Note - The investigation that is reported in the document below was originally carried out in 1988 and was at that time given some very limited distribution within a single governmental agency. The author of this document has recently been encouraged by several colleagues that the general subject material, what is represented in the document, is not found in most of the so-called ‘open literature’ [i.e., non-classified]. Very little research pertaining to study of the factors and characteristics involved in the adjudication process of psychological evaluation of those being considered for high-level security clearances has been actually carried out; almost none has been reported in the ‘open literature.’ There is also no reason to believe that more recent edition versions of the two psychological tests, that were employed, would have produced different results than what are herein described. 
 

     Security clearances, as granted by the US. Government, are of several different types and levels. Although most persons are rather familiar with the concepts, secret and top-secret clearances, very few have ever encountered the Sensitive Compartmented Information (SCI) clearance concept. SCI access is, for all practical purposes, the highest major security clearance level granted to any fairly large group of people by the government. The decision to grant access to the SCI clearance level is determined by guidelines specified in a Directive issued by the Director of Central Intelligence which is best known as the DCID 1/14 (see Director of Central Intelligence, 1984**). Although such information may be very difficult to estimate and obtain, Washington Post reporters (see the June 8, 1986 issue, A Section, page 18) have stated that as of March, 1985 there were 98,715 persons (civilian and military) within the Department of Defense (DoD) and another 9,576 in industry who held SCI access. In order to obtain SCI access, individuals may/must undergo what might be considered as extremely rigorous screening or vetting which employs psychological testings and interviews, polygraph examinations, extensive and depthy background investigations and records checks. Reinvestigations are supposedly routinely conducted every five years for those who hold this particular clearance. 
                                                                                                                                                                         2 
     Although the psychological testing instruments which may be involved in evaluating persons as part of the clearance screening/vetting process may differ from one agency or organization to another, there are at least two popular personality measuring tests which have been regarded as appropriate for this particular kind of purpose or situation. Historically, the Minnesota Multiphasic Personality Inventory (MMPI) immediately comes to mind as perhaps useful for this type of assessment situation. In more recent years, the Millon Multiaxial Inventory (MCMI) is another testing instrument which has had some use in similar type assessment situations. The question of whether these two very popular, frequently utilized, personality-measuring instruments do provide psychometric information that can be used in prediction of whether eventual high-level security clearances (SCI access in particular) will or will not be subsequently granted was the raison d’être for this investigation. The research reported in this paper is of the kind that was recently suggested, by a Congressional Committee (House of Representatives, 1987), be conducted. They urged that "research attention [be given to] . . . personality-based profiling . . . as an element to be considered in the screening process [pages 7-8]." 

     The MMPI was administered to a sample of 102 contractor employees (these were individuals who were employees of companies/corporations which were providing services to the DoD) who were being psychologically evaluated and processed for possible granting of SCI access; 61 were eventually and finally successful in obtaining SCI access, 41 were not. In this particular sample of 102, their ages ranged from 22 to 60 years (M = 36, SD = 10), 17% were female and 83 were male, average education in years was 15.5 years (SD = 2.7). Approximately, 25-30 different contractor firms were represented as sponsors for this sample. Most of the employees were from either the West or East Coast regions. Based upon their MMPI clinical scales scores, this sample appeared to be rather representative of the general population as their MMPI scores seemed to be rather close to recognized recent norm values (e.g., see Colligan, Osborne, Swenson & Offord, 1983). This particular sample can be regarded as being somewhat unusual in that the rate for non-obtaining access/clearance can be considered to be unusually high. However, this sample can be regarded as well-suited for a validity testing of the MMPI with respect to the question of eventually obtaining/non-obtaining SCI access. 

     In actuality, the criterion or validity variables were three in number; the first of the three was whether the interviewing clinical psychologists gave positive or negative recommendations for the individuals involved. The second criterion variable was whether the chief of the clinical psychologists also gave positive or negative recommendations following his review of all pertinent information. Only in a couple of cases did the chief psychologist offer a different recommendation than did the interviewing psychologists; in all such cases he made a negative recommendation whereas the interviewing psychologists had given positive recommendations. Also, it should be noted that the only real differences between the second criterion variable and 

                                                                                                                                                                      3 
                                                                                                                                                                                                                    the third (which, in fact, was the final adjudication decision, made by the Office of Security, as to whether the individual would or would not be granted SCI access), were that the decisions made in the third were generally more stringent. Correlation coefficients between all three criterion variables were high. The correlation coefficient for between the interviewing psychologists’ recommendations and the chief psychologist’s review recommendations was 0.96; between the interviewing psychologists’ recommendations and whether SCI access was granted or not was 0.836; and between the chief psychologist’s review recommendations and whether SCI access was granted was 0.837. It should be noted that none of the interviewing clinical psychologists (nor was anyone else in the investigation/adjudication chain) were aware that this presently reported investigation was being conducted while the date were being collected. 

     To analyze the mean differences between the contractor employees judged Favorable versus Unfavorable by the interviewing psychologist, two-tailed t -tests were calculated, based upon T-scores from the basic MMPI scales. The results of these two comparisons revealed that only for the K, Hs, and the Sc scales were the differences between means perhaps even close to being statistically significant; these three differences between means could only be regarded as being significant at the 0.10 level and not significant at the 0.05 level. Also, most of the differences were in what might be considered as an unusual direction; higher psychopathology was indicated with the favorably judged group than was observed for the unfavorably judged group with two of the three scales. Such a directional difference is exactly opposite to what one would predict. It seems reasonably safe to conclude that al of the tested differences could have easily occurred, simply on the basis of chance alone. 

     To analyze the differences between means with respect to the contractor employee judged Favorable versus Unfavorable by the chief psychologist upon case review, two-tailed t-tests were calculated based upon the T-scores from the basis MMPI scales. The results of these comparisons reveal that only for the K, Hs, and the Si scales were the differences between means perhaps even close to being statistically significant. The difference between Si means could only be regarded as significant at the 0.10 probability level, whereas with the K and the Hs scales the differences could be regarded a significant at the 0.05 level but not at the 0.02 level. Again, only with the Si scale difference did the direction of the difference make any sense, very lightly more psychopathology indicated for the Unfavorable group than with the Favorable group. Again, it seems reasonably safe to conclude that all of the tested differences could have most likely easily occurred, simply on the basis of chance alone. 

     To analyze the differences between means of the contractor employees judged Favorable versus Unfavorable by the complete SCI access adjudication process, two-tailed t-tests were calculated, based on the scores from the basic MMPI scales. Inspection of the computed t-tests revealed that only for the Sc scale was the difference between means perhaps even close to being statistically significant; the difference could 

                                                                                                                                                                       4 

only be regarded as being significant at the 0.10 probability level. Since this one noted difference was just one of the 13 t-tests computed with these data, then the most likely explanation for the size of the difference was that it occurred simply on the basis of chance alone. 

     It should be noted that the t-test statistic was not the only one used to explore whether MMPI scores were systematically related to the three criterion variables. Multiple (two-group) discriminant functions were also computed with each of the criterion variables and included all 13 studied MMPI scale variables. All three discriminant functions were found to be statistically non-significant (i.e., p-levels were not even remotely close to any frequently utilized alpha levels). 

     It seems rather safe to conclude that any differences noted between means (for the Favorable versus Unfavorable groupings) were simply based only on chance alone. In other words, there would appear to be no linear association noted on the MMPI with respect to whether contractor employees, being psychologically evaluated and finally evaluated by SCI access adjudication process, were later regarded favorably or unfavorably in the psychological evaluations or more finally by the Office of Security. Psychopathology, as measured by the MMP, seems to be independent of psychological evaluations (at least with contractor employees) and of SCI adjudication decisions. 
 

    The MCMI was administered to a sample of 117 male contractor employees who were being psychologically processed for being possibly grant SCI access. With the exception that this sample was composed of all males, it resembled very closely the sample that was employed in Study I, which was just reported. The rate for successfully obtaining SCI access was however a bit higher for this group than noted in Study I. With this sample of 117 male employee, 88 (or 75%) were successful in finally obtaining SCI access status. 

    Exactly the same statistical analysis procedures that were used in Study I were also utilized here in Study II. It was decided that separate analyses for the evaluating psychologists’ recommendations were not to be carried out, but rather only the final SCI access adjudication decision would be regarded as the single, only criterion variable. To analyze the differences between MCMI scales means of the contractor employees judged Favorable (n = 88) versus Unfavorable (n = 29) by the complete SCI access adjudication process, tow-tailed t-tests were calculated, based on the scores from the 20 MCMI scales. Inspection of the computed t-tests revealed not one of them being significant at any generally accepted alpha level setting. The largest t value noted was 1.76 and, with 115 degrees of freedom, this was not significant at the 0.05 level. A two-group discriminant function was also computed and a ‘less than unity’ F-test value (dfs = 20 and 96) was obtained, indicating that there was no systematic or reliable relationship between the MCMI data and the two-group adjudication decisions regarding the granting of SCI access. 

                                                                                                                                                                       5

     It is interesting to note that neither the MMPI nor the MCMI result in psychometric information which is predictively associated with success or failure in obtaining SCI access status, at least not with the defense contractor employees in the two samples studied. Although there has not been a great deal of reported research dealing with this particular kind of problem, these just reported findings are actually consistent with the results from a couple earlier reported investigations. About a decade and a half ago, two psychologists (Muha & May, 1972) conducted a rather similar investigation; they studied the relationship of MMPI scales scores (and some other demographic and psychometric information) of applicants for employment with a governmental intelligence agency and they essentially found the same kind of results. Namely, with male applicants (and it should be noted that in the currently reported ‘Study I,’ 83%of the studied contractor employee sample were males) they found that only two MMPI scales (the F and K scales) might be associated with interviewing psychologists’ favorable-unfavorable recommendations. They concluded that "there were no substantive mean differences between Fit and Unfit male Ss on the clinical MMPI scales [p. 61]." 

     More recently, a study (Hibler, Kolski, & Chapman, 1988) was reported which also had some findings that were perhaps similar in fact to the presently reported findings but which appear to have been confusingly interpreted by the principal investigator of that study. They reported (based on a study of 122 active-duty military members who were involved with SCI access) that "a comparison of individuals with elevated MMPI clinical scales (>70 T) to those with non-elevated MMPI clinical scales, suggested that elevated MMPIs do not differentiate those receiving a recommendation to deny access from those retaining access [page 2]." This kind of finding seemed to be about equally true with their MCMI data scores information. Rather surprisingly, they then concluded, based on their research data pertaining to the matter of maintaining or obtaining SCI access, that "the objective personality tests, in particular, were most useful for identifying significant pathology [page 5]." In actuality, their results rather cleanly suggested just the opposite conclusion. 

     With all these findings (including those presently reported), it would seem that it would be most unwise to attempt to build any type of paradigm or scheme for predicting whether applicants or candidates would be later favorably or unfavorably adjudicated for SCI access and base such a paradigm on MMPI or MCMI scales score information. What appears to be the major problem with using this kind of particular psychometric information is perhaps due to response set or style problems and difficulties. The investigator is very familiar with this kind of psychometric information used as part of psychological evaluations conducted in the SCI access adjudication process. It has been repeatedly observed that some rather potent personality-test types of response styles very frequently and significantly influence responses of applicants/candidates who are being considered for possible SCI access. It has been observed that a good percentage of such 

                                                                                                                                                                                 6 

individuals seemingly attempt to minimize psychopathology reporting and this kind of minimization seems to be relatively independent of true psychopathology. Also, it is not at all unusual to encounter other individuals who attempt to be ‘more-than-candid’ or to be more than usually honest in their self-reporting of psychopathology and its possibly related behavioral correlates. For example, it is not unusual to find that some individuals, who are sometimes noted to be associated with particular religious faiths, tend to exhibit a more-than-usually-encountered level of honesty in their own self-descriptions (particularly on psychometric inventories/questionnaires). This type of unexpected degree of honesty, with respect to self-description, seems to also be somewhat unrelated to actual or true psychopathology. Because of these kinds of sources of measurement error, it becomes very difficult to conclude that low scores on a MMPI or MCMI type instrument actually reflect low levels of psychopathology or that high scores on these kinds of instruments actually reflect high levels of psychopathology. 

     It should be noted that the research results described in this reported investigation should not be interpreted to mean that the MMPI or the MCMI should not be clinically used as psychological tests in the psychological evaluation process involved in SCI access adjudication matters. The results though do indicate that MMPI and MCMI scores, based upon the instruments’ basic or major scales, should not be used in any fashion which would suggest that they were the sole or even major basis for any favorable or adverse determination and/or for prediction purposes in the SCI adjudication process. 
 

Colligan, R. C., Osborne, D., Swenson, W. M., & Offord, K. P. (1983). The MMPI: A contemporary normative study. New York: Praeger 

Director of Central Intelligence (1984). Director of Central Intelligence Directive No. 1/14. Washington, DC: Author. 

Hibler, R. J., Kolski, A. R., & Chapman, R. K. (1988). Pathological indices in personnel with high security clearances. Paper read at the 12th Symposium, Psychology in the Department of Defense. Colorado Springs, Colorado. 

House of Representatives (1987). Report by the Permanent Select Committee on Intelligence. (Report 100-5). Washington, DC: U.S. Government Printing Office. 

Muha, T. M., & May, J. R. (1972). II. An employment index for identifying unfit job applicants. National Security Agency Studies in Psychological Assessment, 1-9. 
 

*Since this presently reported research was conducted/described back in the very late 1980s, the DCID 1/14 documentation has undergone a couple updating revisions. However, any changes in the revisions have mainly been with a ‘liberalizing’ of the standards or requirements criteria for the granting of SCI access. There is seen no reason to suspect that were this reported investigation carried out today, rather than back in the late 1980s, that the results obtained would be much different. Some slight changes in the SCI access adjudication standards would seem to not affect the conclusions that have been drawn from the presently reported results. 

**Dr. Stone’s mailing address is: P.O. Box 395, Harpers Ferry, WV 25425. His email 
address is: lastone@intrepid.net and his Web Site can be found at: http:www.home.earthlink.net/~lastone2/home.html.