This article is published in Psychology of Espionage Reports, Vol. II, (2001)
(this journal is located at:
A Two-Factor Motivational Theory for Spying Behavior#

LeRoy A. Stone, Ph.D., ABPP##
School of Medicine, West Virginia University

     Many attempts to describe espionage behavior go back in time as early as around 510 B.C., when it is believed that the Chinese sage, Sun Izu, wrote his still mentioned and quoted work, Ping Fa, which dealt with the arts of war and espionage (Deacon, 1987). Spying, which is defined (Merriam-Webster, 1986) as "acting in [a] clandestine manner or on false pretenses to obtain information on the zone of operations or a belligerent with the intention of communicating it to [a] hostile party;" when involving a citizen and his own country, can be considered to represent an act of treason. Treason is the only crime given special notice in the United States Constitution (e.g., Article III, Section 3). Spies, who commit espionage on their own country, traditionally have been regarded as being criminals who are understood to be rather different from those who are involved in other forms of crime. In an earlier paper, I have presented an argument that this kind of special regard for treasonous spies is not really justified by the facts and that spies should be regarded as merely being white-collar criminals who involve themselves in matters which sometimes have extremely important national security and international ramifications (Stone, 1989). However serious the consequences of their acts, such spies can be best understood as simply being motivated much like many other white-collar criminals. This seems to be especially true during the past few decades when just about all American spies, who have been discovered spying on their own country, have do so mainly for money. Attempts to develop some kind of identifying demographic and psychological profile for spies have been no more successful than have the attempts which have been made to construct such profiles for white-collar criminals. 

     The present writer, who has been engaged in attempts to study, in a scientific fashion, an extensive U.S. spy database, which was constructed for research purposes over the past several or more years, has seen only one previous attempt to build a model.** Pincher (1987), a well-known British journalist, who had for over 40 years specialized on writing about espionage, did attempt to describe a mathematically appearing model for such kind of treacherous behavior. Pincher’s model simply was the product of his thinking on the subject and, in no way, emerged from any formal quantitatively-oriented research. Another problem with Pincher’s model was that he suggested no direction for quantifying any of the model’s parameters. Even though it appears, from his writing on the subject, that Pincher did not really or truly understand the logic of factor or principal components analysis, he did attempt to describe his model using some of the ideas inherent in this kind of mathematical conceptualization. Pincher’s model was conceptually a rather complex one; he suggested that it should involve 6-7 different motivational constructs. Other than my later multivariate statistical research efforts (e.g., Stone, 1991b, 1991c), which have led to explicit suggestions for building models that could describe and explain spying behavior, I can find no other previous attempts, in the vast reaches of the espionage and spying literature, to accomplish any formal model building, except that presented by Pincher. 


     The purpose of the present paper, is to describe, in specific form, the development of a two-factor theory or model for motivations that lead to traitorous spying behaviors; I will also attempt to describe, in explicit detail, the means and logic for obtaining quantified scores or measurement on both of these factors. Prior accomplished research (which has led to this thinking can be found in two previous papers (Stone 1991b, 1991c). It should be noted that the development of the to-be-described two-factor theory was directly based upon statistical analysis of empirical quantitative data and was not the product of some intuitive, ‘off the top of the head,’ or ‘armchair’ philosophizing kinds of thinking. The theory is founded in quantitative data-based research and can be expected to be able to "fit" a large portion of the known traitorous spy population. 

Stages of Development of the Theory


     In a recent attempt to apply complex multivariate statistical procedures to an American spy database, we computed a canonical correlation between two sets of variables (Stone, 1991c). One set consisted of a four-group classification of spying motivations (i.e., money [or greed], ideology, disaffection, and other [which included such matters as blackmail, coercion based upon threat to family members, etc.]). The second set of variables consisted of 10 different security clearance adjudication matters which are normally and routinely studied during a security clearance investigation (i.e., foreign preference, security responsibility, [past] criminal conduct, mental and emotional disorders, foreign connections, financial matters, alcohol abuse, drug abuse, falsification, and sexual misconduct). A rather high and very statistically significant canonical correlation (canonical correlation coefficient = 0.70) was obtained regarding the relationship between these two sets of variables, based upon data from 100 uncovered U.S. spies. The most major canonical factor, which emerged, was seen to identify two of the spying motivations (money and ideology) as being opposing extremes seen and identified. The second most potent factor was seen to have the disaffection spying motivation as one polar extreme, with the money spying motivation somewhat associated with the opposing pole. The third canonical factor was identified with the "other" spying motivation on one pole and the money spying motivation also seen to be occupying, in a nonextreme fashion, the other bipolar position. In a rather interpretable fashion, the 10 various security clearance adjudication criteria were seen to "load" on different positions on the three extracted canonical factors. It was from this particular research that the logic for the first factor (of the proposed two-factor theory) developed. It is interesting to note that the present conceptualization of four different motivations for spying do correspond well with the espionage motivation conceptualization described by Herbig (1988). 


     The first factor for spying motivation was clearly bipolar with money (or greed) at one end and ideology at the other. This bipolar factor was the subject of a second research project that endeavored to determine whether spying motivation (money vs. ideology) could be predicted, for a large number of U.S. spies (N = 132), based on available Biographical and demographically information (Stone, 1991b). Multiple regression procedures were employed; the result being that a high level of success was obtained (i.e., R = .92) with respect to being able to predict this kind of dichotomous spy motivation classification from a knowledge of scoring on 13 different biographic/demographic measures. These measures included: gender, years of federal or national defense employment, civilian at time of being caught/identified, in Air Force at time of being caught, being Jewish, employment level, age at which spying began, birthyear, in Navy at time of being caught, year caught/identified, homosexual history, months engaged in spying, and in Army at time of being caught. Using information from these 13 measures, the matter of whether a spy was motivated by greed or by ideology could be very accurately predicted (only a two and a half percent error rate was observed when making these individual spy predictions). The Stone (1991b) investigation did lend a good deal of additional credence to the two-category, spying motivational concept by showing that measured position on it, the bipolar spying motivational scale could be reliably predicted from the weighted sum of information from a number of other very different kinds of measures/variables. A kind of construct validity belief regarding the two-category motivational concept, of the kind suggested by Cronbach and Meehl (1955), could be deduced from this discovered relationship. 

     Results from Stone (1991b, 1991c) have been discussed with personal friends and associates who are either currently or in the past have been involved as governmentally-employed counterintelligence officials. All of these particular individuals agreed with the basic idea of greed and ideology being "opposites" with respect to spying motivation matters. One individual, who has had an unusually lengthy background in counterintelligence efforts, suggested that there was even an underlying more basic consideration than simply a greed/ideology dichotomy. He conceptualized that some type of self-orientation and other-orientation was key to whether an individual spy was operating on greed or ideology demands. He suggested that this self- and other-orientation conceptualization might also be involved with those fewer number of spies who seemingly were not motivated by either greed or ideology, but were instead motivated by other causes. 

     It was this last thought, expressed by this particular counterintelligence expert, which resulted in a refinement of that kind of thinking. The opposing directions of self- and other-directed could rather nicely be applied to the two spying motivational classifications, disaffection and other. Disaffection seemingly represents a kind of self-orientation. One becomes disaffected or dissatisfied with how one’s own self has been treated by other persons, organizations or political entities. The "other" spy motivation category was intended (see the fourth paragraph in this paper) to indicate the kind of miscellaneous catch-all category that would include rather rare spying motivations based upon blackmail, coercion based upon threat to family, and the like. It was not difficult to conceptualize this "other" motivation category as being representative of some kind of other-orientation, i.e., based upon a concern for the welfare of others. The self- and other-orientations may bear some resemblance to what Rotter (1966) has suggested with his internal and external locus of control theorizing. 



     The spy database was used to construct a new dichotomously scaled variable, that of Disaffection vs. Other (representing the two different determined spying motivations); however, this new constructed variable only involved 21 spies (nine who were regarded as having spied because of being disaffected and 12 because of "other" motivations). Determination of parameters in theory building, based upon small samples of data, is always risky, but no other choice was available. When this newly created dichotomous variable, Disaffection vs. Other, was correlated with the other measurement variables in the database, a few rather moderately sized relationships were noted. For example, this new dichotomous variable had its highest correlation (r = 0.69, p < .001) with Drug Abuse; indicating that drug abuse seemed to be a historical background matter for many of those spies who had been identified as having spied because of them having been disaffected with some part of their lives associated with the U.S. Government. Another relationship observed was with Foreign Relatives (r = -0.49, p < .05); indicating that having foreign relatives was negatively associated with being disaffected but seemingly positively associated with being motivated for spying due to "other" causes. Since "other" has been explicitly defined as including such matters as blackmail and being coerced due to threats to family, this particular relationship seems quite interpretable and even expected. A somewhat unexpected relationship between the new dichotomous variable and Amount of Money Obtained for Having Spied was observed (r = 0.46, p < .05). This suggests that disaffected spies seemingly were prone to also receive larger sums of money for their spying efforts. Perhaps, also receiving money for their treachery satisfied aspects of their disaffections. Another noted relationship was with the [past] Criminal Conduct measure (r = 0.43, p < .05); seemingly, having a history of some previous criminal conduct seems to be an antecedent to being a spy because of disaffection motivations. It is suggested that perhaps a prior criminal behavior history may be caused by previous proneness to behave generally in a criminal fashion as a function of being disaffected. The "Other" pole direction of this new dichotomously scaled variable was also seen to be related to Sexual Misconduct (r = -0. 41, p < .10); this is exactly as would be expected knowing that a "blackmail" spying motivation was one of the inclusions in this "catch-all" motivational classification. 



     Utilizing the same statistical procedures that were employed in Stone (1991b), we computed a multiple correlation between some selected biographical variables (in the spy database) and the above-described newly-created dichotomous variable, Disaffected vs. Other. The final multiple correlation only involved three predictor variables (Criminal Conduct, Drug Abuse, and Sexual Misconduct [all dichotomously scored security clearance adjudication variables]) and was sizable enough to be regarded as being statistically significant (R = 0.82, df = 3 and 9 [dfs were based on the number of spies with complete data], F = 5.66, p < .02). For 13 of the 21 spies (who had measurement scores on the dichotomous dependent variable) there was complete enough data to predict their position on this two-point variable, measured as 1.0 and 2.0. Using 1.5 as a cutoff point, correct classification was obtained 85% of the time using the prediction regression equation. Only two of the 13 were incorrectly classified on this dichotomous measure; these two were Cpl. Robert L. Johnson and William P. Kampiles, both of whom have been perceived (in the spy literature) as having been disaffected with their own country this being their primary motivation for spying. Other spies (e.g., SSgt. Allen J. Davies, Edward L. Howard, and Douglas R. Bacon), who have been regarded as having spied also because of being disaffected, were correctly predicted (each had the same predicted score, 2.2). All eight spies (Lt. Christopher Cooke, MSgt. Walter T. Perkins, Petty Officer Brian P. Horton, Joseph S. Peterson, Victor N. Hamilton, Sharon M. Scrange, Ruby L. Schuler-Harper, and Barbara Walker***) who had scores of 1.0, indicating having membership in "Other," were all correctly predicted by the regression equation. These data indicate that it would be statistically proper to consider the Disaffection vs. Other dichotomous variable to be somewhat bipolar with respect to it being interpretable. 

Assembling the Research Facts Into a Theory


     Up to this point, what is known? We are now aware that conducted multivariate statistical research reveals that there are two bipolar dimensions, on which a total of 153 U.S. spies (out of 175, which is believed to be fairly close to the absolute total number since the end of World War II)**** can be placed, based upon information derived from the nonclassified spy literature. These two bipolar dimensions, because of the original classification strategy, can be considered to be orthogonal, i.e., independent, with respect to one another. Two separately developed multiple regression equations (one described in paragraph) have been built which can be used to predict position on these two bipolar dimension measures. The prediction information for both regression equations can be regarded as being of the biographical/demographic kind. 

     Two orthogonal bipolar variables can be rather easily represented graphically. This was accomplished and the results shown as Figure 1. [Note - Figure 1 is shown as the last ‘page’ in this present report presentation.] The only metric change which was accomplished for this graphic representation was that the number identifying pole positions were transformed from 1.0 and 2.0 to -1.0 and 1.0. Positive unity values identify the Money and Disaffection pole positions whereas negative unity values identify the Ideology and the Other pole positions. The previously employed cutoff value of 1.5 is now equal to 0.0 on the graphing. So as to possibly render Figure 1 more interpretable and understandable, biographical/demographic information for a dozen well-known American citizen spies (Christopher J. Boyce; James Hall, III; Alger Hiss; William P. Kampiles; Karl F. Koecher; Richard W. Miller; Ronald W. Pelton; Jonathan J. Pollard; Ethel Rosenberg; Julius Rosenberg; Barbara Walker; and John A. Walker, Jr.), all of whom can be considered to possess rather high-profiles due to extensive past press coverage, were input into the multiple-regression equations for predicting or establishing position on each of the two bipolar, orthogonal dimensions. Because it is possible, when employing these multiple-regression equations, to produce predicted values, which have numerical value in excess of the theoretical polar values, some of these 12 spies do have scores on the motivational dimensions greater than the theoretical maximum scorings. 



     Inspection of Figure 1 clearly reveals some of the extremely different predicted positions, in two-dimensional motivational space, for many of the 12 chosen, well-known spies. It should be recognized that the actual scorings or positionings of these 12 spies were accomplished by employment of the two developed multiple-regression equations rather than the very simple dichotomous measurement arrangement which constituted the ‘true’ dependent variables against which the multiple-regression equations were built. Had these spies been graphically located, based on their original dichotomously-measured scores, they would have been placed only at the zero and the +/-1.0 positions on the two dimensions. Even without actually knowing the singular most-important spying motivation, such can be quite accurately predicted from a knowledge of several or more biographical/demographical variables using the developed multiple-regression equations. It is these ‘predicted’ locations in two-dimensional, spying motivational space that are shown in Figure 1. 

     As was noted and discussed in Stone (1991b), the ability to locate a spy or suspected spy on the single bipolar motivational dimension of Money vs.. Ideology can be seen to have practical value. When only considering this particular single bipolar dimension, I had suggested that it could have value as a police investigative tool (when ongoing investigation of a possible spying case is still incomplete and underway) to ‘validate or verify early impressions regarding motivational causes for spying on the part of individual spies, for possible use in security clearance investigations and reinvestigations, and for development of espionology research measures or variables. Exactly the same suggestions can be made now for the two-dimensional (or factor) model for spying motivation. Instead of having only the potential to locate a studied individual on a single bipolar dimension, we now have the ability to measure the individual on two orthogonal bipolar dimensions; which are believed to ‘explain’ combinations of spying motivations. 

     A few years ago, when I was preparing a manuscript, pertaining to the psychology of spying, I received comments from colleagues, who were reviewing pre-publication versions of the manuscript, that what I was saying about spies might also apply to non-spy white-collar criminals (Stone, 1989). This same kind of comment may also be applicable with the currently described two-factor motivational model or theory for spying; it too might be potentially helpful in attempting to understand motivations which produce white-collar criminality. In this regard, there seems to be no problem with the money or greed end of the bipolar Money vs. Ideology dimension, however the ideology pole seems more difficult to interpret with generalized white-collar criminal behavior. Although perhaps unlikely, it is possible to conceive of a white-collar criminal act, such as bank fraud, being caused by a ‘Robin Hood’ type of ideological belief concerning the wrongness of banking institutions having so much money when poor or homeless people abound. In other words, there probably are some white-collar criminal acts committed under the rationale of a "rob the rich, give to the poor" philosophy. An actual, not too long ago example of this kind of situation, which was widely reported in newspapers and news-magazines nationwide, was the case in which a real estate broker, in the Washington, DC area, stole millions of dollars from the Housing Urban Development (HUD) Office so as to have been able to personally distribute the money to the poor and homeless. The press soon gave this criminally charged individual the moniker of "Robin HUD" ("Robbing HUD," 1989). 



     The second bipolar factor (Disaffection vs. Other) can be, seemingly without any difficulty, applied to the general category of white-collar criminal behavior. White-collar criminality directed towards institutions, such as banks or corporations, not infrequently, seems to be motivated by anger or disaffection towards the victimized institutions. The "Other" end of this bipolar motivational factor or dimension also seems to be applicable when generalized white-collar criminal behavior is considered. Although not frequently seen, it is not believed to be actually rare that individuals are sometimes blackmailed or coerced in some fashion to commit or to participate in a white-collar criminal act.***** 



Barron J. (1987). Breaking the ring. Boston: Houghton Mifflin. 
Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. 
Psychological Bulletin, 52, 281-302. 
Deacon, R. (1987). Spyciopedia: The comprehensive handbook of espionage. New 
York: William Morrow. 
Eoyang, C. K. (1988). The etiology of espionage. In C. K. Eoyang (Ed.), Proceedings of
Personnel Security Research Symposium (pp. 79-97). Monterey, California: 
Defense Personnel Security Research and Education Center. 
Herbig, K. L. (1988). Why spy? A history of recent American espionage. In C. K. 
Eoyang (Ed.,), Proceedings of the Personnel Security Research Symposium
Merriam Webster Inc. (1986). Webster’s ninth new collegiate dictionary. Springfield, 
Ma: Author. 
Pincher, C. (1987). Traitors: The anatomy of treason. New York: St. Martins. 
Robbing HUD. (1989, December 4). U.S. News & World Report, 107, 15. 
Rotter, J. B. (1966). Generalized expectancies for internal versus external control of 
reinforcement. Psychological Monographs, 80, Whole No. 609, pp. 1-28. 
Scientists convicted of selling secrets. (1991, March 8) Washington Times, p. A2. 
Stone, L. A. (1989). On the psychological makeup of a spy. Forensic Reports, 2, 215- 
221. Also, as a summary in (1989). Bulletin of the American Academy of Forensic
Psychology, 11, 1. 


Stone, L. A. (1991a). I spy a myth. Security Management, 35, 26-32. 
Stone, L. A. (1991b). The use of demographic and biographical information to predict
specific spying motivations. Unpublished manuscript. 
Stone, L. A. (1991c). The canonical-correlation between security-clearance adjudication 
concerns and later motivational causes for espionage behavior. Bulletin of the
American Academy of Forensic Psychology, 12, 10. 
Stone, L. A. (1992). A prinicipal components analysis of 59 variables descriptive of 
uncovered spies. In D. Sands & J. Ellis (Eds.), Proceedings of the 24th Annual
Conference of the Military Testing Association, Vol. II, (pp. 576-81). San Diego, CA: 
Navy Personnel Research and Development Center. 



#This research/theory report was originally developed back in 1992 and was given some highly focused distribution among some selected  members of this country's intelligence agencys' counterintelligence personnel.  No real availability of this report consequently was available in the so-called 'open literature.'  In the past year or so the author of the report was given some suggestion that the report should be given some type of publication so that its contents might be available to a more wide audience than it had been in the past. 

##Dr. Stone’s mailing address is: P.O. Box 395, Harpers Ferry, WV 25425; his email address is: 

*This investigator’s U.S. spy database is given considerable description, as to how it was originally planned and built, in Stone (1991a, 1992). 

**When the writer first encountered the symposium paper by Eoyang (1988), he initially assumed that Eoyang was describing models for espionage that had emerged from analysis of data-based research. Although the models may have some appearance of having been developed, based upon bonafide research results, in reality they are nothing more than some very interesting graphic representations of the results of ‘armchair thinking’ regarding the subject. The present writer believes that most of Eoyang’s ideas are researchable but the stage which Eoyang described in his 1988 presentation could be regarded as theorizing and conceptualizing at a pre-research level. 

***Although Barbara Walker has never been charged with espionage, there would seem to be no doubt that she actually assisted her husband, John Walker, for at least a couple of years, when he made "dead-drop" exchanges. Together, they would drop off bundles of classified information and later go to other sites to pick up payment, bags of money. There seems to be no doubt at all that Barbara assisted her spy husband in these kinds of activities especially during the 1969-71 years. An interpretation that she knowingly participated in some of her husband’s spying activities, as well as financially benefiting from same, has been stated by several investigative authors (e.g., Barron, 1987). 

****When the statistical analyses, upon which the present reported investigation is based, were completed in the Fall of 1991, the involved spy database then contained information pertaining to 175 spies. Since then this database has been continued to be enlarged over the years, as of Spring 2001, it now is much larger. 



*****Shortly after the last manuscript version, upon which the present articles is based, was completed, the writer discovered, in a major newspaper, a short article that described a then quite recent industrial espionage case where two scientists had been convicted of conspiring to sell trade secrets belonging to large pharmaceutical companies ("Scientists Convicted," 1991). The present writer managed to obtain some biographical/demographically information pertaining to the two convicted industrial spies and then input this information into the two multiple-regression equations, described in the present article, for spies. Predicted values of above 2.0 were found on the Monetary vs. Ideology motivational factor, which most clearly indicated that greed would be the expected position they would occupy on this dimension. Also, it was found that both of the industrial spies had quite low-positive score-values (i.e., between 0.33 and 0.40) on the Dissatisfaction vs. Other motivational dimension. Such scores can be interpreted as being indicative that neither one of these polemic motivational descriptions was likely operating. These predicted values on the two orthogonal motivational dimensions seem to be very consistent with their actually known motivations for stealing and conspiring to sell trade secrets. These two cases, where convictions for industrial espionage rather than national security espionage were involved, cannot be regarded as being the kind of proof that the described theory for spying motivation can be clearly generalized also to the more broad industrial espionage field. However, the apparent accuracy of the computed motivational scores with just these two studied industrial espionage cases can be considered as a beginning indicator that the spying motivational theoretical model might also be applicable with this kind of spy. More research is clearly called for. 



Click Here to Return to Index Page of 
Psychology of Espionage Reports