Causality and phenomenon of contradictory outcomes in research
http://www.100md.com
《中华医药杂志》英文版
Causality and phenomenon of contradictory outcomes in research(pdf)
Emeq Yezreel Academic College & Oranim Academic College
Correspondence to Marwan Dwairy, D.Sc. Associate Professor, Ora Street 3b, P.O. Box 14710, Nazerat Ellit 17000, Israel
Tel:Work: 972 46561541, Mobile:972 054 6334515, E-mail:psy@marwandwairy.com
[Abstract] A considerable number of empirical studies that examined the causal relationships between factors arrive at mixed or contradictory results. Typically, these results are attributed to methodological differences. This article proposes that contradictory results are inherent in the simple causality and reductionistic approach that dominates the experimental paradigm. Such results created because the influence of factor A on B is expected to be different when in the presence or absence of other intervening or intermediary variables. Our claim is that thinking in terms of causality is too simple way of understanding complex phenomena in nature and in society. Research of questions regarding causality must focus on the causal mechanism by applying a dynamic-systemic view, cooperation between quantitative and qualitative research, different fields of knowledge, and use of meta-analysis and retention of a general philosophic perspective.
[Key words] causality;contradictory outcomes;research
A considerable number of empirical studies that examined the influence of one or more factors on other factor/s arrive at mixed even contradictory outcomes. The phenomenon of “contradictory outcomes” is common in nearly all research areas in psychology and medicine. For example, studies that examined the influence of personality factors on health (Smith and Gallo, 2001)[31], thinking on pain (Turk, 1996, 2001)[34,35], debriefing on persons exposed to traumas (Meichenbaum, 2001; Orner, & Schnyder, 2003)[21,24], and viewing of violent films on violent behavior (Bandura, 1973, Feshbach, & Singer, 1971)[4,12] arrived at contradictory outcomes. Such outcomes appear in the following manner: Certain studies reveal that factor “A” amplifies “B,” while other research found that factor A decreases B, and yet other studies found that there is no influence by A on B. The common tendency in literature is to assume that contradictory outcomes are due to methodological differences that exist between the research studies: differences in the definition of variables, in measurement tools, in the sample, type of analysis, and so forth. This article proposes a more basic explanation for the phenomenon of contradictory outcomes, one that relates it specifically to the causal and reductionist forms of thinking that characterize experimental research paradigms in science, in general, and in the social and behavioral sciences, in particular. In effect, these methodological differences create substantive differences in the field or in the system of the phenomenon and create different combinations of variables, such that the outcomes received are contradictory in regard to the influence of independent variables on the dependent variables. In my opinion, contradictory results are created because the influence of factor A on B is expected to be different when in the presence or absence of other intervening or intermediary variables. Thus, when examining the influence of factor A on B, we arrive at contradictory outcomes due to the interaction between the factors that the researcher chose to examine. Yet, contradictory outcomes are not necessarily a result of a methodological mistake in measurement of the phenomenon, rather they reflect the authentic reality of different phenomena that are created through different combinations of variables.
Statistical Significance Does not Prove Causality
Every first year student in the behavioral sciences studies the experimental research model as the scientific method for examining causal relations (Campbell & Stanley, 1966)[10]. This is undertaken by means of activating independent variable/s (IV) on dependent variables (DV) with an experimental group while maintaining the remaining variables as constant or neutral, whether they are relevant or intervening. Any significant change in the DV in comparison with the control group is related to the IV. According to this paradigm, the IV are perceived by the researchers to be the causes of changes in the DV (Breakwell, Hammond, & Fife-Schaw, 2000)[8]. The preference for the experimental paradigm for examination of hypotheses about causality is well-established in the research literature (Feuer, Towne, & Shavelson, 2002)[13], reinforced by respected research centers (National Research Council, 2002)[23], and identified in the literature as the regularity approach that is based on differential analyses of variance theory (Mohr, 1982)[22].
Cause is defined in philosophy as a sufficient and necessary condition for the occurrence of an event: if event A is the cause of event B, then the existence of A is sufficient and necessary for the existence of B. In contrast, statistical significance, however strictly defined, is based on probability. Thus, finding an influence of factor A on factor B at the significance level of 0.01 or even 0.001 is not strong enough a proof that factor A is the cause of factor B, since the explained variance of B by A is still much lower than 100% (n. b. if, for example, the correlation between A and B is 0.60, then the explained variance is still only 36%). This means that factor A had only a partial influence on B and that it is possible that it didn't have any influence on some of those examined or that, even, its influence had the opposite effect on some of them. In addition, statistical significance in an experiment still is not proof that event B did not take place in the absence of factor A. If so, then statistical significance is not proof that the IV is a sufficient and necessary condition for a change in the DV.
Following the theory of Dollard, et al. on the influence of frustration on aggressive behavior (Dollard, Doob, Miller, Mowrer, & Sears, 1939)[11], Barker et al. conducted an experiment in which children in the experimental group waited a certain period of time outside a playroom (Barker, Dembo, & Levin, 1941)[7]. While waiting, the children were able to see the toys, but they did not have access to them. Thus, the researchers created a frustrating situation. After waiting awhile, the children were allowed to enter and to engage in play in the room. At such time, they displayed more aggressive behavior than the control group that had entered the playroom without having to wait. According to the experimental model, the outcomes verified Dollard et al's theory and demonstrated that frustration (IV) caused aggression (DV).
What undermines the certainty claimed regarding the conclusions of this experiment is that in spite of the statistical significance achieved, not all of the children in the second stage of the experiment were aggressive, despite this being a frustrating situation. Some continued to be calm, others cried, or displayed behavior such as avoidance or depression. If frustration is a sufficient factor for aggression, then why were not all of the children aggressive? What in fact happened was that the situation of frustration caused a variety of responses that included aggression, crying, depression, while among other children it did not cause any difference in their behavior.
Other experiments in the area of aggression have shown that aggression can appear as a result of biological, personality, learning, cultural, and many additional factors without being related to frustration. If so, then frustration is not a necessary cause of aggression (Bandura, 1973; Bandura, & Walters, 1963; Feshbach, & Singer, 1971)[4].
Influence of a Factor Depends on the Presence or Absence of Additional Actors
The theory applied by the researcher to examine hypotheses usually guides selection of the factors defined as DV and IV in a research study. Most of the theories in psychology are reductionist in that they focus on a limited number of factors. In psychodynamic theories, the central factors are unconscious drives and processes originating in childhood; in behavioral theories the factors are environmental stimuli and reinforcement received; in humanistic theories, the factors are the self and free desire; while in cognitive theories they are internal thoughts. The same phenomena dealt with by psychology are related to by psychiatry as biological factors and by sociology as social, economic, and political factors. Few of the theories are holistic or systemic. In fact, most researchers select a limited number of factors according to a particular theory and ignore other causes. As a result, the outcome will be depended upon a combination of specific causes selected for the study. This being the case, activating IVs in the experimental group and neutralization of other variables is not neutralizing in any sense. Rather, the researcher creates a combination of factors according to a particular theoretical approach and, by doing so, the researcher creates a phenomenon compatible to the theory assumed. Such an approach deals selectively with part of the phenomenon and is generally reductionist. Accordingly, the inability to generalize the outcomes of experimental research in regard to natural reality is not due to a reasonable limitation of internal validity in the research method, but rather the reductionist method creates a distortion of the phenomenon being studied. Thus, in returning to the topic of aggression, if we examine the influence of viewing violent films on aggression, we see that selecting different combinations of IVs will lead to contradictory outcomes.
Let us consider, for example, a classic experiment in which children are exposed to a film that shows a person hitting a plastic figure. Following viewing of the film, there was a rise in the children's aggressive behavior (Bandura, 1973; Bandura, Ross, & Ross, 1961)[4,5]. Based on this outcome, the researchers concluded that viewing violent films (IV) caused the aggression (DV). A deeper analysis of the outcomes reveals that there were children in this experiment that did imitate the aggressive behavior in the film and who continued to be calm. If so, then here too, viewing violent films cannot be a cause (sufficient and necessary) for aggression. Additionally, it is interesting to note that the influence of viewing violent films on violent behavior depend on many other factors such as reinforcement or punishment for violent behavior. In another series of experiments in regard to the influence of viewing violent films, Bandura et al. (Bandura, Ross, & Ross, 1963)[6] showed children portrayals of violent characters who received positive reinforcement and others who are punished for violent behavior. This time the outcome revealed that there was imitation of the violence, principally, when the violent character received reinforcement. Thus, the influence of viewing violent films depend upon whether or not the violence receives reinforcement. In other words, the influence of violent behavior in a film together with reinforcement is different from its influence together with punishment.
Many additional causes can change the influence of viewing violent films on aggressive behavior. For example, when the influence of films on youth living in institutions was examined, the experimental group that had viewed violent films was found to have less violent behavior than the control group that had viewed non-violent films (Feshbach, & Singer, 1971)[12]. Several of the observer's testimonies revealed that viewing of violent films by youth in institutions led some of them to be tranquil. The explanation offered by researchers for such behavior was that for some of the youth in institutions viewing violent films served as a catharsis, hence they felt less of a need for behavioral expressions of violence. In addition, the influence of viewing violent films was different on different measures of violence, as measured by behavioral measures regarding authority, regarding peers, or in fantasies. The different influence on different measures was depended on personality measures of the youth.
These experiments show that the influence of viewing violent films on aggressive behavior depends upon a number of factors, such as the type of violence presented, the context in which the violence appears in the film, the outcome of the violence in the movie, the personality of the violent character and the degree of attraction to the character, as well as, the justification of the violence in the drama. All of these are in addition to factors related to viewer characteristics, such as age, self-control, values, and emotional state. All combination of variables creates a field or system that changes the outcomes. Thus, persons seeking a simple causal answer to the influence of viewing violent films may arrive at three contradictory conclusions depending upon the combination selected for the study:(1) viewing violent films causes violence (e.g. when viewing violence is accompanied by reinforcement);(2) viewing violent films does not cause violence (e.g. when viewing is accompanied by punishment);(3) viewing violent films reduces violence (e.g. when viewing violence is accompanied by special personality factors of children living in institutions).
In order to understand the contradictory outcomes regarding the influence of violent films we must relate to the relationship between the violent films and the aggressive behavior in the systemic context together with the remaining relevant factors, such as: inherited and biological factors, frustration, previous learning factors of each child, character and personality factors, and social and cultural factors. Here it is important to understand that there is no exclusive influence of viewing violent films as a cause of aggressive behavior. Thus, the contradiction in the outcomes of the influence of violent films on behavior appears is an apparent contradiction because if we would relate it to these outcomes according to the systemic conception, that is beyond simple causality, it would appear that these are authentic outcomes of different viewing phenomena that were created by the researchers. Accordingly, the response to every question that asks - “how does variable A influence B?” is that “it depends upon the presence or absence of variables C, D, E, and so forth.” Similarly, the response to a question, such as “what are the causes of the appearance of B?, is that a very special combination of variables is present.
The Interaction between Factors is no less Important than the Factors Themselves
If the behavior of variable A in the presence of variable B is different from its behavior in the absence of variable B, then that indicates that there is an interaction between them that is, in and of itself, an additional factor to A and B. The presence or absence of certain factors can determine the interactions between the factors. A simple illustration from the world of chemistry can clarify this point. The mutual influence between oxygen and hydrogen is non-identical when particles of sulfur are present or absent. Simple chemistry teaches us that combining oxygen and hydrogen without sulfur creates water (H2O), under certain conditions: sulfur oxide (H2SO4) is created in the presence of sulfur, while another material is created in the presence of sodium (NaOH, sodium hydroxide). The mutual influence between the oxygen and the hydrogen changes over and over in the presence or absence of tens or hundreds of bases and other factors. Every combination of bases will cause another interaction and another outcome. Thus, there can be no influence or causality that is context or system-free since phenomena occur within a system. Again, this is not a case of contradictory results vis-a-vie the mutual influence of the oxygen and hydrogen, as it would appear to the naked eye, but rather in this case different phenomena are created as a result of different interactions.
The interaction between the factors is very important in both the social and natural sciences. In a classic experiment, Schacher & Singer (1962)[30] injected Epinephrine (an artificial substance equivalent to adrenalin) in subjects of an experimental group in order to cause physiological arousal. Subjects in the control group were injected with a placebo that contained a substance that was thought to lack any specific influence on the body. All the subjects were told that they had been injected with a vitamin in order to study its influence on perception. The real goal of the experiment was to examine the influence of ephinephrine on emotions in different cognitive and environmental conditions. There were three cognitive conditions that the researchers created by means of different explanations that were given to the subjects in regard to the substance injected into their bodies. One experimental group was told that the injection could cause physiological arousal, the real effect that ephinephrine does have on the body. A second experimental group was given a misleading explanation, such as: “the injection may cause dizziness, a headache, or itchiness.” Members of the control group were told that the injection (the placebo) has no physiological influence. As part of the experiment, the subjects spent some time in a waiting room where they were asked to complete a questionnaire. The researchers created two environmental conditions by means of an actor present in the waiting room:(1) a cheerful state created by an actor who laughed and played with a paper airplane as he sat in the waiting room; and (2) an angry situation caused by an actor who became upset with the researchers, claimed that the questionnaires were too invasive, tore it up, and left the room (Schachter & Singer, 1962)[30].
As noted above, the real purpose of the experiment was to examine the influence of the epinephrine on feelings when the subjects in the two experimental groups were distinguished in their understanding and expectations regarding the influence of the injection, in the social environment to which they exposed. The outcomes of the experiment revealed significant interaction between the three IVs: epinephrine, the explanation, and the environment. Those who received epinephrine and the false explanation (dizziness, headache, and itchiness) were influenced by the climate that the actors created and reported having feelings that were similar to the same climate: those who were with the cheerful actor reported feeling cheery while those who were with the angry actor reported being angry. In contrast, those who received the truthful explanation regarding the influence of epinephrine and those injected with the placebo were not influenced by either the cheery or the angry actors. Subjects in the control group have not influenced by the placebo nor of the climate.
The influence of each IV on the feelings depended on the other two IVs:(1) the influence of epinephrine on the feelings depended on the contents that they thought about (the explanation) and social environment (the actor) such that each combination created a different outcome;(2) the influence of the social environment depended on the physiological status and expectation based on the explanation provided; and,(3) the influence of thought depended on the physiological status and social environment. If we add that the differences between the groups were based upon statistically significant differences, then it is possible to understand that not all of members of the same group reacted in the same manner and that there were very significant individual differences within the group. These individual differences add to the three variables that the experiment examined another influencing factors such as biological, personality, and situational factors. If so, what conclusions can be drawn about the influence of the epinephrine on the feeling? It is depended on the field (the system) in which the epinephrine acted upon the feeling.
Here is another example: in spite of many experiments that have shown that morphine causes addiction, Alexander, et al. demonstrated in a series of experiments that examined the influence of morphine that while rats living in laboratory cages did become addicted to morphine, a group of rats that lived in a “rat park” (a rich, expansive environment) did not become addicted to it. They also demonstrated that the addicted rats living in the laboratory ceased their morphine use and addiction when they were transferred to the “rat park” (Alexander, 2001, 2002)[1,2]. Here, too, the experiments demonstrated that the influence of the morphine was differential and depended on the environment. If the morphine in and of itself was the cause, then why did it lose its effect in a spacious environment?
The absurdity of searching for simple causality and the need for systemic thinking is clear when searching for causes of complex phenomena such as, causing spring blooming, rain, wars, or divorce. Every answer to these questions requires going beyond the borders of causal thought and adoption of systemic thinking that views the “particulars” as part of the “whole,” and not separate from it. If this is not undertaken, then experimental research that seeks causes to phenomena will arrive at contradictory conclusions.
Through MANOVA analysis it is possible to examine the interactive influence between variables. When there are many variables, many interactions are created between the variables that are themselves new factors that have no less importance than the influence of the variables in and of themselves. In the cases of complex phenomena, such as psychological and social phenomena, the interactions between the variables becomes all the more central and examination of the influence of the single factors becomes secondary. Regretfully, the human brain is only capable of perceiving and understanding the interactions of close to 3 factors only. In other words, the meaning of the interaction between 5 factors, for example, cannot be perceived and, thus, researchers tend to refrain from examining such complex interactions. Accordingly, we remain far from understanding the activity of complex systems and retain the illusion of understanding simple and solitary connections. This is similar to looking under a streetlight because of the difficulty of looking in darker areas. As a result, the knowledge that accumulates from these types of explorations is partial and sometimes contradictory. In seeking to understand a system, we should not be deluded into thinking that discovery of all solitary connections between the variables enables us to understand the complete system of interactions. We need to remember that the sum of all the parts is not identical to the whole.
It is insufficient when researching a complex phenomenon to point to a group of factors that are the “cause” or “causes” for creation of the phenomenon. For example, it is not sufficient to state that aggression is caused by frustration, imitation, reinforcements, thinking, character, heredity, biology, poverty, culture, and lack of education. All of these factors are components of a system that acts in a very special manner. The medical and psychological literature that discuss diseases and syndromes are complete with series of causes included under a category, such “etiology” or “causal factors” which are in fact a collection of factors connected to the disease. For example, the literature indicates that the causes of heart disease include inheritance, gender, diet, smoking, physical exercise, personality type, mental pressure, and many other additional causes (Suchday, Tucker, Krantz, 2002; Welicker, 2002)[33,36].
Research that focuses on identification of the factors involved in a system connected to diseases is only a small step toward understanding the system. In order to understand the phenomenon, we need to understand the interaction between different combinations of factors and the regularity according to which the combination is connected to aggression or heart disease. Usually it is not the case that one from among them is the sole cause that has proven to be the sufficient and necessary for creation of the disease. Even when there is a sole cause (as in the case of the AIDs virus), this cause can be neutralized in the presence of another factor, such as safe behavior or a preventive medicine. On the other hand, we should not forget that the phenomenon (or disease) can take place in the presence of a partial combination of factors, as seen in studies on aggression that were mentioned earlier in regards to frustration, imitation, reinforcements, and character. In addition, we need to remember that aggression is part of a system and exists in interaction and in itself influences the other factors. For example, it could cause frustration, provide reinforcements, and influence the remainder of the factors in the system.
When examining the causal relation between some of the factors connected to a system, while neutralizing or controlling other factors, we neutralize meaningful interaction between meaningful variables that are not marginal “intervening variables,” but rather they are the “cause.” If factors A, B, C, and D are the cause of variable “E”, then we are unable to examine specifically how factor A influences factor E. Sayer (2000)[29] even claimed that the attempt to neutralize selected variables within a context in which an event takes place damages and distorts the causal processes.
Simple Causality versus Causality as Systemic
Having seen the limitations of experimental research in discovery of the causes and hence its limitations in understanding systems, we proceed with a discussion of the concept of causality as well as of the contributions and limitations of causal thinking.
What causes bodies to fall? What causes a forest fire? What causes people to succeed? Why do people commit suicide? All these questions have occupied researchers and scientists. This line of questioning assumes that there is one or a number of causes for every event. Such an assumption is very much consistent with the mechanistic school of thought of Newton that proposed mechanistic laws of physics. The claim that every action has a corresponding opposing action encouraged researchers initially in the natural sciences and then later in the social sciences to identify the driving force (i.e. cause) for every event.
Thus far cause has been defined as a necessary and sufficient cause for an event to take place. In this regard, we need to remember that a literature exists that claims that there is no way to define what is a necessary and sufficient, and that this principle is misleading (Brennan, Fall 2003)[9]. The primary limitation of this principle is revealed when it is applied to attempts to understand how events take place and their causes. Even in such a simple event as spilling a glass, it is difficult to find one necessary and sufficient cause for its falling and shattering when it hits the floor. Is the cause the force that drove the glass off the table; is it gravity; the height of the table; the materials that make up the glass; the floor? Or, is it the absence of additional causes, such as: a force that will restrain the glass and prevent it from falling; a flexible way of holding the glass; a flexible floor? All these are causes that in different combinations determine the outcome. For example, the result of the fall of a thin glass from high distance onto a stiff floor will be different from the fall of a thick glass from a low height onto a dirt floor. Even when we identify a sufficient cause of an event, this is still not its sole cause. For example, if a shot fired in the head it is sufficient to cause death, yet death can still be caused by many other causes without a shot hitting a head. Among the causes that are known in nature and in society, it is difficult to find one cause that is necessary and sufficient of a particular event. Events always take place due to a combination of causes.
David Hume, the Scottish philosopher, became famous due to his theory about the question of causality (in Magee, 1987)[19]. He claimed that even if event “A” always appears before event “B,” even when event “B” always happens after event “A,” we cannot conclude that event “A” is the cause (a necessary a sufficient cause) for event “B” to take place. For example, “evening” takes place prior to “daytime” that always takes place after the evening. In spite of this, we cannot claim that the “evening” causes “daytime,” or the opposite. According to Hume, it is impossible to observe a causal connection in nature. We are only able to observe necessary regularity or proximity between events. Our mind is what makes the connection between regularity and causality. The German philosopher, Immanuel Kant (1969)[17], also claimed that causality is a characteristic of the mind and does not exist in nature.
In my opinion, the causality produced by the mind is too simple way of understanding complex phenomena in nature and in society, for a human mind that is incapable of such perception. Every system is composed of a number of factors that simultaneously interact in multiple directions. Further, every system is part of another system that is composed of other systems whose inter-connection is simultaneously multi-directional. Systemic perception is necessary when we are speaking about systems such as our physical body, a social system such as the family, or ecological systems. For example, the human body is a continuous and developing multi-dimensional, interactive system that links thinking, feeling, behavior, personality, heredity, biology, and environment (Bernard, & Krupat, 1994; Forgas, 2001; Sarafino, 2002)[3,28]. We need to remember that even the most minuscule phenomena such as cell is in and of itself a system in which very complex, multi-directional connections exist simultaneously. Similarly, neither is the atom inorganic, rather it is itself a dynamic energy field within which complex actions by electrons, protons, and neutrons take place in equilibrium within a special interactive system that exists internally within the atom as well as externally with its surroundings.
In juxtaposition to linear-causal thinking, there are theories that claim that the processes in nature and in society have a cyclical-rhythmic nature (Lloyd, & Rossi, 1992; Rossi, 1986)[18,25]. For example, the heart beat, blood circulation, the day and the night, the tides, the cycle of water in nature, the life cycle, historical periods, and the movement of the planets. Indeed, that which appears to be linear-serial is only part of a very complex cyclical system. Maxwell (2004)[20] criticized the regularity approach in research and variance theory (Mohr, 1982)[22] that seek the influence of the variables, and proposes the approach of scientific realism and process theory (Salmon, 1984, 1998; Sayer, 2000)[26,27,29] that deals with events and causal processes that take place between the incidents. This is a systemic approach that deals with processes in the context in which the incidents take place and seeks the causal mechanisms (Maxwell, 2004)[20].
In summary, the linear-serial and reductionist approach in scientific research is an attempt to adapt complex phenomena to the limitations of our minds. Causality is a cognitive structure or characteristic of the human mind that enables us to perceive and to interpret what is observed, but it is not part of the observed. There is no doubt that this approach assisted scientists to achieve a partial understanding of phenomena and to make relatively giant advancements in science, but such are minuscule as we proceed down a long and obscure path. We should not delude ourselves and believe that the causal connections observed actually exist in nature or believe that nature or society is a simple sum of the linear connections discovered to date. With our approaching understanding of systems in nature and in society through causal thinking, we need to be aware as well of the distortion and confusion that causality causes in most of the domains of knowledge (Sayer, 2000)[29]. Part of the confusion is recognizable, as noted, in the accumulation of contradictory outcomes in nearly all research that deals with environmental factors in regard to psychological, medical, and social phenomena.
The Systemic Approach
Science developed in three stages: The first stage started with Greek philosophy and continued through the Middle Ages when scientists sought to understand the characteristics of matter and the movement of bodies in relation to their own characteristics. In the mechanistic thinking that developed in the second stage, beginning with Galileo and Newton in the 17th century, the concept of “power” was recognized as a “cause” of the movement of bodies. Since then equations and scientific laws were developed to explain the causal connections between factors. The “field” theory that developed in the third stage (20th century) perceived events in their totality such that individual parts are identified quantitatively and qualitatively only in terms of their relation with other parts (Frank, 1989; Hawking, 1989; Smith, & Smith, 1996)[14,15,32]. Smith and Smith (1996)[32] claim that while physics developed over these three stages, psychology remains in the second, the mechanistic, stage. In their view, aside from a few exceptional cases such as Kurt Lewin's, Gestalt theory and other ecological theories, psychology has remained at the level of investigation of single causation, explanation by way analogy, biological reduction, or psycho-physio dualism. In their view, “field theory” is the necessary and sufficient condition for understanding psychological events and processes (ibid). Though the influence of mechanistic physics on causal thinking in the social science is well known among the different domains of the natural sciences, what is less noticed is the influence of chemistry and biology that had significant achievements due to adoption of more systemic, rather than mechanistic thought. The science of chemistry is characterized more for its understanding of interactions between the bases and less for causal influence of a base on another base. The science of biology too is characterized by its systemic understanding of the body and less for its search for simple causality. Similarly, psychology could realize great benefits if it too would learn from the systemic approach of chemistry and biology.
The Quantum Field Theory in physics that appeared during the era of Albert Einstein's relativity theory provided an answer to the apparent contradiction between Newtonian laws of gravity that dealt with mass on one hand, and laws of electromagnetism and energy that were discovered later on the other. It did not negate Newtonian laws of mechanics, but rather ground them together with the laws of electromagnetism in one field that are explained by means of uniform laws (Kaku, 1993)[16]. Similarly, systemic approaches in the social sciences do not negate the existence of causal connections in nature and in society, but see all the causal connections between the single factors as dynamic and changing components within the system. The systemic approach has great importance since it is capable of unifying worlds of research and disciplines that appear separate or contradictory in an all-encompassing systemic conception.
Conventional psychological research deals with the mind separate from the body (that is dealt with by medicine) and in distinction from society and culture (the domains of sociology). This is in and of itself a form of reductionism that artificially divides the “field” or the system into separate parts. Within each domain of study, the impact of reductionism is much stronger and research is reduced to smaller and smaller components. For example, in medicine there are specializations and separate fields of research for nearly every system and organ in the body (heart, brain, skin, ear-nose-throat, etc). Psychological theories, as well, focus on partial factors in the psychological experience. In contrast, a systemic approach, one that views psychological theories as components of one system, is capable of resolving the apparent contradictions between the single-cause theories.
Though systemic theories do exist in psychology, they function primarily in the field of treatment and intervention and less in the field of research. For example, family therapy theories attempt to understand the individual in the context of the familial system. Similarly, some intervention approaches in educational psychology view a pupil's problems as part of the school as a system and attempt to intervene in the system. These are relative systemic theories since they place less emphasis on biological factors, on the one hand, and social-ecological factors, on the other hand. It may be the case that the bio-psycho-social approach in psychology and medicine (Sarafino, 2002, Turk, 1996)[28,34] is a more promising systemic approach since it views biological, psychological, and social phenomena as one general system.
Grounding Experimental Research in the Systemic Perspective
Though the limitations of simple causal thinking are more salient in the social sciences than in physics, it was physics that led the confrontation with them by means of field theory, when as noted it ground Newton's laws of mechanics in a more general field. Such a move is all the more necessary in the social sciences in order that systemic approaches can ground the experimental approach of cause and outcome. In this regard Gestalt theories teach us that a connection exists between the part and the whole. This is not a matter of right or not right, as both are correct. That is, we need to recognize that a part is but part of a whole and not apply it to the whole; and similarly we need to be aware that the whole is different from the sum of its parts.
Qualitative research, action research, field research, and case study that seek to understand a system within its natural context are no less scientific than experimental studies that deal with questions regarding the influence of a certain factor on others. In my view, the scientific nature of research is not measured according to its degree of accuracy and objectivity, but rather in the degree of its appropriateness to natural and social systems. In this regard, it is possible that an experimental study is exceptionally accurate in regard to validity and internal reliability, but lacks external validity and does not provide us with a real understanding of the system. In this regard we need to realize that tests of validity and reliability need to be undertaken in relation to the systems in nature and in society and not in relation to a select number of variables or to an artificial experimental structure chosen by the researcher according to certain criteria.
Maxwell (2004)[20] claims that qualitative research is capable of providing more valid answers to questions regarding causality than experimental research. In his view, juxtaposed to experimental research that seeks causality in isolated variables while ignoring the context, qualitative research is capable of providing answers regarding the causal processes as well as the mechanisms of the occurrence of phenomena in the events' context.
Further, it appears that meta-analysis research that studies all the research outcomes in a particular field can serve as the basis for grounding apparent contradictory outcomes of different experiments in a more general systemic view. Meta-analysis research that analyzes tens of specific studies that differ in methodology and in the combination of variables selected for study within a general phenomenon is capable of contributing to the understanding of systems in nature or in society and to provide a satisfactory answer to the problem of “contradictory outcomes.” In addition, in order to overcome the limitations of reductionism, there is also a need to expand research in inter-disciplinary fields. Such research opens up the possibility for mutual enrichment between different fields of knowledge and contributes to illumination of the murky areas that exist between domains. What remains to be recognized is the need to strengthen the relation between scientific research in different fields and philosophy that serves as an overarching framework that is capable of retaining the overview of the causal relations within a more general system.
In summary, I am not claiming that is possible or needed to discontinue examining the influence of causes through experimental research. My claim is that we should not forget that the regularity of the character of nature and society is far greater than the causal connections that our mind is capable of understanding and that examination of simple connections is a reduction of more complex phenomenon that can result in what appear to be contradictory outcomes. We should not forget that the behavior of variables when they are neutralized from other variables is not identical to their behavior when they exist within the system. Accordingly, research of questions regarding causality must focus on the causal mechanism by applying a dynamic-systemic view, cooperation between quantitative and qualitative research, different fields of knowledge, and use of meta-analysis and retention of a general philosophic perspective.
Acknowledgement：
I wish to thank my colleagues - Prof. Michael Moore, Prof. Yeshiyahu Tadmor, Dr. Avi Avital, Dr. Miriam Welicker, Dr. Sivan Raz, and Balig Abdu - for interesting comments that contributed to the ideas and presentation of this article.
REFERENCES
1. Alexander, R. B. (2001). The myth of drug-induced addiction. A paper delivered to the Canadian senate, January, 2001, www.parl.gc.ca/37/1/parlbus/commbus/senate/com-e/ill e-e/presentation-e. Retrieved December 12, 2004.
2. Alexander, R. B.Arresting addictions: Drug education and relapse prevention in corrections. Lanham, MD: American Correctional Association, 2002.
3. Bernard, L. C., & Krupat, E. Health psychology: Biopsychosocial factors in health and illness. New York: Harcourt Brace College Publishers, 1994.
4. Bandura, A.Aggression: A social learning analysis. Englewood Cliffs, NJ: Prentice Hall,1973.
5. Bandura, A., Ross, D., & Ross, S. A.Transition of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology,1961,63(3):575-582.
6. Bandura, A., Ross, D., & Ross, S. A. Imitation of film-mediated aggressive model. Journal of Abnormal and Social Psychology,1963,66:3-11.
7. Barker, R., Dembo, t. & Levin, K. (1941). Frustration and aggression: An experiment with young children. University of Iowa Studies in Child Welfare.
8. Breakwell, G. M., Hammond, S., & Fife-Schaw, C.(Eds). Research methods in psychology. 2nd ed. London: Sage Publications,2000.
9. Brennan, A. (Fall 2003). Necessary and Sufficient Conditions, The Stanford Encyclopedia of Philosophy (Fall 2003 Edition), Edward N. Zalta (ed.), URL =
10. Campbell, D. R., & Stanley, J. C.Experimental and quasi-experimental designs for research. New York: Rand McNally,1966.
11. Dollard, J., Doob, L., Miller, N. E., & Mowrer, O. H., Sears, R. Frustration and aggression. New Haven, CT: Yale University Press,1939.
12. Feshbach, S. & Singer, R. D.Television and aggression. San Francisco: Jossy-Bass,1971.
13. Feuer, M. J., Towne, L., & Shavelson, R. J. Scientific culture and educational research. Educational Researcher,2002,31(8):4-14.
14. Frank, P.Einstein: His life and times. New York: Da Capo Press, Inc. (Originally published in 1953),1989.
15. Hawking, S.A brief history of time. New York: Bantam,2005.
16. Kaku, M.Quantum field theory. New York: Oxford University Press,1993.
17. Kant, I.Critique of pure reason. New York: St. Martin's Press. (Originally published, 1781),1969.
18. Lloyd, D., & Rossi, E.,(Eds.). Ultradian rhythms in life processes: A fundamental inquiry into chronobiology and psychobiology. New York: Springer-Verlag,1992.
19. Magee, B. The great philosophers: An introduction to Western philosophy. New York: Oxford Paperbacks,2000.
20. Maxwell, J. A.Causal explanation, qualitative research, and scientific inquiry in education. Educational Researcher,2004，33(2), 3-11.
21. Meichenbaum, D. A clinical handbook: Practical therapist manual for assessing and treating adults with post-traumatic stress disorder (PTSD). Ontario, Canada: Institute Press，2001.
22. Mohr, L. B.Explaining organizational behavior. San Francisco: Jossey-Bass，1982.
23. National Research Council. Scientific research in education. Washington, DC: National Academy Press，2002.
24. Orner, R. J. & Schnyder, U.Reconstructing early intervention after trauma. Oxford: Oxford University Press，2003.
25. Rossi, E. Altered states of consciousness in everydaylife: The ultradian rhythms. In B. Wolman & M. Ullman (Eds.), Handbook of altered states on consciousness. New York: Van Nostrand，1986，97-132.
26. Salmon, W. C.Scientific explanation and the causal structure of the world. Princeton, NJ: Princeton University Press，1984.
27. Salmon, W. C. Causality and explanation. New York: Oxford University Press，1998.
28. Sarafino, E. P.Health psychology: Biopsychosoial interactions. New York: John Wiley & Sons, Inc，2002.
29. Sayer, A.Realism and social science. London: Sage Publications，2000.
30. Shachter, S., & Singer, J. H.Cognitive social and physiological determinants of emotional state. Psychological Review,1962，69, 379-397. 1,
31. Smith, T. W., & Gallo, L. C.Personality trait as risk factors for physical illness. In A. W. Baum, T. A. Revenson, & J. E. Singer, (Eds). Handbook of health psychology. New York: Lawrence Erlbaum Associates, Inc,2001.
32. Smith, N. W., & Smith L. L.Field theory in science: Its role as a necessary and sufficient condition in psychology. Psychological Reports,1996，46:3-19.
33. Suchday, S., Tucker, D. L., & Krantz, D. S.Diseases，2002.of the circulatory system. In T. J. Boll.Handbook of clinical health psychology. Washington, DC.: American Psychological Association，2002.
34. Turk, D. C.Biopsychosocial perspective on chronic pain. In R. J. Gatchel, and D. C. Turk (Eds.). Psychological Approaches to Pain Management: A Practitioner's Handbook,1996,3-32.
35. Turk, D. C. Physiological and psychological bases of pain. In A. W. Baum, T. A. Revenson, & J. E. Singer, (Eds). Handbook of health psychology. New York: Lawrence Erlbaum Associates, Inc，2001.
36. Welicker, M.Labria'ut mekol halev [Heartfelt Health]. Haifa: Department for Teaching Technology and Science, The Technion & Jerusalem: Curriculum Department, Ministry of Education，2002.
(Editor Jaque)(Marwan Dwairy)
Emeq Yezreel Academic College & Oranim Academic College
Correspondence to Marwan Dwairy, D.Sc. Associate Professor, Ora Street 3b, P.O. Box 14710, Nazerat Ellit 17000, Israel
Tel:Work: 972 46561541, Mobile:972 054 6334515, E-mail:psy@marwandwairy.com
[Abstract] A considerable number of empirical studies that examined the causal relationships between factors arrive at mixed or contradictory results. Typically, these results are attributed to methodological differences. This article proposes that contradictory results are inherent in the simple causality and reductionistic approach that dominates the experimental paradigm. Such results created because the influence of factor A on B is expected to be different when in the presence or absence of other intervening or intermediary variables. Our claim is that thinking in terms of causality is too simple way of understanding complex phenomena in nature and in society. Research of questions regarding causality must focus on the causal mechanism by applying a dynamic-systemic view, cooperation between quantitative and qualitative research, different fields of knowledge, and use of meta-analysis and retention of a general philosophic perspective.
[Key words] causality;contradictory outcomes;research
A considerable number of empirical studies that examined the influence of one or more factors on other factor/s arrive at mixed even contradictory outcomes. The phenomenon of “contradictory outcomes” is common in nearly all research areas in psychology and medicine. For example, studies that examined the influence of personality factors on health (Smith and Gallo, 2001)[31], thinking on pain (Turk, 1996, 2001)[34,35], debriefing on persons exposed to traumas (Meichenbaum, 2001; Orner, & Schnyder, 2003)[21,24], and viewing of violent films on violent behavior (Bandura, 1973, Feshbach, & Singer, 1971)[4,12] arrived at contradictory outcomes. Such outcomes appear in the following manner: Certain studies reveal that factor “A” amplifies “B,” while other research found that factor A decreases B, and yet other studies found that there is no influence by A on B. The common tendency in literature is to assume that contradictory outcomes are due to methodological differences that exist between the research studies: differences in the definition of variables, in measurement tools, in the sample, type of analysis, and so forth. This article proposes a more basic explanation for the phenomenon of contradictory outcomes, one that relates it specifically to the causal and reductionist forms of thinking that characterize experimental research paradigms in science, in general, and in the social and behavioral sciences, in particular. In effect, these methodological differences create substantive differences in the field or in the system of the phenomenon and create different combinations of variables, such that the outcomes received are contradictory in regard to the influence of independent variables on the dependent variables. In my opinion, contradictory results are created because the influence of factor A on B is expected to be different when in the presence or absence of other intervening or intermediary variables. Thus, when examining the influence of factor A on B, we arrive at contradictory outcomes due to the interaction between the factors that the researcher chose to examine. Yet, contradictory outcomes are not necessarily a result of a methodological mistake in measurement of the phenomenon, rather they reflect the authentic reality of different phenomena that are created through different combinations of variables.
Statistical Significance Does not Prove Causality
Every first year student in the behavioral sciences studies the experimental research model as the scientific method for examining causal relations (Campbell & Stanley, 1966)[10]. This is undertaken by means of activating independent variable/s (IV) on dependent variables (DV) with an experimental group while maintaining the remaining variables as constant or neutral, whether they are relevant or intervening. Any significant change in the DV in comparison with the control group is related to the IV. According to this paradigm, the IV are perceived by the researchers to be the causes of changes in the DV (Breakwell, Hammond, & Fife-Schaw, 2000)[8]. The preference for the experimental paradigm for examination of hypotheses about causality is well-established in the research literature (Feuer, Towne, & Shavelson, 2002)[13], reinforced by respected research centers (National Research Council, 2002)[23], and identified in the literature as the regularity approach that is based on differential analyses of variance theory (Mohr, 1982)[22].
Cause is defined in philosophy as a sufficient and necessary condition for the occurrence of an event: if event A is the cause of event B, then the existence of A is sufficient and necessary for the existence of B. In contrast, statistical significance, however strictly defined, is based on probability. Thus, finding an influence of factor A on factor B at the significance level of 0.01 or even 0.001 is not strong enough a proof that factor A is the cause of factor B, since the explained variance of B by A is still much lower than 100% (n. b. if, for example, the correlation between A and B is 0.60, then the explained variance is still only 36%). This means that factor A had only a partial influence on B and that it is possible that it didn't have any influence on some of those examined or that, even, its influence had the opposite effect on some of them. In addition, statistical significance in an experiment still is not proof that event B did not take place in the absence of factor A. If so, then statistical significance is not proof that the IV is a sufficient and necessary condition for a change in the DV.
Following the theory of Dollard, et al. on the influence of frustration on aggressive behavior (Dollard, Doob, Miller, Mowrer, & Sears, 1939)[11], Barker et al. conducted an experiment in which children in the experimental group waited a certain period of time outside a playroom (Barker, Dembo, & Levin, 1941)[7]. While waiting, the children were able to see the toys, but they did not have access to them. Thus, the researchers created a frustrating situation. After waiting awhile, the children were allowed to enter and to engage in play in the room. At such time, they displayed more aggressive behavior than the control group that had entered the playroom without having to wait. According to the experimental model, the outcomes verified Dollard et al's theory and demonstrated that frustration (IV) caused aggression (DV).
What undermines the certainty claimed regarding the conclusions of this experiment is that in spite of the statistical significance achieved, not all of the children in the second stage of the experiment were aggressive, despite this being a frustrating situation. Some continued to be calm, others cried, or displayed behavior such as avoidance or depression. If frustration is a sufficient factor for aggression, then why were not all of the children aggressive? What in fact happened was that the situation of frustration caused a variety of responses that included aggression, crying, depression, while among other children it did not cause any difference in their behavior.
Other experiments in the area of aggression have shown that aggression can appear as a result of biological, personality, learning, cultural, and many additional factors without being related to frustration. If so, then frustration is not a necessary cause of aggression (Bandura, 1973; Bandura, & Walters, 1963; Feshbach, & Singer, 1971)[4].
Influence of a Factor Depends on the Presence or Absence of Additional Actors
The theory applied by the researcher to examine hypotheses usually guides selection of the factors defined as DV and IV in a research study. Most of the theories in psychology are reductionist in that they focus on a limited number of factors. In psychodynamic theories, the central factors are unconscious drives and processes originating in childhood; in behavioral theories the factors are environmental stimuli and reinforcement received; in humanistic theories, the factors are the self and free desire; while in cognitive theories they are internal thoughts. The same phenomena dealt with by psychology are related to by psychiatry as biological factors and by sociology as social, economic, and political factors. Few of the theories are holistic or systemic. In fact, most researchers select a limited number of factors according to a particular theory and ignore other causes. As a result, the outcome will be depended upon a combination of specific causes selected for the study. This being the case, activating IVs in the experimental group and neutralization of other variables is not neutralizing in any sense. Rather, the researcher creates a combination of factors according to a particular theoretical approach and, by doing so, the researcher creates a phenomenon compatible to the theory assumed. Such an approach deals selectively with part of the phenomenon and is generally reductionist. Accordingly, the inability to generalize the outcomes of experimental research in regard to natural reality is not due to a reasonable limitation of internal validity in the research method, but rather the reductionist method creates a distortion of the phenomenon being studied. Thus, in returning to the topic of aggression, if we examine the influence of viewing violent films on aggression, we see that selecting different combinations of IVs will lead to contradictory outcomes.
Let us consider, for example, a classic experiment in which children are exposed to a film that shows a person hitting a plastic figure. Following viewing of the film, there was a rise in the children's aggressive behavior (Bandura, 1973; Bandura, Ross, & Ross, 1961)[4,5]. Based on this outcome, the researchers concluded that viewing violent films (IV) caused the aggression (DV). A deeper analysis of the outcomes reveals that there were children in this experiment that did imitate the aggressive behavior in the film and who continued to be calm. If so, then here too, viewing violent films cannot be a cause (sufficient and necessary) for aggression. Additionally, it is interesting to note that the influence of viewing violent films on violent behavior depend on many other factors such as reinforcement or punishment for violent behavior. In another series of experiments in regard to the influence of viewing violent films, Bandura et al. (Bandura, Ross, & Ross, 1963)[6] showed children portrayals of violent characters who received positive reinforcement and others who are punished for violent behavior. This time the outcome revealed that there was imitation of the violence, principally, when the violent character received reinforcement. Thus, the influence of viewing violent films depend upon whether or not the violence receives reinforcement. In other words, the influence of violent behavior in a film together with reinforcement is different from its influence together with punishment.
Many additional causes can change the influence of viewing violent films on aggressive behavior. For example, when the influence of films on youth living in institutions was examined, the experimental group that had viewed violent films was found to have less violent behavior than the control group that had viewed non-violent films (Feshbach, & Singer, 1971)[12]. Several of the observer's testimonies revealed that viewing of violent films by youth in institutions led some of them to be tranquil. The explanation offered by researchers for such behavior was that for some of the youth in institutions viewing violent films served as a catharsis, hence they felt less of a need for behavioral expressions of violence. In addition, the influence of viewing violent films was different on different measures of violence, as measured by behavioral measures regarding authority, regarding peers, or in fantasies. The different influence on different measures was depended on personality measures of the youth.
These experiments show that the influence of viewing violent films on aggressive behavior depends upon a number of factors, such as the type of violence presented, the context in which the violence appears in the film, the outcome of the violence in the movie, the personality of the violent character and the degree of attraction to the character, as well as, the justification of the violence in the drama. All of these are in addition to factors related to viewer characteristics, such as age, self-control, values, and emotional state. All combination of variables creates a field or system that changes the outcomes. Thus, persons seeking a simple causal answer to the influence of viewing violent films may arrive at three contradictory conclusions depending upon the combination selected for the study:(1) viewing violent films causes violence (e.g. when viewing violence is accompanied by reinforcement);(2) viewing violent films does not cause violence (e.g. when viewing is accompanied by punishment);(3) viewing violent films reduces violence (e.g. when viewing violence is accompanied by special personality factors of children living in institutions).
In order to understand the contradictory outcomes regarding the influence of violent films we must relate to the relationship between the violent films and the aggressive behavior in the systemic context together with the remaining relevant factors, such as: inherited and biological factors, frustration, previous learning factors of each child, character and personality factors, and social and cultural factors. Here it is important to understand that there is no exclusive influence of viewing violent films as a cause of aggressive behavior. Thus, the contradiction in the outcomes of the influence of violent films on behavior appears is an apparent contradiction because if we would relate it to these outcomes according to the systemic conception, that is beyond simple causality, it would appear that these are authentic outcomes of different viewing phenomena that were created by the researchers. Accordingly, the response to every question that asks - “how does variable A influence B?” is that “it depends upon the presence or absence of variables C, D, E, and so forth.” Similarly, the response to a question, such as “what are the causes of the appearance of B?, is that a very special combination of variables is present.
The Interaction between Factors is no less Important than the Factors Themselves
If the behavior of variable A in the presence of variable B is different from its behavior in the absence of variable B, then that indicates that there is an interaction between them that is, in and of itself, an additional factor to A and B. The presence or absence of certain factors can determine the interactions between the factors. A simple illustration from the world of chemistry can clarify this point. The mutual influence between oxygen and hydrogen is non-identical when particles of sulfur are present or absent. Simple chemistry teaches us that combining oxygen and hydrogen without sulfur creates water (H2O), under certain conditions: sulfur oxide (H2SO4) is created in the presence of sulfur, while another material is created in the presence of sodium (NaOH, sodium hydroxide). The mutual influence between the oxygen and the hydrogen changes over and over in the presence or absence of tens or hundreds of bases and other factors. Every combination of bases will cause another interaction and another outcome. Thus, there can be no influence or causality that is context or system-free since phenomena occur within a system. Again, this is not a case of contradictory results vis-a-vie the mutual influence of the oxygen and hydrogen, as it would appear to the naked eye, but rather in this case different phenomena are created as a result of different interactions.
The interaction between the factors is very important in both the social and natural sciences. In a classic experiment, Schacher & Singer (1962)[30] injected Epinephrine (an artificial substance equivalent to adrenalin) in subjects of an experimental group in order to cause physiological arousal. Subjects in the control group were injected with a placebo that contained a substance that was thought to lack any specific influence on the body. All the subjects were told that they had been injected with a vitamin in order to study its influence on perception. The real goal of the experiment was to examine the influence of ephinephrine on emotions in different cognitive and environmental conditions. There were three cognitive conditions that the researchers created by means of different explanations that were given to the subjects in regard to the substance injected into their bodies. One experimental group was told that the injection could cause physiological arousal, the real effect that ephinephrine does have on the body. A second experimental group was given a misleading explanation, such as: “the injection may cause dizziness, a headache, or itchiness.” Members of the control group were told that the injection (the placebo) has no physiological influence. As part of the experiment, the subjects spent some time in a waiting room where they were asked to complete a questionnaire. The researchers created two environmental conditions by means of an actor present in the waiting room:(1) a cheerful state created by an actor who laughed and played with a paper airplane as he sat in the waiting room; and (2) an angry situation caused by an actor who became upset with the researchers, claimed that the questionnaires were too invasive, tore it up, and left the room (Schachter & Singer, 1962)[30].
As noted above, the real purpose of the experiment was to examine the influence of the epinephrine on feelings when the subjects in the two experimental groups were distinguished in their understanding and expectations regarding the influence of the injection, in the social environment to which they exposed. The outcomes of the experiment revealed significant interaction between the three IVs: epinephrine, the explanation, and the environment. Those who received epinephrine and the false explanation (dizziness, headache, and itchiness) were influenced by the climate that the actors created and reported having feelings that were similar to the same climate: those who were with the cheerful actor reported feeling cheery while those who were with the angry actor reported being angry. In contrast, those who received the truthful explanation regarding the influence of epinephrine and those injected with the placebo were not influenced by either the cheery or the angry actors. Subjects in the control group have not influenced by the placebo nor of the climate.
The influence of each IV on the feelings depended on the other two IVs:(1) the influence of epinephrine on the feelings depended on the contents that they thought about (the explanation) and social environment (the actor) such that each combination created a different outcome;(2) the influence of the social environment depended on the physiological status and expectation based on the explanation provided; and,(3) the influence of thought depended on the physiological status and social environment. If we add that the differences between the groups were based upon statistically significant differences, then it is possible to understand that not all of members of the same group reacted in the same manner and that there were very significant individual differences within the group. These individual differences add to the three variables that the experiment examined another influencing factors such as biological, personality, and situational factors. If so, what conclusions can be drawn about the influence of the epinephrine on the feeling? It is depended on the field (the system) in which the epinephrine acted upon the feeling.
Here is another example: in spite of many experiments that have shown that morphine causes addiction, Alexander, et al. demonstrated in a series of experiments that examined the influence of morphine that while rats living in laboratory cages did become addicted to morphine, a group of rats that lived in a “rat park” (a rich, expansive environment) did not become addicted to it. They also demonstrated that the addicted rats living in the laboratory ceased their morphine use and addiction when they were transferred to the “rat park” (Alexander, 2001, 2002)[1,2]. Here, too, the experiments demonstrated that the influence of the morphine was differential and depended on the environment. If the morphine in and of itself was the cause, then why did it lose its effect in a spacious environment?
The absurdity of searching for simple causality and the need for systemic thinking is clear when searching for causes of complex phenomena such as, causing spring blooming, rain, wars, or divorce. Every answer to these questions requires going beyond the borders of causal thought and adoption of systemic thinking that views the “particulars” as part of the “whole,” and not separate from it. If this is not undertaken, then experimental research that seeks causes to phenomena will arrive at contradictory conclusions.
Through MANOVA analysis it is possible to examine the interactive influence between variables. When there are many variables, many interactions are created between the variables that are themselves new factors that have no less importance than the influence of the variables in and of themselves. In the cases of complex phenomena, such as psychological and social phenomena, the interactions between the variables becomes all the more central and examination of the influence of the single factors becomes secondary. Regretfully, the human brain is only capable of perceiving and understanding the interactions of close to 3 factors only. In other words, the meaning of the interaction between 5 factors, for example, cannot be perceived and, thus, researchers tend to refrain from examining such complex interactions. Accordingly, we remain far from understanding the activity of complex systems and retain the illusion of understanding simple and solitary connections. This is similar to looking under a streetlight because of the difficulty of looking in darker areas. As a result, the knowledge that accumulates from these types of explorations is partial and sometimes contradictory. In seeking to understand a system, we should not be deluded into thinking that discovery of all solitary connections between the variables enables us to understand the complete system of interactions. We need to remember that the sum of all the parts is not identical to the whole.
It is insufficient when researching a complex phenomenon to point to a group of factors that are the “cause” or “causes” for creation of the phenomenon. For example, it is not sufficient to state that aggression is caused by frustration, imitation, reinforcements, thinking, character, heredity, biology, poverty, culture, and lack of education. All of these factors are components of a system that acts in a very special manner. The medical and psychological literature that discuss diseases and syndromes are complete with series of causes included under a category, such “etiology” or “causal factors” which are in fact a collection of factors connected to the disease. For example, the literature indicates that the causes of heart disease include inheritance, gender, diet, smoking, physical exercise, personality type, mental pressure, and many other additional causes (Suchday, Tucker, Krantz, 2002; Welicker, 2002)[33,36].
Research that focuses on identification of the factors involved in a system connected to diseases is only a small step toward understanding the system. In order to understand the phenomenon, we need to understand the interaction between different combinations of factors and the regularity according to which the combination is connected to aggression or heart disease. Usually it is not the case that one from among them is the sole cause that has proven to be the sufficient and necessary for creation of the disease. Even when there is a sole cause (as in the case of the AIDs virus), this cause can be neutralized in the presence of another factor, such as safe behavior or a preventive medicine. On the other hand, we should not forget that the phenomenon (or disease) can take place in the presence of a partial combination of factors, as seen in studies on aggression that were mentioned earlier in regards to frustration, imitation, reinforcements, and character. In addition, we need to remember that aggression is part of a system and exists in interaction and in itself influences the other factors. For example, it could cause frustration, provide reinforcements, and influence the remainder of the factors in the system.
When examining the causal relation between some of the factors connected to a system, while neutralizing or controlling other factors, we neutralize meaningful interaction between meaningful variables that are not marginal “intervening variables,” but rather they are the “cause.” If factors A, B, C, and D are the cause of variable “E”, then we are unable to examine specifically how factor A influences factor E. Sayer (2000)[29] even claimed that the attempt to neutralize selected variables within a context in which an event takes place damages and distorts the causal processes.
Simple Causality versus Causality as Systemic
Having seen the limitations of experimental research in discovery of the causes and hence its limitations in understanding systems, we proceed with a discussion of the concept of causality as well as of the contributions and limitations of causal thinking.
What causes bodies to fall? What causes a forest fire? What causes people to succeed? Why do people commit suicide? All these questions have occupied researchers and scientists. This line of questioning assumes that there is one or a number of causes for every event. Such an assumption is very much consistent with the mechanistic school of thought of Newton that proposed mechanistic laws of physics. The claim that every action has a corresponding opposing action encouraged researchers initially in the natural sciences and then later in the social sciences to identify the driving force (i.e. cause) for every event.
Thus far cause has been defined as a necessary and sufficient cause for an event to take place. In this regard, we need to remember that a literature exists that claims that there is no way to define what is a necessary and sufficient, and that this principle is misleading (Brennan, Fall 2003)[9]. The primary limitation of this principle is revealed when it is applied to attempts to understand how events take place and their causes. Even in such a simple event as spilling a glass, it is difficult to find one necessary and sufficient cause for its falling and shattering when it hits the floor. Is the cause the force that drove the glass off the table; is it gravity; the height of the table; the materials that make up the glass; the floor? Or, is it the absence of additional causes, such as: a force that will restrain the glass and prevent it from falling; a flexible way of holding the glass; a flexible floor? All these are causes that in different combinations determine the outcome. For example, the result of the fall of a thin glass from high distance onto a stiff floor will be different from the fall of a thick glass from a low height onto a dirt floor. Even when we identify a sufficient cause of an event, this is still not its sole cause. For example, if a shot fired in the head it is sufficient to cause death, yet death can still be caused by many other causes without a shot hitting a head. Among the causes that are known in nature and in society, it is difficult to find one cause that is necessary and sufficient of a particular event. Events always take place due to a combination of causes.
David Hume, the Scottish philosopher, became famous due to his theory about the question of causality (in Magee, 1987)[19]. He claimed that even if event “A” always appears before event “B,” even when event “B” always happens after event “A,” we cannot conclude that event “A” is the cause (a necessary a sufficient cause) for event “B” to take place. For example, “evening” takes place prior to “daytime” that always takes place after the evening. In spite of this, we cannot claim that the “evening” causes “daytime,” or the opposite. According to Hume, it is impossible to observe a causal connection in nature. We are only able to observe necessary regularity or proximity between events. Our mind is what makes the connection between regularity and causality. The German philosopher, Immanuel Kant (1969)[17], also claimed that causality is a characteristic of the mind and does not exist in nature.
In my opinion, the causality produced by the mind is too simple way of understanding complex phenomena in nature and in society, for a human mind that is incapable of such perception. Every system is composed of a number of factors that simultaneously interact in multiple directions. Further, every system is part of another system that is composed of other systems whose inter-connection is simultaneously multi-directional. Systemic perception is necessary when we are speaking about systems such as our physical body, a social system such as the family, or ecological systems. For example, the human body is a continuous and developing multi-dimensional, interactive system that links thinking, feeling, behavior, personality, heredity, biology, and environment (Bernard, & Krupat, 1994; Forgas, 2001; Sarafino, 2002)[3,28]. We need to remember that even the most minuscule phenomena such as cell is in and of itself a system in which very complex, multi-directional connections exist simultaneously. Similarly, neither is the atom inorganic, rather it is itself a dynamic energy field within which complex actions by electrons, protons, and neutrons take place in equilibrium within a special interactive system that exists internally within the atom as well as externally with its surroundings.
In juxtaposition to linear-causal thinking, there are theories that claim that the processes in nature and in society have a cyclical-rhythmic nature (Lloyd, & Rossi, 1992; Rossi, 1986)[18,25]. For example, the heart beat, blood circulation, the day and the night, the tides, the cycle of water in nature, the life cycle, historical periods, and the movement of the planets. Indeed, that which appears to be linear-serial is only part of a very complex cyclical system. Maxwell (2004)[20] criticized the regularity approach in research and variance theory (Mohr, 1982)[22] that seek the influence of the variables, and proposes the approach of scientific realism and process theory (Salmon, 1984, 1998; Sayer, 2000)[26,27,29] that deals with events and causal processes that take place between the incidents. This is a systemic approach that deals with processes in the context in which the incidents take place and seeks the causal mechanisms (Maxwell, 2004)[20].
In summary, the linear-serial and reductionist approach in scientific research is an attempt to adapt complex phenomena to the limitations of our minds. Causality is a cognitive structure or characteristic of the human mind that enables us to perceive and to interpret what is observed, but it is not part of the observed. There is no doubt that this approach assisted scientists to achieve a partial understanding of phenomena and to make relatively giant advancements in science, but such are minuscule as we proceed down a long and obscure path. We should not delude ourselves and believe that the causal connections observed actually exist in nature or believe that nature or society is a simple sum of the linear connections discovered to date. With our approaching understanding of systems in nature and in society through causal thinking, we need to be aware as well of the distortion and confusion that causality causes in most of the domains of knowledge (Sayer, 2000)[29]. Part of the confusion is recognizable, as noted, in the accumulation of contradictory outcomes in nearly all research that deals with environmental factors in regard to psychological, medical, and social phenomena.
The Systemic Approach
Science developed in three stages: The first stage started with Greek philosophy and continued through the Middle Ages when scientists sought to understand the characteristics of matter and the movement of bodies in relation to their own characteristics. In the mechanistic thinking that developed in the second stage, beginning with Galileo and Newton in the 17th century, the concept of “power” was recognized as a “cause” of the movement of bodies. Since then equations and scientific laws were developed to explain the causal connections between factors. The “field” theory that developed in the third stage (20th century) perceived events in their totality such that individual parts are identified quantitatively and qualitatively only in terms of their relation with other parts (Frank, 1989; Hawking, 1989; Smith, & Smith, 1996)[14,15,32]. Smith and Smith (1996)[32] claim that while physics developed over these three stages, psychology remains in the second, the mechanistic, stage. In their view, aside from a few exceptional cases such as Kurt Lewin's, Gestalt theory and other ecological theories, psychology has remained at the level of investigation of single causation, explanation by way analogy, biological reduction, or psycho-physio dualism. In their view, “field theory” is the necessary and sufficient condition for understanding psychological events and processes (ibid). Though the influence of mechanistic physics on causal thinking in the social science is well known among the different domains of the natural sciences, what is less noticed is the influence of chemistry and biology that had significant achievements due to adoption of more systemic, rather than mechanistic thought. The science of chemistry is characterized more for its understanding of interactions between the bases and less for causal influence of a base on another base. The science of biology too is characterized by its systemic understanding of the body and less for its search for simple causality. Similarly, psychology could realize great benefits if it too would learn from the systemic approach of chemistry and biology.
The Quantum Field Theory in physics that appeared during the era of Albert Einstein's relativity theory provided an answer to the apparent contradiction between Newtonian laws of gravity that dealt with mass on one hand, and laws of electromagnetism and energy that were discovered later on the other. It did not negate Newtonian laws of mechanics, but rather ground them together with the laws of electromagnetism in one field that are explained by means of uniform laws (Kaku, 1993)[16]. Similarly, systemic approaches in the social sciences do not negate the existence of causal connections in nature and in society, but see all the causal connections between the single factors as dynamic and changing components within the system. The systemic approach has great importance since it is capable of unifying worlds of research and disciplines that appear separate or contradictory in an all-encompassing systemic conception.
Conventional psychological research deals with the mind separate from the body (that is dealt with by medicine) and in distinction from society and culture (the domains of sociology). This is in and of itself a form of reductionism that artificially divides the “field” or the system into separate parts. Within each domain of study, the impact of reductionism is much stronger and research is reduced to smaller and smaller components. For example, in medicine there are specializations and separate fields of research for nearly every system and organ in the body (heart, brain, skin, ear-nose-throat, etc). Psychological theories, as well, focus on partial factors in the psychological experience. In contrast, a systemic approach, one that views psychological theories as components of one system, is capable of resolving the apparent contradictions between the single-cause theories.
Though systemic theories do exist in psychology, they function primarily in the field of treatment and intervention and less in the field of research. For example, family therapy theories attempt to understand the individual in the context of the familial system. Similarly, some intervention approaches in educational psychology view a pupil's problems as part of the school as a system and attempt to intervene in the system. These are relative systemic theories since they place less emphasis on biological factors, on the one hand, and social-ecological factors, on the other hand. It may be the case that the bio-psycho-social approach in psychology and medicine (Sarafino, 2002, Turk, 1996)[28,34] is a more promising systemic approach since it views biological, psychological, and social phenomena as one general system.
Grounding Experimental Research in the Systemic Perspective
Though the limitations of simple causal thinking are more salient in the social sciences than in physics, it was physics that led the confrontation with them by means of field theory, when as noted it ground Newton's laws of mechanics in a more general field. Such a move is all the more necessary in the social sciences in order that systemic approaches can ground the experimental approach of cause and outcome. In this regard Gestalt theories teach us that a connection exists between the part and the whole. This is not a matter of right or not right, as both are correct. That is, we need to recognize that a part is but part of a whole and not apply it to the whole; and similarly we need to be aware that the whole is different from the sum of its parts.
Qualitative research, action research, field research, and case study that seek to understand a system within its natural context are no less scientific than experimental studies that deal with questions regarding the influence of a certain factor on others. In my view, the scientific nature of research is not measured according to its degree of accuracy and objectivity, but rather in the degree of its appropriateness to natural and social systems. In this regard, it is possible that an experimental study is exceptionally accurate in regard to validity and internal reliability, but lacks external validity and does not provide us with a real understanding of the system. In this regard we need to realize that tests of validity and reliability need to be undertaken in relation to the systems in nature and in society and not in relation to a select number of variables or to an artificial experimental structure chosen by the researcher according to certain criteria.
Maxwell (2004)[20] claims that qualitative research is capable of providing more valid answers to questions regarding causality than experimental research. In his view, juxtaposed to experimental research that seeks causality in isolated variables while ignoring the context, qualitative research is capable of providing answers regarding the causal processes as well as the mechanisms of the occurrence of phenomena in the events' context.
Further, it appears that meta-analysis research that studies all the research outcomes in a particular field can serve as the basis for grounding apparent contradictory outcomes of different experiments in a more general systemic view. Meta-analysis research that analyzes tens of specific studies that differ in methodology and in the combination of variables selected for study within a general phenomenon is capable of contributing to the understanding of systems in nature or in society and to provide a satisfactory answer to the problem of “contradictory outcomes.” In addition, in order to overcome the limitations of reductionism, there is also a need to expand research in inter-disciplinary fields. Such research opens up the possibility for mutual enrichment between different fields of knowledge and contributes to illumination of the murky areas that exist between domains. What remains to be recognized is the need to strengthen the relation between scientific research in different fields and philosophy that serves as an overarching framework that is capable of retaining the overview of the causal relations within a more general system.
In summary, I am not claiming that is possible or needed to discontinue examining the influence of causes through experimental research. My claim is that we should not forget that the regularity of the character of nature and society is far greater than the causal connections that our mind is capable of understanding and that examination of simple connections is a reduction of more complex phenomenon that can result in what appear to be contradictory outcomes. We should not forget that the behavior of variables when they are neutralized from other variables is not identical to their behavior when they exist within the system. Accordingly, research of questions regarding causality must focus on the causal mechanism by applying a dynamic-systemic view, cooperation between quantitative and qualitative research, different fields of knowledge, and use of meta-analysis and retention of a general philosophic perspective.
Acknowledgement：
I wish to thank my colleagues - Prof. Michael Moore, Prof. Yeshiyahu Tadmor, Dr. Avi Avital, Dr. Miriam Welicker, Dr. Sivan Raz, and Balig Abdu - for interesting comments that contributed to the ideas and presentation of this article.
REFERENCES
1. Alexander, R. B. (2001). The myth of drug-induced addiction. A paper delivered to the Canadian senate, January, 2001, www.parl.gc.ca/37/1/parlbus/commbus/senate/com-e/ill e-e/presentation-e. Retrieved December 12, 2004.
2. Alexander, R. B.Arresting addictions: Drug education and relapse prevention in corrections. Lanham, MD: American Correctional Association, 2002.
3. Bernard, L. C., & Krupat, E. Health psychology: Biopsychosocial factors in health and illness. New York: Harcourt Brace College Publishers, 1994.
4. Bandura, A.Aggression: A social learning analysis. Englewood Cliffs, NJ: Prentice Hall,1973.
5. Bandura, A., Ross, D., & Ross, S. A.Transition of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology,1961,63(3):575-582.
6. Bandura, A., Ross, D., & Ross, S. A. Imitation of film-mediated aggressive model. Journal of Abnormal and Social Psychology,1963,66:3-11.
7. Barker, R., Dembo, t. & Levin, K. (1941). Frustration and aggression: An experiment with young children. University of Iowa Studies in Child Welfare.
8. Breakwell, G. M., Hammond, S., & Fife-Schaw, C.(Eds). Research methods in psychology. 2nd ed. London: Sage Publications,2000.
9. Brennan, A. (Fall 2003). Necessary and Sufficient Conditions, The Stanford Encyclopedia of Philosophy (Fall 2003 Edition), Edward N. Zalta (ed.), URL =
10. Campbell, D. R., & Stanley, J. C.Experimental and quasi-experimental designs for research. New York: Rand McNally,1966.
11. Dollard, J., Doob, L., Miller, N. E., & Mowrer, O. H., Sears, R. Frustration and aggression. New Haven, CT: Yale University Press,1939.
12. Feshbach, S. & Singer, R. D.Television and aggression. San Francisco: Jossy-Bass,1971.
13. Feuer, M. J., Towne, L., & Shavelson, R. J. Scientific culture and educational research. Educational Researcher,2002,31(8):4-14.
14. Frank, P.Einstein: His life and times. New York: Da Capo Press, Inc. (Originally published in 1953),1989.
15. Hawking, S.A brief history of time. New York: Bantam,2005.
16. Kaku, M.Quantum field theory. New York: Oxford University Press,1993.
17. Kant, I.Critique of pure reason. New York: St. Martin's Press. (Originally published, 1781),1969.
18. Lloyd, D., & Rossi, E.,(Eds.). Ultradian rhythms in life processes: A fundamental inquiry into chronobiology and psychobiology. New York: Springer-Verlag,1992.
19. Magee, B. The great philosophers: An introduction to Western philosophy. New York: Oxford Paperbacks,2000.
20. Maxwell, J. A.Causal explanation, qualitative research, and scientific inquiry in education. Educational Researcher,2004，33(2), 3-11.
21. Meichenbaum, D. A clinical handbook: Practical therapist manual for assessing and treating adults with post-traumatic stress disorder (PTSD). Ontario, Canada: Institute Press，2001.
22. Mohr, L. B.Explaining organizational behavior. San Francisco: Jossey-Bass，1982.
23. National Research Council. Scientific research in education. Washington, DC: National Academy Press，2002.
24. Orner, R. J. & Schnyder, U.Reconstructing early intervention after trauma. Oxford: Oxford University Press，2003.
25. Rossi, E. Altered states of consciousness in everydaylife: The ultradian rhythms. In B. Wolman & M. Ullman (Eds.), Handbook of altered states on consciousness. New York: Van Nostrand，1986，97-132.
26. Salmon, W. C.Scientific explanation and the causal structure of the world. Princeton, NJ: Princeton University Press，1984.
27. Salmon, W. C. Causality and explanation. New York: Oxford University Press，1998.
28. Sarafino, E. P.Health psychology: Biopsychosoial interactions. New York: John Wiley & Sons, Inc，2002.
29. Sayer, A.Realism and social science. London: Sage Publications，2000.
30. Shachter, S., & Singer, J. H.Cognitive social and physiological determinants of emotional state. Psychological Review,1962，69, 379-397. 1,
31. Smith, T. W., & Gallo, L. C.Personality trait as risk factors for physical illness. In A. W. Baum, T. A. Revenson, & J. E. Singer, (Eds). Handbook of health psychology. New York: Lawrence Erlbaum Associates, Inc,2001.
32. Smith, N. W., & Smith L. L.Field theory in science: Its role as a necessary and sufficient condition in psychology. Psychological Reports,1996，46:3-19.
33. Suchday, S., Tucker, D. L., & Krantz, D. S.Diseases，2002.of the circulatory system. In T. J. Boll.Handbook of clinical health psychology. Washington, DC.: American Psychological Association，2002.
34. Turk, D. C.Biopsychosocial perspective on chronic pain. In R. J. Gatchel, and D. C. Turk (Eds.). Psychological Approaches to Pain Management: A Practitioner's Handbook,1996,3-32.
35. Turk, D. C. Physiological and psychological bases of pain. In A. W. Baum, T. A. Revenson, & J. E. Singer, (Eds). Handbook of health psychology. New York: Lawrence Erlbaum Associates, Inc，2001.
36. Welicker, M.Labria'ut mekol halev [Heartfelt Health]. Haifa: Department for Teaching Technology and Science, The Technion & Jerusalem: Curriculum Department, Ministry of Education，2002.
(Editor Jaque)(Marwan Dwairy)