Abstract
This study tested Agnew's (1992) general strain theory (GST) of delinquency and crime. GST predicts that strain has an increasing effect on delinquency, and that the effect of strain is conditional upon several variables, including delinquent peers. I analyzed longitudinal data of 87 Japanese college students' deviant behavior in a classroom with a one-week interval, using Ordinary Least Squares (OLS) regression and applying an Expectation Maximization (EM) algorithm to estimate missing values. I found a significant main effect of strain and a significant interaction between strain and delinquent peers, after controlling for variables such as gender, school year, the number of previous delinquencies, and other variables based on social control and differential association theories. These results are consistent with GST. However, I also found that delinquent peers had a decreased effect for unstrained students, and an increasing one for strained students. The implications of the results are discussed.
