1. An example of such confluence between economic theory and psychology can be found in Edwards (1954 Edwards W. The theory of decision making. Psychological Bulletin 1954; 51: 380–417[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]), which contains a scholarly discussion regarding the particular components of individual decision-making that overlaps quite nicely with some of the main tenets of economic rationality. See also Camerer (1995 Camerer C. Individual decision making. Handbook of experimental economics, J. Kage, A. E. Roth. Princeton University Press, Princeton, NJ 1995; 587–703 [Google Scholar]) for a thoughtful opus on decision making from an economic perspective. Both Cameron (2000 Cameron S. Nicotine addiction and cigarette consumption: A psycho-economic model. Journal of Economic Behavior and Organization 2000; 41: 211–219 [Google Scholar]), who refers to psycho-economics, and Rabin (2002 Rabin M. A perspective on psychology and economics. European Economic Review 2002; 46: 657–685[Crossref], [Web of Science ®] , [Google Scholar]), who refers to behavioral economics, present an excellent discourse on the potential cross-fertilization (i.e., a merger) between psychology and economics. In a related vein, Lewin (1996 Lewin S. B. Economics and psychology: Lessons for our own day from the early twentieth century. Journal of Economic Literature 1996; 34: 1293–1323[Web of Science ®] , [Google Scholar]) presents an excellent review of how the two fields need each other to proffer explanations for behavior.

2. Although early studies of ecstasy users contributed to the belief that ecstasy had generally a few negative health effects (Peroutka, Newman, and Harris, 1988 Peroutka S. J., Newman H., Harris H. Subjective effects of 3,4-methylenedioxy methamphetamine in recreational users. Neuropsychopharmacology 1988; 1: 273–277[PubMed], [Web of Science ®] , [Google Scholar]; Solowij et al., 1992 Solowij N., Hall W., Lee N. Recreational MDMA use in Sydney: A profile of ‘ecstasy’ users and their experiences with the drug. British Journal of Addiction 1992; 87: 1161–1172[Crossref], [PubMed] , [Google Scholar]), more recent studies have shown that club drugs users recognize and do report symptoms of abuse and dependence (Cottler et al., 2001 Cottler L. B., Womack S. B., Compton W. M., Ben Abdallah A. Ecstasy abuse and dependence among adolescents and young adults: Applicability and reliability of DSM-IV criteria. Human Psychopharmacology 2001; 16: 599–606[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]; Topp et al., 1997 Topp L., Hall W., Hando J. Is there a dependence syndrome for ecstasy?. NDARC Publications, Sydney 1997, (Tech. Report 51) [Google Scholar]).

3. We were unable to obtain a satisfactory model positing distinct constructs of abuse and dependence. The statistical artifact of an almost perfect correlation between the two constructs necessitated modeling only one construct. Also, collapsing these two into a single latent factor might go against the grain of certain nosological criteria; thus, we decided to proceed in our modeling efforts specifying only the dependence construct. This enables us to gain certain clarity with respect to negative sequelae of ecstasy consumption and use a construct possessing greater reliability (alpha for the abuse construct was. 91, whereas alpha for the dependence construct was. 96). From the perspective of model fit, a model positing two constructs produced a correlation of r =. 98, p ≤. 001 and the model fit well, χ²(43) = 81.01, p ≤. 001, NFI =. 939, CFI =. 970, RMR =. 03, RMSEA =. 04. A model constraining the correlation between Abuse and dependence to unity also fit well, and the nested difference between the base and constrained model was not significant, Δ χ²(1) =. 52, p = n.s., indicating the constrained model fit the data equally well.

4. Based on the psychological meaning associated with sexual risk-taking, we expected greater sexual risk to be associated with greater dependence. This expectation is consistent with the literature on sexual risk and drug use (for an excellent review, see, for example, Leigh and Stall, 1993 Leigh B. C., Stall R. Substance use and risky sexual behavior for exposure to HIV: Issues in methodology. American Psychologist 1993; 48: 1035–1045[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]). In essence, and using this argument, sexual risk behaves like addiction, driving the individual to act compulsively (and perhaps impulsively) leading to the same dependence problems that occur with substance use. The data support this contention once we strip the model down to the bare essentials. In other words, trimming the model of all the exogenous constructs and leaving only the economic measures and sexual risk, the path from sexual risk to ecstasy use is positive and significant (β =. 09, p ≤. 05), as is the path from sexual risk to dependence (β =. 14, p ≤. 01). However, once we begin incrementally adding exogenous constructs to the base model, there is statistical evidence of suppression. Once the endogenous constructs (dependence and ecstasy use) are residualized on depression and motives (in addition to the economic indicators), a completely different picture emerges. We conducted a series of analyses to isolate the change in sign and found the reversal occurred with the introduction of the motivation construct. Stripped of everything exogenous in the model (including the economic indicators), we obtained a positive relation between sexual risk and dependence (β =. 14, p ≤. 01) and this effect remained consistent with the introduction of the depression factor (β =. 12, p ≤. 01). Furthermore, the association between depression and sexual risk was positive (r =. 11, p ≤. 01). However, introduction of the motivation construct changed the sign appreciably between sexual risk and dependence (β = −.11, p ≤. 05). Other parameters of importance in this model included the coefficient from motivation to dependence (β =. 66, p ≤. 001) as well as the associations between sexual risk and depression (r =. 25, p ≤. 001) and the association between sexual risk and motivation (r =. 37, p ≤. 001). When we ran this model regressing consumption and dependence on the economic indicators, the same results were obtained. That is, there was a positive and significant path from sexual risk to dependence (β =. 14, p ≤. 01). Likewise, the relation between sexual risk and ecstasy use was positive (β =. 09, p ≤. 05).

5. The first step in testing whether the U.S. and Sydney samples are drawn from the same population constrains the factor loading (lambda) matrix to equality. This model fit well, χ²(311) = 748.297, p ≤. 0001, NFI =. 828, CFI =. 891, SRMR =. 102, and RMSEA =. 068. By all indications, releasing any of the factor loading constraints would not make a significant improvement in the overall model fit (a nested comparison of the fully constrained model with a less restricted model positing relaxed factor loadings would require a difference of at least 3.84 χ² points for each degree of freedom corresponding to a single parameter change). The next model constrained the factor intercorrelations (phi matrix) to equality between sites. This model fit well, χ²(316) = 758.314, p ≤. 0001, NFI =. 826, CFI =. 890, SRMR =. 108, and RMSEA =. 068. Again, there was no indication that freeing any of the cross-group constraints in the factor correlation matrix would improve significantly on the model. A third and final step involved testing equality of the obtained structural (predictive) parameters. This model included the economic indicators as predictors of consumption and dependence. Fit statistics showed this model was adequate, χ²(454) = 946.144, p ≤. 0001, NFI =. 797, CFI =. 882, SRMR =. 093, and RMSEA =. 060. Importantly, while none of the model fit indices achieved their suggested benchmarks, the purpose of this tiered model testing strategy was to examine the tenability of cross-group constraints. Since no constraints needed to be relaxed to achieve a superior model fit, the two samples could be said to come from the same population.