Chapter 13 Chapter 9 Busting the Myth of the “Wonder Bullet”—The Problem of Multiple Causes

In Chapter 8 we focus on the importance of aggregation operations and the search for more efficient research methods that can establish a single link between variables.In this chapter, we will not only look at a single link between two variables, but also focus on another important idea, that is, human behavior is determined by multiple factors. Any given behavior is not caused by a single variable, but by many different variables.Assuming that there is a significant causal relationship between variable A and behavior B does not mean that variable A is the only factor that causes behavior B.For example, some researchers have found that there is a correlation between the time spent watching TV and academic performance, but they would not conclude that the time spent watching TV is the only factor affecting academic performance.The reason is simple, academic performance is also affected to some extent by a large number of other variables (eg, family environment, quality of schooling, etc.).In fact, relative to these variables, watching TV is only a secondary factor affecting academic performance.Likewise, viewing high levels of television violence is not the only cause of aggressive behavior in children, it is just one of many.

But people often forget that behavior is determined by multiple causes, and they seem to search for that so-called "magic bullet"—the only cause of behavior that interests them.Psychologist Teodoire Wachs (2000) cites as an example the way people have tried to explain school shootings in the United States between 1998 and 1999, noting that the reasons people think were involved included easy access to guns, parental control of children Low attention, internet, video violence, peer influence and mental illness.Watts believes that "few people feel that the surge in school shootings is the result of the combination of the above reasons, and any solution should not only target a certain underlying reason" (px).

As with many of the other principles discussed in this book, the idea of ​​diversity of causes is important.On the one hand, it reminds us not to rely too much on a single explanation.Because the world is deeply intertwined, and the factors that affect behavior are diverse and complex.Although we can prove that a certain variable causes a certain behavior, it does not mean that we have discovered the only reason affecting the behavior, or even the most important reason.In order to make a comprehensive explanation for a particular behavior, researchers must explore the influence of various variables on it, and integrate these research results to completely describe all the causal relationships related to the behavior.

On the other hand, just because a variable is only one of many factors that affect a particular behavior and can explain only a small portion of that behavior does not mean that the variable is insignificant.First, this relationship may have profound theoretical implications.Second, this relationship may have practical value, especially when the influencing variable can be controlled manually, as in the example of TV violence mentioned above.If controlling for this one variable can reduce violence by 1% per year, I don't think anyone would think it is insignificant.In conclusion, knowing how to control even a small cause of a problem behavior is of great value if the problem behavior is critical.

Rosenthal (1990) cites the example of treating heart disease. In one experiment, a certain treatment regimen improved the survival rate of patients by less than 1 percentage point; however, even this result was considered to be The significance is so great that the experimenters had to terminate the study early due to ethical considerations: since the experimental treatment is so effective, it is obviously unethical for patients who are randomly assigned to the control group to still use the placebo.Likewise, anything that reduces motor vehicle fatalities by 1% is critical -- saving 450 lives a year.A 1% reduction in the homicide rate would save more than 170 lives per year.In sum, the fact that an outcome is determined by multiple variables does not diminish the importance of any one variable that is causally related to the outcome—even if that variable makes only a small difference in the outcome.

The idea of ​​multiple causes leads to another important concept, that of interaction.This concept has been introduced in detail in many methodological books, so I will not praise it here, but just mention it briefly: When one factor that affects behavior and another factor work together, they will have different effects on the behavior than each other alone. Distinctly different impacts at work.This is what we often call an interaction: the effect of one independent variable depends on the different levels of another independent variable.An experiment directed by Simmons, Burgeson, Carlton-Ford, & Blyth, 1987 provides an example of this.The researchers looked at the academic achievement averages of a group of adolescents to see whether life events such as school transfers, puberty, puppy love, moving, and family breakdowns had an impact on academics.They found that the aforementioned life events, taken together, were key contributors to academic underachievement.

Another example is Michael Rutter's (1979) review of research on factors associated with childhood mental illness, in which he states: To understand the logic of how interactions such as those described by Nutt occur, imagine a risk scale with a score of 80-110 representing low risk, 110-125 representing moderate risk, and 125-150 representing high risk.Suppose we find that children have an average risk score of 82 in a non-stressful situation, 84 in stressor A, and 86 in stressor B.When studying the joint influence of both factors A and B on children, if it is found that the risk index reaches 126, that is to say, the combined risk index far exceeds the results predicted when studying a single factor independently, which shows that the factor There is an interaction between A and B.

Developmental psychology is full of examples similar to those described by Nutt.Bonnie Breitmeyer and Craig Ramey studied two groups of infants, one group of infants with non-optimal perinatal periods and the other group of normal infants (Breitmeyer & Ramey, 1986).After the two groups of babies were born, they were randomly assigned to two groups—an experimental group and a control group—and then put on a special parenting regimen designed to prevent mild mental retardation.Babies in the control group did not receive any special care.When these children were 4 years old, their cognitive development ability was tested, and it was found that under the special baby care program, children born in non-optimal perinatal periods had no significant difference in cognitive ability from normal children.However, children in the non-optimal perinatal period in the control group who received no special care performed below normal levels of cognitive development.The interaction of physiological and environmental factors in this study illustrates that a complex behavioral outcome (cognitive development) is determined by multiple factors.Negative cognitive developmental outcomes occur when children born during suboptimal perinatal periods are not adequately cared for.The researchers concluded: "The findings support a theoretical framework in which congenital disabilities and adverse environmental factors play a role in the developmental cumulative risk factors" (p. 1151).

Similarly, another study (Metalsky & Joiner, 1992) on the validation of the "Constitution-Stress Theory" of depression showed that negative life events combined with three psychological factors of vulnerability lead to the greatest likelihood of developing depression .These three factors are: attributional style (the tendency to attribute negative events to some stable, global factor), negative inferences about the self, and a general tendency to make negative inferences about the consequences of any behavior (see Alloy, Abramson, & Francis, 1999). Many negative behavioral and cognitive consequences follow a similar logic.For example, aggressive behavior in children is caused by the interaction of genetics and adverse social environment (Pennington & Ozonoff, 1996).Similarly, Pettit et al. (1999) found that adolescents who spent a lot of time socializing with their peers in early development and who received little parental supervision at home were more prone to externalizing behavior problems.

The positive results can also be explained by multiple factors and their interactions.Knight et al. (1994) examined psychological factors associated with children's propensity to help behaviors (eg, donating money to children in need) in their study of prosocial behavior in children aged 6-9 years.They found that some variables -- such as empathy, emotional reasoning, and knowledge about money -- had low correlations with prosocial behavior when they were alone.However, when these variables were combined, they were good predictors of prosocial behavior.For example, children who had greater empathy, stronger emotional reasoning, and an understanding of money donated four times as much as children who performed lower on these variables.

Thus, the concept of causal diversity may be more complex than you initially assume.Not only do you need to track and measure the various possible factors that influence problem behavior, you must also examine how these variables work together. The basic idea that complex events are determined by multiple causes seems easy to understand.In fact, this idea is easy to grasp and apply when the issue is not too controversial; Forget about the principle of diversity of causes.Countless times have we heard that debates over emotionally charged topics—the causes of crime, the distribution of wealth, discrimination against women and minors, the causes of poverty, the role of the death penalty, and the standard of taxation—are in debates. Do it in such a way that it feels like the problems are simple, one-dimensional, and have only one cause for the effect.These examples further reflect what Nisbett & Ross (1980) said: "While people sometimes acknowledge a diversity of causes, people act more consistently with beliefs about a single cause. In a sense In other words, people behave as if they treat causes as "hydraulic," or that possible causes compete with each other as in a 'zero-sum' game" (p. 128). "Zero-sum" games -- one person's gain is another's loss -- often mirror how we discuss emotionally charged topics.Under the influence of emotions, people often forget the principle of multiplicity of causes.Consider how two rival political parties discuss crimes in society.Liberals would argue that those with low socioeconomic status commit crimes because they themselves are victims of adverse social conditions (e.g., unemployment, poor housing conditions, lack of education, and loss of hope for the future, etc.).More conservatives would argue that there are also many poor people who do not commit crimes, so socioeconomic conditions are not the main reason.Instead, they believe that individual values ​​and personality traits are what really determine criminal behavior.Neither side seems to recognize that individual and circumstantial factors combined to drive criminal behavior. Political commentator Richard Cohen has also written about how we often turn “single cause” explanations 180 degrees based on preconceived biases.He cites the example of a 63-year-old farmer in Iowa, which has experienced a severe agricultural recession, which left him deeply in debt and lost his farm.In desperation feeling that he had nowhere to pay, he shot and killed the manager of the bank from which he had borrowed money, then shot his wife before killing himself.Neighbors and conventional media have suggested that the farmer "broke" because he was in dire financial difficulties.The media reports gave him great sympathy.Cohen wrote that the rancher was portrayed as a "hard-working business owner who fought doggedly against nature, the banks, and the Chicago produce merchants. He was honest and frugal, with the virtues of most Americans—self-employed, earning his own living , and fear God” (1985, p. 11). But what Cohen wonders is, if this person really killed and killed himself because of financial difficulties, can we use the same (single cause) theory to explain the killings that occurred in the slums? "If this is the cause of the collapse for the stranded farmer, why is it different in the slums? Why is it that poverty, lack of opportunity, third-rate schools and brutality are the causes of crime when it is suggested that , will receive so much blame?" (p. 11).Of course, Cohen points out another fallacy that arises in single-cause explanations: we use a single explanation to reinforce pre-existing biases.Cohen argues that this fallacy might be avoided by recognizing from the outset that the killings on farms and slums in the above examples are likely to be determined by a variety of factors.Both Iowa ranchers and ghetto homicides are influenced by both individual psycho-physiological traits and environmental pressures.No single cause alone can explain the crime.Criminal behavior is affected by many factors, partly environmental and partly individual. A discussion of the complex economic situation provides such an example.A socially important topic that has sparked decades-long debate is why the gap between rich and poor in the United States is growing (Beatty, 1996; Bronfenbrenner et al., 1996; Frank, 1999; Karger, 2005; Madrick , 2006).As with the Clever Hans example mentioned in Chapter 6 of this book, this fact is not in dispute; what is in dispute is the interpretation of this fact. Since 1979, real earnings (ie, adjusted for inflation) of male workers in the United States have fallen by more than 10 percent (Cassidy, 1995; Mishel, 1995).Of course, there are individual groups that are doing well.In the more than ten years from 1977 to 1990, the income of the top 1% of the total population increased by 74% (this value is also after inflation is excluded, Slemrod & Bakija, 1996); At the same time, the income of the middle class in the United States increased by only 3%; while the income of the bottom 20% of the population fell by 13%. In 1977, the richest 20% of society earned four times as much as the poorest 20%.By 1991, this figure was seven times (Frank & Cook, 1995). The massive transfer of wealth from one class of citizens to another has sparked a contentious political debate about its causes and effects.What is most striking about this debate is that these contenders are all focused on a single cause.Each side of the argument bases its argument on one cause only, and then goes out of its way to attack all arguments in favor of the other causes.In fact, econometric research (Beatty, 1996; Cassidy, 1995; Frank & Cook, 1995; Mishel, 1995) has focused on four variables (more than four have been proposed, but these four are the most extensive attention and research).One of these factors is technology.One of the arguments, for example, is that computers increase the productivity of their users, leading to an increase in their income.Conversely, computers also displace many jobs of unskilled workers (such as mail sorters, bank tellers, etc.), thereby reducing their wages.The second element of the debate is the continued influx of new immigrants, many of them unskilled workers, who have created an oversupply of unskilled labor that keeps wages already low.A third reason is globalization, which further exacerbates income inequality, as firms can outsource their operations and employ unskilled and semi-skilled (which is becoming skilled) workers in countries with lower wages, which This has exacerbated the surplus of unskilled labor in the country.A fourth reason is the ebb and flow of influence between labor unions and big business.The argument is that in the 1990s, labor strikes decreased, while management strikes (i.e., companies gave up a certain area as their production base, because production in other areas can achieve higher return on investment) increased , thereby reducing the value of labor while increasing the value of capital. What did economics discover when they studied these four variables?You've guessed it.All four of these factors work together to create rising social inequality.This example also demonstrates the previously mentioned concept of interaction.Cassidy (1995) pointed out in the article, “Certain factors may interact and reinforce each other. As global competition intensifies, business managers weaken labor unions and invest capital in computer technology. Similarly, corporate relocation Both the threat of the location and the increase of cheap labor from outside would lead to a further weakening of the power of the trade unions” (p. 122). Like problems in economics, almost all complex problems studied in psychology are determined by multiple causes.Take learning disabilities as an example, a problem that has been extensively studied by educational, cognitive, and developmental psychologists.Brain lesions were found to be associated with learning disabilities (Galaburda, 1994; Hynd, Clinton, & Hiemenz, 1999).Studies have also found that learning disabilities have genetic causes (Cardon et al., 1994; Olson, 1999).These two research results seem to allow us to draw a conclusion that learning disabilities are purely physiological-brain problems, but such conclusions are wrong, because further research has found that part of the cause of learning disabilities is in the early Lack of certain guiding experiences in schooling (Pressley, 2002), and poor home environments (Senechal & LeFevre, 2002; Snow, Burns, & Griffin, 1998).Learning disabilities are thus not caused by a single cause; rather, they are the result of an interaction of physiological and environmental factors. This chapter is simple, but very important.When examining the causes of behavior, think in terms of the principle of diversity.Don't fall into the trap of thinking that a particular behavior is caused by a particular cause.Most complex behaviors are determined by multiple causes.A variety of factors work together to produce a certain behavior.Sometimes multiple factors interact when combined.In other words, the overall effect of variables acting together will be completely different from the effect obtained when they act alone.