Study 1, an exploratory study, was to determine if there was a generic fake good profile that applied across contexts in which one might be motivated to fake good. Undergraduate University psychology students were randomly assigned to four analogue conditions (n = 16 per condition) and required to role-play different fake good scenarios (i.e. Good Parent; Good Person; Healthy Psychiatric Patient; Healthy Drug User). Results demonstrated that there were far more similarities than differences between participants in all four conditions, with all conditions obtaining a version of the normal quartet (elevations on Y-Desirability, 4-Histronic, 5-Narcissistic, and 7-Compulsive scales and with very low scores on most other scales). Differences that were found were either not clinically meaningful with very low base rate (BR) scores or possibly attributable to unequal difficulty in the instructions used in two of the conditions. Study 2 was conducted as a manipulation check to determine if the instructions were perceived to be a confounding variable in Study 1. Participants were 20 adults who were supplied all four instruction sets from Study 1 and asked how difficult they believed it would have been to follow each individual role. Additionally, they were required to rank from one to four the easiest to hardest instructional set. The conclusion drawn from Study 1 was confirmed that some instructions (primarily for the Healthy Psychiatric Patient condition) were perceived to be significantly harder than others for participants to role-play. The strongest conclusion that could be drawn from Study 1 was that regardless of context, even with different instructions and difficulties, when respondents were required to fake good, they generally obtained a fake good profile that was an exaggerated version of the normal quartet.
Study 3 was conducted to answer the questions raised at the end of Study 1; that is, why are respondents elevating on scales designed to measure psychopathology, despite instructions to fake good, and what strategies were they using that might be contributing to these elevations? A sample of parents (n = 10) were administered the MCMI-III and required to answer out loud as if they were taking the test as part of a Family Court assessment and how they respond will help determine with whom the child/ren will reside. The participants were asked to report why they answered True or False to each item and how important it was to answer in that way. The data were explored both quantitatively and qualitatively. Results illustrated that participants used a variety of strategies ranging from a single theme of denial of perceived pathology to more sophisticated strategies by assessing each item carefully and even promoting minor negative attributes. Despite the faking good strategy adopted, most participants interpreted a number of items on the 4-Histrionic, 5-Narcissistic, and 7-Compulsive scales as positive attributes (mostly true-keyed items) that they willingly endorsed, at times rating it important to do so if one wants to look like a good parent. Participants answered False and rated it important to do so for most of the test items with the intention of hiding perceived pathology in line with the fake good instructions. However, by answering False, participants inadvertently scored heavily on these items across the 4-Histrionic, 5-Narcissistic, and 7-Compulsive scales given the disproportionately greater number of false-keyed items found on these three scales. Study 4 was developed after the completion of Studies 1 to 3 due to the release of the new Millon Clinical Multiaxial Inventory – fourth edition (MCMI-IV; Millon, Grossman, & Millon, 2015). The purpose of this exploratory analysis was to determine if the same issues revealed with the MCMI-III in high-stakes settings were still present in the revised edition. A replication of the Lenny and Dear (2009) study was undertaken. Lenny and Dear used the MCMI-III results from participants in a fake good parent condition and compared them against an Honest condition. The authors concluded that the normal quartet obtained by those in the Fake Good condition was more likely due to factors other than psychopathology. For Study 4, MCMI-IV results were obtained from 60 parents that were randomly assigned to either the Fake Good condition (n = 30) or the Honest condition (n = 30). The Honest condition participants were asked to answer the test in an honest manner and their final BR results were compared to that of the Fake Good condition participants. The Fake Good participants were required to answer the test items as if they were undertaking a Family Court assessment and that how they answer will help determine with whom the child/ren will reside. A normal quartet was still present in the Fake Good condition but now comprised BR score elevations on the revised Y-Desirability, 4A-Histrionic, 7-Compulsive scales, the new 4B-Turbulent scale and with most other scales very low. A borderline subclinical elevation on the 5-Narcissistic scale was also obtained, which was an improvement from the MCMI-III results. The new 4B-Turbulent scale was clinically elevated when participants attempted to fake good on the MCMI-IV and in this study was the highest elevation of a normal quartet.
The most important findings from the four studies in this research project were (1) that the normal quartet elevations were likely due to psychometric issues with the test design combined with the way respondents interpreted items when instructed to fake good on both the MCMI-III and MCMI-IV; (2) elevations on the MCMI-III were directly linked to participants actively endorsing items they misperceived as positive attributes, particularly on true-keyed items, and attempting to deny or hide perceived pathology but inadvertently scoring on the false-keyed items; and (3) when using the MCMI-IV, the BR score profile will likely be similar to that of the MCMI-III in genuine high-stakes settings, such as child custody evaluations, with the addition of an elevated 4B-Turbulent scale. Qualitative findings from Study 3 with the MCMI-III are likely to be transferable to the MCMI-IV and explain the significant elevations on the equivalent scales and new scale that was seen in Study 4. Given the significant BR score elevations on both the MCMI-III and MCMI-IV in simulated high stakes contexts, which closely resemble results from real-world examples, serious caution is warranted when using either edition in high stake contexts. The final conclusion from this research project is that the MCMI-III and MCMI-IV should not be used in high-stakes settings until further research is conducted.
Access to this thesis –the full text is not available by author's request. Email request to firstname.lastname@example.org
Lenny, P. (2017). Faking good on the MCMI-III and MCMI-IV. Retrieved from http://ro.ecu.edu.au/theses/1969