The goal of this article is to support the use of the rational choice model for criminal behavior and question the model. Matsueda and the other researchers hypothesize that the rational choice model is an effective way to examine criminal behavior. This article devises 12 other hypotheses to creating an updating model. The first hypothesis is labeled Prior Perceived Risk and states that the perceived risk in the future correlates positively with the prior perceived risk. Their second hypothesis is labeled Bayesian Learning Based on Personal Experience with Arrest and states that the those who have experienced the risk of arrest is positively correlated to their perceived risk of arrest. Their third hypothesis is labeled the Bayesian Learning Based on Personal Experience with Crime and states that individuals who have experienced risk of arrest and not been caught are negatively correlated to the perceived risk of arrest. The fourth hypothesis is labeled Shell of Illusion and states that newer criminals overestimate the risk of arrest when compared to more experienced offenders. The fifth hypothesis is labeled Bayesian Learning Based on Vicarious Experience and states that Delinquents have a negative correlation to their perceived risk of arrest. The sixth hypothesis is labeled Social Structure and Perceived Risk and states that Perceived risk is correlated to where they are located, their family structure, race, gender, and residential stability. The seventh hypothesis is labeled Deterrence and states that criminal acts are reduced when there are greater costs in comparison to the rewards that can be accumulated from committing a crime. The eighth hypothesis is labeled Opportunity Costs and states that schooling, work, and other opportunity costs will reduce the crime rate. The ninth hypothesis is labeled Psychic Returns to Crime and states that individuals partaking in crime are correlated to the thrills and social status of succeeding at the crime. The tenth hypothesis is labeled Criminal Opportunities and states that criminal behavior rises in correlation with opportunities that present the possibility of succeeding with the crime. The eleventh hypothesis is labeled Limited Rationality and Discounting which states that criminal behavior is tied by the rewarding opportunities a criminal at gives but not by their risk perception of getting arrested. The final hypothesis is labeled Instrumental versus Expressive Crimes and states that rational choice and deterrence affect theft more so than violence.
Rational choice has developed considerable over the years with researchers such as Morrow in 1994, Posner in 1998, Sunstein in 1999, and many more. This theory has become an individual-level theory of motivation that ties to macrolevel theories of social structure. In 1994, researchers named Hechter and Kanazawa found that rational theories can be used in explaining areas beyond market or rational behavior; however, there still needs more research in order to solidify the theory as a proper method. This article states that criminal behavior is very complex and difficult to analyze through this method due to the fact that criminal behavior is typically irrational and suboptimal. This contrasts considerably from market behavior, which is deemed as rational and suboptimal behavior. Researchers believe that using the rational choice theory to analyze criminal behavior will aid in its development because it will offer that utilitarian principles serve as a foundation for our legal institution. A researcher in 1789 named Bentham argued that happiness is the root to maximizing pleasure and minimizing pain, binding to the utilitarian principle that greater happiness is due to greater numbers. He also believed that political, moral, physical, and religious sanctions are what causes one to feel pleasure or pain. Beccaria in 1764 believed that giving protection to those willing to relinquish their freedom to violate other people’s rights, and presented that punishment should be just enough to deter those willing to carry out a criminal offense. These classical theorists’ views are contrasted by critical legal scholars such as Pashukanis from 1929 and Garland from 1990. These two believe that the legal system is too ideological and can spark inequality between classes. What this article hopes to achieve is to examine rational choice as a whole rather than take sides when considering legal institution. Becker in 1968 created a model for criminal decision making. Becker’s model theorizes that individuals will commit criminal acts when it is more rewarding than it is punishing when perpetrating a crime. Those using Becker’s model utilize a system that measures the effects of risk of imprisonment (imprisonment per capita) or risk of arrest (arrests per reported crimes). The results are controversial because they assume that the criminals already know the amount of risk they are undergoing and the likelihood it takes to get caught. In order to contrast to this, researchers made a Subjective expected utility model which replaces the one probability with multiple subjective probabilities. They are still considered as a rational model due to the fact that the average of subjective probability distribution should correlate to the value of objective probability. The rational choice theory being tested in this article assumes that perceiving the risk of crime is correlated in reality. It believes that in order to test rational choice and deterrence theories, there needs to be a model that considers the perception of risk. One model is considered to be the Bayesian Learning Model. This model dictates that individuals already have prior knowledge of the probability of getting caught. They then consider new information on recent arrests to help gage and update their chances of getting caught. This is known as posterior probability and is a combination of past preconceptions as well as new preconceptions. Many psychologists and other researchers criticized this view due to the fact that it can lead to biased opinions on the perceiving risks. This article states that the Bayesian learning perspective is not a plausible learning model to follow due to the fact that people may overestimate the probability that they will become arrested and fail to act upon their criminal tendencies. Thus, Matsueda and the other researchers believe that three important sources of information need to be addressed; their personal experience with crime, their knowledge of others with crime, as well as the environment they are located in.
In order to test the models that the researchers have created, they conducted and extracted data from Huizinga and colleagues. This became known as the Denver Youth Survey (DYS), and is a study that tests those who are delinquents and drug abusers in high risk neighborhoods in Denver, Colorado. High risk neighborhoods were identified by analyzing them based on their family structure, race, housing, mobility, age difference, SES, and marital status. Those with the most disorganized clusters were selected. They also selected places within each cluster where there was the greatest number of arrest rates as according to the Denver Police Department. The researchers then selected 20,300 of 48,000 houses and interviewed those between the ages of 7-15. 1,528 youths were interviewed in the first wave of screening. Those 11 and older were given a youth questionnaire. The total sample added up to 1,459 respondents and 3,298 person-years. Thus, a table is formed depicting the descriptive statistics which comprise of ethnicity, gender, social status, those employed, in school or not in school, and more. They then tested their Bayesian updating model of perceived risk of arrest by controlling for different variables such as concentrated disadvantage, residential mobility, the amount of blacks, and the amount of crime in 1984 as well as age, race, gender, impulsivity, etc. Afterwards, they tested their models of perceived risk of arrest using random-effects Tobit models with lagged regressors. These Tobit models have three different assumptions. The first assumption is that the perceived risk is measured based on percentage from 0-100. Second, the person-year data is dependent, so they used random effects models that aid in estimating the un observed heterogeneity. Lastly, the experienced certainty and peer delinquency is related to the perceived certainty, which poses as a problem. To combat this, they used a three-wave panel model with a lagged endogenous predictor so that they can manipulate it to their liking. The random effects model helps combat any potential bias of any of the tests.
The next model is of rational choice as well as criminal behavior. They assumed that delinquent accounts are generated from a Poisson distribution. They found that the variance was too high as compared to the average count of delinquency, which would ultimate cause errors in their study. In order to combat this they underwent negative binomial models that included lagged time-varying covariates so that everything is consistent and it included the lagged endogenous predictor. Again, they used a random effects specification so that they can remove any potential bias from any of the tests.
Based on Model I in table 2, it is revealed that females and younger individuals have a higher risk of arrest for criminal acts. Females believe that there is a higher risk for arrest than males by eleven percent on average when it comes to theft, and are nine percent on average when it comes to risk for arrest when it comes to violence. Younger individuals that have no sibling perceive that there is a lower risk for getting caught for violence, with a non-significant variable when considering theft. They state that their variables fail to exert significant effects on certainty and that it may be due to the fact that the places sampled where the neighborhoods that had the most crime. Model II found that individuals who believe in risky behavior believe that there is a lower chance of getting arrested for both theft and violence. Model III discovered that there is no correlation between experienced certainty and perceived risk for criminal acts both theft and violence. They found that their negative coefficients support their fourth hypothesis which state that those who have not committed a crime will overestimate the risk of arrest in comparison to more experienced criminals. The gambler’s fallacy is disproven due to the fact that the coefficients would be positive. They found that their research supports the hypothesis that those who have not gotten arrested for their actions exhibit a lower perception of risk for arrest (15 points lower than those who have no prior experience). Those more experienced perceive the risk as 8 points lower for theft, and 12 points lower when considering violence. These correlations support the hypothesis of Bayesian Learning. They were unable to support their hypothesis that young individuals ignore new information when it comes to arrests. They found support for the Bayesian learning hypothesis where on average, certainty of arrest is inversely correlated to the amount of offenses not being punished or sanctioned. Those who have committed a high amount of crimes see a 10-point lower risk perception for violence than criminals who have no prior experience while those with a medium amount see an 8-point lower risk perception for violence. For cases where it pertains to theft, those who have committed a high amount of crimes see a 10-point lower risk perception while those with a medium amount see a 4-point lower risk perception. These results support their hypothesis of Bayesian learning. There is more support for the Bayesian learning hypothesis that a higher delinquency in peers will correlate with the lower risk perception of arrest. The found that prior perceived risk is an important effect on the risk perception even when subjected to new information.
For Table 3 on Parameters of Criminal Behavior Models, they found that males and those with very high impulsive tendencies commit more crimes than those that do not. Youth on the older side of the spectrum commit slightly more violence and a negligible difference of theft. They also discovered that black individuals commit more violence than that of whites, but do not exhibit more criminal acts of theft. Hispanics commit slightly more of each than whites. Biological parents have a negative correlation on theft but not when it pertains to violence, and income only negatively affects violence. They discovered that their research supports their claim that those with prior criminal acts will commit more crimes. Model II on Table 3 presents that younger individuals who like to commit daring acts will be more likely to commit criminal acts. This supports their rational choice model which hypothesizes that those who don’t take risks will avoid criminal behaviors as compared to those willing to take risks. They then looked into individuals’ grades, and discovered that higher grades inversely correlate with criminal behavior such as acts of violence and theft. Employment has a positive effect when compared to violence but not on theft. They also found that those not attending school commit less criminal behavior than those who do. When pertaining to perceived costs and rewards, they found that their research supports their rational choice perspective. They found a significant deterrent effect of arrest, in contrast to research in the past. Those who believe they will be caught will commit fewer acts of violence or theft. When it pertains to the rational choice process’ reward side, it was found that those who commit crimes due to excitement affect only theft and not violence. Younger individuals who enjoy excitement and value excitement from stealing commit more acts of theft. This supports the hypothesis that thrills are positively correlated to acts of theft. Many younger individuals also commit more crimes when they view them as making them look cooler. This discovery supports ethnographic research theories that show crime as an avenue to gain status within certain groups. This also proves that this image is a psychological reward when committing a crime. Model III discovered that those who find opportunities to commit theft or violence will inherently increase their likelihood to commit said crime. This supports their hypothesis of Criminal Opportunities. Finally, they failed to support their discounting hypothesis, whilst succeeding in supporting both sides of the rational choice model, costs and returns.
The researchers obtained multiple results that help support the Bayesian learning model of perceived risk formation, the rational choice model of criminal behavior, and a deterrence hypothesis of perceived risk of arrest. These studies obtained critics who believed that the temporal ordering of crime and perceived certainty contradicted the causal ordering. Their results contrast drastically with other studies. In order to justify their research, they stated that they had more refined measures of certainty, a larger sample of younger individuals in high risk neighborhoods, and random effects negative binomial estimators so that their research is free of bias and inefficiency. They found that consistent deterrents affect the risk perception of arrest. They state that their needs to be more research on perceived risk using both self-report outcomes and behavioral intentions in order to figure out their vignette results. They found that their results contrast to that of prior studies, but state four different caveats to support their research. One is that they sampled high risk disorganized neighborhoods and that they did not generalize them when compared to low risk neighborhoods. Secondly, they have no formally tested for heuristics. Thirdly, they found that they did not exhaust every possible incentive for crime in their research even though they view that their models give strong and somewhat accurate measurements. Finally, they state that they relied on a on lagged effects to their perceived costs and returns to crime, which may have skewed their results, understating certain values. They state that rational choice theories and institutional theories have a place together and that they can be complementary. They found that certain education factors can cause a student to commit a crime. They also discovered that social status among groups is a key factor in providing as incentive to commit a crime. Thus, they believe that reducing the probability of decreasing the potential reward of social status will drastically decrease their criminal behavior. The problem is that the United States has one of the highest arrests and imprison rates among Western nations, so increasing it will be harder and more demanding of the criminal justice system. Knowing this, enacting policies to increase the probability of punishment may prove to be of little value when it comes to reducing crime rate substantially.
Their results show that there are multiple promising gateways to intervene with criminal behavior. Some examples include: school programs that are directed towards decreasing the social status of criminal actions; programs that can increase opportunity costs to crime; and recreational programs that can help in providing for young individuals with risk-taking tendencies. They also believe that educational programs that can teach young individuals that inhibit them from believing that crime is easy to get away with.
They state in the article that policies that increase punishment on crime resonate well in America, and that they believe that it is based more on ideology rather than empirical research. They believe as if increasing punishment is detrimental and not the answer to deter criminal acts.