Extant studies have shown academic misconduct, that is, cheating, to be a pervasive problem among college students. A number of avenues have been explored in order to determine those factors related to this type of deviant behavior. This study looks at a variety of traditional theoretical factors, including constructs representing social bond, self‐control, social strain, and differential association theories, using data from a stratified random sample of 674 undergraduates at a Southwestern university. We also enter into the equations a host of other factors that have been found to be related to academic misconduct in previous studies. Separate equations were estimated for each theory, and in the full model all theoretical variables were included for a total of six equations. In the full model, the data did not support strain or social bond theories, and moderate support was found for self‐control theory; however, strong support was found for differential association theory. The final equation accounted for 53.2 percent of the variance in academic misconduct.

In what has become a classic work in the field, Matsueda (1982) tested control theory against differential association theory using Hirschi's (1969) Richmond Youth Project data. Matsueda found that measures of “definitions favorable to law violation” entirely mediated the effect of his social control measures and friends' delinquency, and concluded that differential association theory was supported over control theory. We note several problems with Matsueda's specification of control theory, and we reanalyze the Richmond data including measures of commitment to conventional goals and several attachment to parents variables that Matsueda excluded. We also propose and test a new method of measuring the social bond, conceptualizing the social bond as a second‐order latent construct. In contrast with Matsueda's findings, we find that the social bond and friends' delinquency retain important direct effects on delinquency, and that these effects are greater than those of definitions. Thus, our results are more supportive of control theory than differential association theory.

An extensive amount of criminological research has focused on the relationship between the inability of adolescents to achieve financial success or middle‐class status through legitimate channels and their subsequent involvement in delinquency. This genre of research is commonly recognized as structural‐strain theory. As the efficacy of this theory has periodically been questioned, this study addresses some of the methodological and theoretical issues associated with structural‐strain theory. Using a sample of 8,338 adolescents from a southern state, we use a structural equations model to test a hypothesized sequence ultimately leading to delinquent behavior. We believe that this method is more appropriate than more traditional methods (e.g., multiple regression) for investigating some of the linkages specified in previous strain models. Our results support structural‐strain theory, though the hypothesized model has more predictive power among European American than African American youth. Explanations are given for these differential findings, and recommendations for future studies are suggested.

The linkages among several popular theories in the delinquency literature-social disorganization, strain, and social control theory-as well as their effects on delinquent behavior are investigated in this study. Social disorganization has been primarily studied with aggregate-level data in prior work, and we note certain theoretical and empirical issues associated with this approach. In this study of public high school students in Mississippi (n = 8,338), social disorganization is measured by observed indicators of juveniles' perceptions of neighborhood deterioration. Social strain is operationalized through perceptions of blocked opportunity, whereas social control is measured through the respondents' reported level of attachment, commitment, and involvement in school. Structural equation models with latent variables are estimated separately by race and residence in order to more fully compare the linkages among these