CHAPTERIEXPLANATIONINSCIENTIFICPSYCHOLOGY15explain moreresults is a better theory.If two theories can explain thesame number ofresults, the one with fewer explanatory concepts is preferred.Precision is another important criterion, especially in psychology (where it is oftenlacking).Theories that involvemathematical equations or computerproblemsaregen-erally more precise, and hence better, than those that use loose verbal statements (allother things being equal, of course).Unless a theory is so precise that different investi-gators can agree about its predictions, it is for all intents and purposes useless.Testability goes beyond precision. A theory can be very precise and yet not ableto be tested. For example, when Einstein proposed the equivalence of matter andenergy (E = mc-), nucleartechnology was not ableto test this relationship directlyThe scientist places a very high value on the criterion of testability, because a theorythat cannot be tested can never be disproved At first you might think this would be agood quality since it would be impossible to demonstrate that such a theory was incor-rect.The scientist takes the opposite view.For example,consider EsP (extrasensoryperception). Some believers in ESP claim that the presence of a disbeliever is sufficienttopreventapersongiftedwithEsPfromperforming,becausethedisbelieverputsout"bad vibes"that disrupt EsP.This means that EsP cannot be evaluated, because onlybelievers can be present when it is demonstrated. The scientist takes a dim view of thislogic, and most scientists, especially psychologists, are skeptical about EsP. Belief in atheory increases as it survives tests that could reject it, Since it is logically possible thatsomefuture test may find a flaw,belief in a theory is never absolute.If it is not logicallypossible to test a theory, it cannot be evaluated; hence, it is useless to the scientist. If itis logicallypossiblebut notyettechnicallyfeasible, as wasonce the case withEinstein'stheory, then evaluation of a theory is deferred.Finally, a theory must fit the data it explains. While goodness of fit is not a suf-ficient criterion for accepting a theory (Roberts & Pashler, 2000), there is little point inpursuing a theory that fails to fit the data (Rodgers & Rowe, 2002).InterveningVariablesTheories often use constructs that summarize the effects of several variables. Variables arediscussed atgreaterlength in Chapter3.Fornow,webrieflydescribetwo differentkindsof variables.Independent variables are those manipulated by the experimenter.For ex-ample, not allowing rats to have any water for several hours would create an independentvariable called hours of deprivation.Dependent variables are those observed by the ex-perimenter. For example, one could observe how much water a rat drinks.Science tries to explain the world by relating independent and dependent vari-ables.Intervening variables are abstract concepts that link independent variables todependent variables. Gravity is a familiar construct that accomplishes this goal. It canrelate an independent variable, the feet of height from which an object is dropped,to a dependent variable, the speed of the object when it hits the ground. Gravity alsosummarizestheeffectsofheightonspeedforallmannerof objects.Gravityexplainsfallingapples as well as falling baseballs.Scienceprogresseswhen a single construct,such as gravity, explains outcomes in many different environments.Miller (1959)has explained how a single intervening variable, thirst, organizesexperimental results efficiently. Figure 1.3 shows a direct and an indirect way to relatean independent variable,hours of deprivation,to a dependentvariable, rate of bar
CHAPTER 1 EXPLANATION IN SCIENTIFIC PSYCHOLOGY 15 explain more results is a better theory. If two theories can explain the same number of results, the one with fewer explanatory concepts is preferred. Precision is another important criterion, especially in psychology (where it is often lacking). Theories that involve mathematical equations or computer problems are generally more precise, and hence better, than those that use loose verbal statements (all other things being equal, of course). Unless a theory is so precise that different investigators can agree about its predictions, it is for all intents and purposes useless. Testability goes beyond precision. A theory can be very precise and yet not able to be tested. For example, when Einstein proposed the equivalence of matter and energy (E mc 2 ), nuclear technology was not able to test this relationship directly. The scientist places a very high value on the criterion of testability, because a theory that cannot be tested can never be disproved. At fi rst you might think this would be a good quality since it would be impossible to demonstrate that such a theory was incorrect. The scientist takes the opposite view. For example, consider ESP (extrasensory perception). Some believers in ESP claim that the presence of a disbeliever is suffi cient to prevent a person gifted with ESP from performing, because the disbeliever puts out “bad vibes” that disrupt ESP. This means that ESP cannot be evaluated, because only believers can be present when it is demonstrated. The scientist takes a dim view of this logic, and most scientists, especially psychologists, are skeptical about ESP. Belief in a theory increases as it survives tests that could reject it. Since it is logically possible that some future test may fi nd a fl aw, belief in a theory is never absolute. If it is not logically possible to test a theory, it cannot be evaluated; hence, it is useless to the scientist. If it is logically possible but not yet technically feasible, as was once the case with Einstein’s theory, then evaluation of a theory is deferred. Finally, a theory must fi t the data it explains. While goodness of fi t is not a suf- fi cient criterion for accepting a theory (Roberts & Pashler, 2000), there is little point in pursuing a theory that fails to fi t the data (Rodgers & Rowe, 2002). Intervening Variables Theories often use constructs that summarize the effects of several variables. Variables are discussed at greater length in Chapter 3. For now, we briefl y describe two different kinds of variables. Independent variables are those manipulated by the experimenter. For example, not allowing rats to have any water for several hours would create an independent variable called hours of deprivation. Dependent variables are those observed by the experimenter. For example, one could observe how much water a rat drinks. Science tries to explain the world by relating independent and dependent variables. Intervening variables are abstract concepts that link independent variables to dependent variables. Gravity is a familiar construct that accomplishes this goal. It can relate an independent variable, the feet of height from which an object is dropped, to a dependent variable, the speed of the object when it hits the ground. Gravity also summarizes the effects of height on speed for all manner of objects. Gravity explains falling apples as well as falling baseballs. Science progresses when a single construct, such as gravity, explains outcomes in many different environments. Miller (1959) has explained how a single intervening variable, thirst, organizes experimental results effi ciently. Figure 1.3 shows a direct and an indirect way to relate an independent variable, hours of deprivation, to a dependent variable, rate of bar 59533_02_ch01_p001-023.indd 15 9533_02_ch01_p001-023.indd 15 3/4/08 11:47:15 PM /4/08 11:47:15 PM
16PARTIFUNDAMENTALSOFRESEARCHDependent VariableIndependentVariableInterveningVariableHoursofdeprivationRateofbarpressingHours ofdeprivationThirstRateofbarpressingFIGURE1.3One Setof Variables.pressing.The dependent variable is obtained by placing a rat into a small chamberwhere it can press a bar to obtain drinking water.The experimenter observes the rate(how many presses per minute) at which the rat presses the bar to get water.The directrelationship uses only one arrow to link hours of deprivation to rate of bar pressing.Afterdoingtheexperiment,wecouldbuildamathematicalformulathatdirectlyrelateshours of deprivation to rate of bar pressing.The indirect method in Figure 1.3 uses twoarrows.The first arrow relates hours of deprivation to thirst, an intervening variable.The second arrow relates the intervening variable, thirst, to the rate of bar pressing.Since the indirect method is more complicated, requiring an extra arrow, you mightexpect the scientist to prefer the direct method of explanation.Indeed, if the onlyscientificgoal were torelatehours of deprivation to rate of bar pressing,you wouldbecorrectbecausesciencepreferssimpleexplanationstocomplexexplanations.How-ever, as we shall explain, thescientific goal is moregeneral.Figure 1.4 relates two independent variables, hours of deprivation and feedingdry food, to two dependent variables,rate of bar pressing and volume of water drunk.Again, both direct and indirect explanations are shown. In Figure 1.4, direct and indirectexplanationsareequallycomplex.Eachrequiresfourdistinctarrows.IndependentVariablesIntervening VariableDependent VariablesHoursofdeprivationRateofbarpressingVolumeof waterdrunkFeedingdryfoodHoursofdeprivationRateof barpressingThirstFeeding dryfoodVolumeofwaterdrunkFIGURE1.4Two Setsof Variables
16 PART 1 FUNDAMENTALS OF RESEARCH pressing. The dependent variable is obtained by placing a rat into a small chamber where it can press a bar to obtain drinking water. The experimenter observes the rate (how many presses per minute) at which the rat presses the bar to get water. The direct relationship uses only one arrow to link hours of deprivation to rate of bar pressing. After doing the experiment, we could build a mathematical formula that directly relates hours of deprivation to rate of bar pressing. The indirect method in Figure 1.3 uses two arrows. The fi rst arrow relates hours of deprivation to thirst, an intervening variable. The second arrow relates the intervening variable, thirst, to the rate of bar pressing. Since the indirect method is more complicated, requiring an extra arrow, you might expect the scientist to prefer the direct method of explanation. Indeed, if the only scientifi c goal were to relate hours of deprivation to rate of bar pressing, you would be correct because science prefers simple explanations to complex explanations. However, as we shall explain, the scientifi c goal is more general. Figure 1.4 relates two independent variables, hours of deprivation and feeding dry food, to two dependent variables, rate of bar pressing and volume of water drunk. Again, both direct and indirect explanations are shown. In Figure 1.4, direct and in direct explanations are equally complex. Each requires four distinct arrows. Hours of deprivation Hours of deprivation Thirst Rate of bar pressing Rate of bar pressing Independent Variable Intervening Variable Dependent Variable ▼ FIGURE 1.3 One Set of Variables. Hours of deprivation Feeding dry food Volume of water drunk Rate of bar pressing Hours of deprivation Feeding dry food Volume of water drunk Rate of bar pressing Thirst Independent Variables Dependent Variables Intervening Variable ▼ FIGURE 1.4 Two Sets of Variables. 59533_02_ch01_p001-023.indd 16 9533_02_ch01_p001-023.indd 16 3/4/08 11:47:15 PM /4/08 11:47:15 PM
CHAPTERIEXPLANATIONINSCIENTIFICPSYCHOLOGY17IndependentVariablesInterveningVariableDependent VariablesHoursofdeprivationRateofbarpressingFeedingdryfoodVolumeofwaterdrunkSaline injectionQuininetostopdrinkingHours of deprivationRateofbarpressing+ThirstFeedingdryfoodVolumeofwaterdrunkSaline injectionQuinineto stopdrinkingFIGURE1.5Three SetsofVariablesFigure 1.5 relates three independent variables, hours of deprivation, feeding dryfood, and saline injection (giving a rat saltwater through a tube inserted in its stomach),to three dependent variables,rate of bar pressing,volume of water drunk, and amountof quinine required to stop the rat from drinking. Again, both direct and indirect ex-planations are shown. Now, it is obvious that the indirect method is less complicatedIt requires six distinct arrows, whereas the direct method requires nine arrows. So asscience tries to relate more independent and dependent variables, intervening variablesbecome more efficient.There is yet another advantage of intervening variables. Thirst, regardless of how itis produced, should have the same effect on all dependent variables. This can be testedin experiments.If it is not true, we can reject the idea of a single intervening variable.Later chapters discuss this issue under the topic of converging operations.FoxesandHedgehogsRoamingthroughPsychologicalTheoryResearch in experimental psychology tends to be organized by sub-fields; indeed, Part Twoof this textbook presents chapters for such sub-fields as perception, memory,and socialinfluence. This approach is the way of the fox who knows many paths (Figure 1.6). Suchwas not always the case in the history of psychology. There were psychologists (e.g.,James
CHAPTER 1 EXPLANATION IN SCIENTIFIC PSYCHOLOGY 17 Figure 1.5 relates three independent variables, hours of deprivation, feeding dry food, and saline injection (giving a rat saltwater through a tube inserted in its stomach), to three dependent variables, rate of bar pressing, volume of water drunk, and amount of quinine required to stop the rat from drinking. Again, both direct and indirect explanations are shown. Now, it is obvious that the indirect method is less complicated. It requires six distinct arrows, whereas the direct method requires nine arrows. So as science tries to relate more independent and dependent variables, intervening variables become more effi cient. There is yet another advantage of intervening variables. Thirst, regardless of how it is produced, should have the same effect on all dependent variables. This can be tested in experiments. If it is not true, we can reject the idea of a single intervening variable. Later chapters discuss this issue under the topic of converging operations. Foxes and Hedgehogs Roaming through Psychological Theory Research in experimental psychology tends to be organized by sub-fi elds; indeed, Part Two of this textbook presents chapters for such sub-fi elds as perception, memory, and social infl uence. This approach is the way of the fox who knows many paths (Figure 1.6). Such was not always the case in the history of psychology. There were psychologists (e.g., James, ▼ FIGURE 1.5 Three Sets of Variables. Hours of deprivation Feeding dry food Volume of water drunk Rate of bar pressing Hours of deprivation Feeding dry food Volume of water drunk Rate of bar pressing Thirst Saline injection Quinine to stop drinking Saline injection Quinine to stop drinking Independent Variables Dependent Variables Intervening Variable 59533_02_ch01_p001-023.indd 17 9533_02_ch01_p001-023.indd 17 3/4/08 11:47:15 PM /4/08 11:47:15 PM
18PARTIFUNDAMENTALSOFRESEARCH"Thefoxknowsmany things,butthehedgehogknowsonebigthing.-ArchilocusFIGURE1.6Segmented (fox)and unified (hedgehog)approaches to psychological theory1890)who tried fora unified explanation of psychological phenomena.This is the way ofthehedgehogwhoknowsonebigthing (seeFigure1.6)Both approaches face significant challenges.The dominant approach builds strongbarriers between fields.Professors are hired within a field and tend to have officeslocated by field of specialization, the better to fend off territorial thrusts from othersub-fields. Graduate students get trained by field, with appropriate course require-ments, and this perpetuates the division. Even panels that evaluate grant proposals areorganized by specialties. The new Ph.D.who takes a job in industry is ill prepared forthe cooperation acrossfields necessaryto solveany importantpractical problem.Recently,sympathyhas been growing for a moreunified (hedgehog)approachtopsychology (Sternberg, Grigorenko, & Kalmar, 2001). These theorists try to knit com-peting theories together, stressing that the explanation role of theory is more crucialthan the predictive role. Theory knitting is a worthy goal if it can be accomplished.However,the present divided system arose bedauseearlierintegratedtheorieswereunable to span all the sub-fields of psychology. Will the new hedgehogs knit betterthan the old hedgehogs?THESCIENCEOEPSYCHOLOGYSome students find it difficult to think of psychology as a science in the same sense thatphysics and chemistry are sciences.They believe that there are aspects of human experi-ence, such as the arts, literature, and religion, that defy scientific analysis. How can thebeauty of a Klee lithograph, a Beethoven sonata, ora Cartier-Bresson photograph bereduced to cold scientific equations? How can the tender feelings of a first romance, thethrill of driving a sports car at 100 miles per hour, or the agony of a defeated footballteam be captured in the objective, disinterested fashion required by science?Some psychologists,known as humanists, would answer these questions in thenegative.Thesehumanists,mostoftenclinical and counselingpsychologists,claimthatitis impossible to evaluate and test objectively much of human feelings and experience bytraditional scientific methods.Even tough, "brass-instrument"experimental psychologists
18 PART 1 FUNDAMENTALS OF RESEARCH 1890) who tried for a unifi ed explanation of psychological phenomena. This is the way of the hedgehog who knows one big thing (see Figure 1.6). Both approaches face signifi cant challenges. The dominant approach builds strong barriers between fi elds. Professors are hired within a fi eld and tend to have offi ces located by fi eld of specialization, the better to fend off territorial thrusts from other sub-fi elds. Graduate students get trained by fi eld, with appropriate course requirements, and this perpetuates the division. Even panels that evaluate grant proposals are organized by specialties. The new Ph.D. who takes a job in industry is ill prepared for the cooperation across fi elds necessary to solve any important practical problem. Recently, sympathy has been growing for a more unifi ed (hedgehog) approach to psychology (Sternberg, Grigorenko, & Kalmar, 2001). These theorists try to knit competing theories together, stressing that the explanation role of theory is more crucial than the predictive role. Theory knitting is a worthy goal if it can be accomplished. However, the present divided system arose because earlier integrated theories were unable to span all the sub-fi elds of psychology. Will the new hedgehogs knit better than the old hedgehogs? ▼ THE SCIENCE OF PSYCHOLOGY Some students fi nd it diffi cult to think of psychology as a science in the same sense that physics and chemistry are sciences. They believe that there are aspects of human experience, such as the arts, literature, and religion, that defy scientifi c analysis. How can the beauty of a Klee lithograph, a Beethoven sonata, or a Cartier-Bresson photograph be reduced to cold scientifi c equations? How can the tender feelings of a fi rst romance, the thrill of driving a sports car at 100 miles per hour, or the agony of a defeated football team be captured in the objective, disinterested fashion required by science? Some psychologists, known as humanists, would answer these questions in the negative. These humanists, most often clinical and counseling psychologists, claim that it is impossible to evaluate and test objectively much of human feelings and experience by traditional scientifi c methods. Even tough, “brass-instrument” experimental psychologists "The fox knows many things, but the hedgehog knows one big thing." –Archilocus ▼ FIGURE 1.6 Segmented (fox) and unifi ed (hedgehog) approaches to psychological theory. 59533_02_ch01_p001-023.indd 18 9533_02_ch01_p001-023.indd 18 3/4/08 11:47:15 PM /4/08 11:47:15 PM
CHAPTERIEXPLANATIONINSCIENTIFICPSYCHOLOGY19concur that the domain of science is limited. We cannot establish or refute the existenceof God by scientific means any more than we could test gravity by theological methodsScience operates where its tools are appropriate (see Chapter 14).This does not implythatknowledge cannotbegained wherever sciencefears to treadthat is, by nonscien-tific means.Many important fields of human endeavor have yet to benefit from extensivescientificanalysis-ethics,morals,andlaw,tonamebutafew.However, most scientists would hold out the hope that scientific analysis eventu-ally might be usefully applied to many such areas. Much of contemporary psychol-ogy was regarded as the sole property of philosophy at one time. As psychologicaltechniques improved, theseaspects of human expertise and behaviormoved into therealm of science. And now most psychologists believe that virtually all facets of humanexperience arefair gamefor the science of psychology.Deriding scientificprogress inpsychology,as did one U.S. senator who criticized the National Science Foundationfor supporting research on romantic love,will nothalt efforts to expand psychologi-cal knowledge. Although concern for the proper and ethical use of such knowledge isvalid and important,ignoranceis nosolutionPsychologyandtheReal WorldScientists, in general, and psychologists,in particular, have many reasons for pursuingtheir profession. Although we think it rather easy to prove that psychological researchdoes serve humanity,we would liketo stress that we do not find this the only,ornecessarilythemajor,justificationfora career asa researchpsychologist,Many scien-tists investigate certain problems simply because they find them interesting. We havecomplete sympathy with a colleague who might state that he or she studies gerbils justbecause gerbils provoke his or her curiosity. It is true that certain studies are performedon animals becausethey are unethical orimpractical to performon humans-for ex-ample, studies of long-term crowding, punishment, drugs, and so on-but it is equallytrue that the behavior of animals is interesting in its own right.Scientific research is often divided into two categories: basic and applied.Appliedresearch aims at solving a specific problem-such as how to cure bedwettingwhereasbasic research has no immediate practical goal. Basic research establishes a reservoir ofdata,theoretical explanations, and concepts that can betapped by the applied researcher.Without this source, applied research would soon dry up and sputter to a halt, unless ap-plied researchersbecame of necessitybasic researchers.It takes quite a while for a con-cept developed by basic research to find some useful application in society. Adams (1972)traced five socially important products to discover the impact, if any,of basic research.Although basic researchaccounted for 70percent of the significant events, the researchoccurred twenty to thirty years before the ultimate use of the product. This long time lagobscures the crucial role of basic research so that many persons incorrectly believe thatbasic research is not very useful to society. It is quite difficult to tell what basic researchbeing donetoday will have an impact thirtyyears from now.But this inabilitytopredicthardlymeansthatweshouldstopdoingbasicresearch.Although most experimental psychologists are content with a scientist-practitionermodel where applied research is based on the fruits of basic research, more recently therehas been a push for a two-track system (Fishman & Neigher, 1982; Howell, 1994) wherebasic and applied research diverge.From a historical perspective (Bevan, 1980) these two
CHAPTER 1 EXPLANATION IN SCIENTIFIC PSYCHOLOGY 19 concur that the domain of science is limited. We cannot establish or refute the existence of God by scientifi c means any more than we could test gravity by theological methods. Science operates where its tools are appropriate (see Chapter 14). This does not imply that knowledge cannot be gained wherever science fears to tread—that is, by nonscientifi c means. Many important fi elds of human endeavor have yet to benefi t from extensive scientifi c analysis—ethics, morals, and law, to name but a few. However, most scientists would hold out the hope that scientifi c analysis eventually might be usefully applied to many such areas. Much of contemporary psychology was regarded as the sole property of philosophy at one time. As psychological techniques improved, these aspects of human expertise and behavior moved into the realm of science. And now most psychologists believe that virtually all facets of human experience are fair game for the science of psychology. Deriding scientifi c progress in psychology, as did one U.S. senator who criticized the National Science Foundation for supporting research on romantic love, will not halt efforts to expand psychological knowledge. Although concern for the proper and ethical use of such knowledge is valid and important, ignorance is no solution. Psychology and the Real World Scientists, in general, and psychologists, in particular, have many reasons for pursuing their profession. Although we think it rather easy to prove that psychological research does serve humanity, we would like to stress that we do not fi nd this the only, or necessarily the major, justifi cation for a career as a research psychologist. Many scientists investigate certain problems simply because they fi nd them interesting. We have complete sympathy with a colleague who might state that he or she studies gerbils just because gerbils provoke his or her curiosity. It is true that certain studies are performed on animals because they are unethical or impractical to perform on humans—for example, studies of long-term crowding, punishment, drugs, and so on—but it is equally true that the behavior of animals is interesting in its own right. Scientifi c research is often divided into two categories: basic and applied. Applied research aims at solving a specifi c problem—such as how to cure bedwetting—whereas basic research has no immediate practical goal. Basic research establishes a reservoir of data, theoretical explanations, and concepts that can be tapped by the applied researcher. Without this source, applied research would soon dry up and sputter to a halt, unless applied researchers became of necessity basic researchers. It takes quite a while for a concept developed by basic research to fi nd some useful application in society. Adams (1972) traced fi ve socially important products to discover the impact, if any, of basic research. Although basic research accounted for 70 percent of the signifi cant events, the research occurred twenty to thirty years before the ultimate use of the product. This long time lag obscures the crucial role of basic research so that many persons incorrectly believe that basic research is not very useful to society. It is quite diffi cult to tell what basic research being done today will have an impact thirty years from now. But this inability to predict hardly means that we should stop doing basic research. Although most experimental psychologists are content with a scientist–practitioner model where applied research is based on the fruits of basic research, more recently there has been a push for a two-track system (Fishman & Neigher, 1982; Howell, 1994) where basic and applied research diverge. From a historical perspective (Bevan, 1980) these two 59533_02_ch01_p001-023.indd 19 9533_02_ch01_p001-023.indd 19 3/4/08 11:47:16 PM /4/08 11:47:16 PM