FADS, FASHIONS, AND FOLDEROL IN PSYCHOLOGY1

MARVIN D. DUNNETTE

University of Minnesota

THIS seemed to be a great idea when the 1965Division 14 program was being arranged.When I chose the title, it seemed very clever,and I looked forward to venting my spleen a bitand to saying sage things about what is wrongwith psychology and how its ills might be cured.

It was not long, however, before I felt misgiv-ings about the whole enterprise and doubtedwhether I could meet the challenge to put up orto shut up that my own foolhardiness had castupon me. At first I thought I knew the things inpsychology that bothered me, but would there beconsensus about this? Suddenly, I had the sober-ing thought that I might be the only one out ofstep and that, really, all was well with psychology.

Soseizing upon one of my own pet types ofmethodological folderolI decided to run a sur-vey! I contacted older or wiser heads,2 to inquirewhat, if anything, was currently bothering them.Their responses might conservatively be describedas a vast outpouring. The volume and intensity oftheir replies caused me to give up my speculationsabout joining the French Foreign Legion and showup here today after all.

With their suggestions, however, 1 was facedsuddenly with a plethora of fads, fashions, andfolderol and the need to make some systematicsense of them. Let me give you the flavor of mysurvey results by simply mentioning some of thethings listed by me and by my respondents.

Fadsthose practices and concepts character-ized by capaciousness and intense, but short-livedinterestincluded such things as brainstorming, Qtechnique, level of aspiration, forced choice, criti-cal incidents, semantic differential, role playing,

1 Invited Address presented to Division 14 at American

Psychological Association, Chicago, September 1965.21 should like to thank the following persons for their

readiness to come to my aid, but please understand thatI do so only because of a strong sense of gratitude andwith no thought of having them join me out at the end ofthe limb. They are: David P. Campbell, John P. Camp-bell, Alphonse Chapanis, Edwin E. Ghiselli, Mason Haire,James Jenkins, Quinn McNemar, Paul Meehl, William A.Owens, Jr. , Bernard Rimland, Auke TcllcRcn, Rains Wal-lace, and Karl Wcick.

need theory, grids of various types, adjectivechecklists, two-factor theory, Theory X and TheoryY, social desirability, response sets and responsestyles, need hierarchies, and so on and so on.

Fashionsthose manners or modes of actiontaking on the character of habits and enforced bysocial or scientific, norms defining what constitutesthe "thing to do"included theorizing and theorybuilding, criterion fixation, model building, nullhypothesis testing, sensitivity training, being pro-ductive at work, developing authentic relationships,devising "cute" experiments, simulation, using "ele-gant" statistics, and so on.

Finally, folderolthose practices characterizedby excessive ornamentation, nonsensical and un-necessary actions, trifles and essentially useless andwasteful fiddle-faddleincluded tendencies to befixated on theories, methods, and points of view,conducting "little" studies with great precision,attaching dramatic but unnecessary trappings toexperiments, asking unimportant or irrelevantquestions, grantsmanship, coining new names forold concepts, fixation on methods and. apparatus,seeking to "prove" rather than "test" theories, andmyriad other methodological ceremonies conductedin the name of rigorous research.

But, even armed with my list, about all I couldsay is that there are many things going on in psy-chology that reasonably responsible people werewilling to label faddish folderol. It accomplishedthe aim of identifying some of the less honorablethings we all are doing, but it seemed rather sterileas a source of prescriptive implications. What wasneeded was a better taxonomy for listing psychol-ogy's ills than the rather artificial trichotomy es-tablished by my title.

One approach might be through some form ofcluster analysisbut my data did not proveamenable to any of the widely used methods suchas Pattern Analysis, Elementary Linkage Analysis,Hierarchical Linkage Analysis, Hierarchical Syn-drome Analysis, Typal Analysis, Rank Order TypalAnalysis, Comprehensive Hierarchical Analysis, oreven Multiple Hierarchical Classification.

343

344 AMERICAN PSYCHOLOGIST

Just when 1 was facing this impasse, 1 receivedthe best-selling book by Eric Berne (1964) titledGames People Play. I was stimulated by thismagnificent book to give thought to the gamespsychologists play. Somewhat to my surprise, Iexperienced no difficulty slipping into the robes ofa medical clinicianintent on describing fullyand completely the behavioral symptomatology ofpsychology's distress in terms of the games we allplaymany of which may reflect underlying pa-thologies leading us down the primrose path tononscience.

The games can be discussed under six broadheadingsThe Pets We Keep; The Fun We Have;The Names We Love; The Delusions We Suffer;The Secrets We Keep; and The Questions We Ask.

THE PETS WE KEEPSubtitled "What Was Good Enough for Daddy

Is Good Enough for Me," this game is character-ized by an early and premature commitment tosome Great Theory or Great Method. One majoreffect is to distort research problems so that theyfit the theory or the method. The theory, method,or both can be viewed as pets inherited by fledg-ling psychologists and kept and nurtured by them,in loving kindness, protecting them from all possi-ble harm due to the slings and arrows and attacksfrom other psychologists who, in turn, are keepingtheir own menageries.

At a general level, the premature commitment toa theory is usually accompanied by the set to proverather than to modify the theory. The problemand its potentially bad outcome was outlined yearsago by T. C. Chamberlin (1965), a well knowngeologist. He stated:The moment one has offered an original explanation fora phenomenon which seems satisfactory, that momentaffection for his intellectual child springs into existence;. . . there is an unconscious selection and magnifying ofthe phenomena that fall into harmony with the theory andsupport it, and an unconscious neglect of those that failof coincidence. . . . When these biasing tendencies set in,the mind rapidly degenerates into the partiality of pa-ternalism. . . . From an unduly favored child, it [thetheory] readily becomes master, and leads its authorwhithersoever it will [p. 7551.

It is not difficult in psychology to recognize thesequence of events described by Chamberlin. Apessimist might, in fact, find it difficult to identifyany psychological theories which do not currentlyenjoy this form of affectionate nurturing. On the

other hand, a more optimistic view might accord totheories the important function of ordering andsystematizing the conduct of research studies.What is to be avoided, of course, is the kind ofpaternal affection and closed mind described byChamberlin.

The problem in psychology is made more severe,however, by the inexplicitness (Fiegel. 1962) and,as Ritchie3 has called it, the "incurable vagueness"with which most theories are statedbut then, itshould be clear that vagueness in theory construc-tion may simply be part of the game, insuringhigher likelihood of a pet theory's long life.

Methodologically, our favored pets include factoranalysis, complex analysis of variance designs, theconcept of statistical significance, and multiple-regression analyses. It is common for psycholo-gists to apply so-called sophisticated methods ofanalysis to data hardly warranting such carefulattention. I shall not try to enumerate the natureof the painstaking activities included in the gameof statistical pet keeping. I refer those of you whoare interested to excellent papers by McNemar(1951) and by Guilford (1960). The net effect,however, is that attention to relevant and impor-tant scientific questions is diminished in favor ofworking through the subtle nuances of methodo-logical manipulation. As my colleague, DavidCampbell remarked,4

We seem to believe that TRUTH will be discovered some-how through using more and more esoteric techniques ofdata manipulation rather than by looking for it in thereal world.

Or, as Platt (1964) has said:Beware of the man of one method or one instrument,either experimental or theoretical. He tends to becomemethod-oriented rather than problem oriented; the methodoriented man is shackled [p. 3SH.

THE NAMES WE LOVE

An alternate title for this game is "What's NewUnder the Sun?" Unfortunately, an undue amountof energy is devoted to the Great Word Gamethe coining of new words and labels either to fitold concepts or to cast new facts outside the kenof a theory in need of protection.

Just one from among many possible examples isthe great emphasis in recent years on Social Desir-

3 B. F. Ritchie, unpublished work.

4 Personal communication, 196S.

FADS, FASHIONS, AND FOLDEEOL 345

abilitya new label for a phenomenon in test-taking behavior dealt with extensively by Meehland Hathaway (1946), Jurgensen, and others manyyears previously, but which did not create muchinterest because they failed at the time to coin alabel sufficiently attractive to "grab" other psy-chologists.

As Maier (1960) has so aptly pointed out, onemajor effect of the Name Game is to sustaintheories even if the facts seem to refute them. Iffacts appear that cannot be ignored, relabelingthem or renaming them gives them their own spe-cial compartment so that they cease to infringeupon the privacy of the theory.

Perhaps the most serious effect of this game isthe tendency to apply new names in psychologicalresearch widely and uncritically before sufficientwork has been done to specify the degree of gen-erality or specificity of the "trait" being dealtwith. Examples of this are numerousanxiety,test-taking anxiety, rigidity, social desirability, cre-ativity, acquiescence, social intelligence, and so onad infinitum.

THE FUN WE HAVEA suitable title for this game would bequite

simply"Tennis Anyone?" But the game hasmany variants, including My Model Is Nicer thanYour Model!, Computers I Have Slept With!, orthe best game of allA Difference Doesn't Needto Make a Difference if It's a Real Difference.

As should be clear, the underlying theme of thegameTennis Anyone?is the compulsion to for-get the problemin essence to forget what we arereally doingbecause of the fun we may be en-joying with our apparatus, our computers, ourmodels or the simple act of testing statistical nullhypotheses. Often, in our zest for this particulargame, we forget not only the problem, but we mayeven literally forget to look at the data!

The most serious yet most common symptom ofthis game is the "glow" that so many of us getfrom saying that a result is "statistically signifi-cant." The song and dance of null hypothesistesting goes on and onapparently endlessly. Inmy opinion, this one practice is as much responsi-ble as anything for what Sommer (1959) hascalled the "little studies" and the "little papers" ofpsychology.

As so many others have pointed out (Binder,1963; Grant, 1962; Hays, 1964; Nunnally, 1960;

Rozeboom, 1960), the major difficulty with psy-chology's use of the statistical null hypothesis isthat the structure of scientific conclusions derivedthereby is based on a foundation of triviality.When even moderately small numbers of subjectsare used nearly all comparisons between meanswill yield so-called "significant" differences. Ibelieve most psychologists will agree, in their moresober and less fun-loving moments, that smalldifferences and inconsequential correlations do notprovide a sufficient yield either for accurately pre-dicting other persons' behavior or for understand-ing theoretically the functional relations betweenbehavior and other variables. Yet, most of us stillremain content to build our theoretical castles onthe quicksand of merely rejecting the null hypoth-esis.

It may seem that my criticism of this particulargame is unduly severe. Perhaps the differencesreported in our journals are not really all thatsmall. In order to examine this question, I askedone of my research assistants, Milton Hakel, tosample recent issues of four APA JournalstheJournal of Applied Psychology, Journal of Abnor-mal and Social Psychology, Journal of Personalityand Social Psychology, and Journal of Experi-mental Psychology. He selected randomly fromamong studies employing either t tests or complexanalysis of variance designs, and converted the tor F values to correlation ratios (eta) in order toestimate the strength of association between inde-pendent and dependent variables.

The distribution of the 112 correlation ratiosranged from .05 to .92 with a median value of .42.Five percent of the studies showed values below.20; over one-sixth were below .25; and nearlyone-third failed to reach .30. The only encourage-ment I derive from these data stems from myidentification with industrial psychology. At a timewhen many in industrial psychology are worriedbecause predictive validities rarely exceed .50, it isat least reassuringthough still disconcertingtonote that our brethren in social and experimentalpsychology are doing little better.

It is particularly informative to note the con-clusions made by the authors of the articles sam-pled by Hakel. Authors of the study yielding theeta of .05 concluded "that rating-scale format is adeterminer of the judgment of raters in this sam-ple [Madden & Bourden, 1964]." In an investi-gation yielding an eta of .14, the authors con-

346 AMEIUCAN PSYCHOLOGIST

eluded "that highly creative subjects give thegreatest number of associations and maintain arelatively higher speed of association throughout a2 minute period [Mednick, Mednick, & Jung,1964]."

Surprisingly, these rather definite conclusionsdiffer little in tone from those based on studiesyielding much stronger relationships. For exam-ple, a study yielding an eta of .77 is summarizedwith "Highly anxious subjects tended to give setsof word associates higher in intersubject variabil-ity than nonanxious subjects |Brody, 1964]." Inlike manner, the conclusion stated for a studyyielding an eta of .63 was simply "It was foundthat reinforcement affected subjects' verbalizations

Ganzer & Sarason, 1964]."It seems abundantly clear that our little survey

provides convincing and frightening evidence thatplaying the game of null hypothesis testing hasled a sizeable number of psychologists to lose sightof the importance of the strength of relationshipsunderlying their conclusions. I could not agreemore fully with Nunnally (1960), who has said:it would be a pity to sec it (psychology) settle for meagerefforts . . . encouraged by the use of the hypothesis test-ing models. . . . We should not feel proud when we seethe psychologist smile and say "the correlation is signifi-cant beyond the .01 level." Perhaps that is the most thathe can say, but he has no reason to smile I p. 6501.

THE DELUSIONS WE SUFEERThis is probably the most dangerous game of

all. At the core, it consists of maintaining delu-sional systems to support our claims that thethings we are doing really constitute good science.The game develops out of a pattern of self-deceitwhich becomes more ingrained and less tractablewith each new delusion. Thus, an appropriate sub-title is "This Above All, to Thine Ownself BeFalse!"

The forms of these delusions are so numerousand so widespread in psychology that time per-mits only brief mention of a few.

One common variant of the game can be called,"Boy, Did \ Ever Make Them Sit Up and TakeNotice!" The argument is often made and seem-ingly almost always accepted that if a new theoryor method stimulates others to do research, itmust be good. Although 1 greatly dislike analogicarguments, I am compelled to suggest that suchreasoning is very similar to stating that accidental

fire must be good simply because it keeps somany firemen busy. Unfortunately, an inestimableamount of psychological research energy has beendissipated in fighting brush fires spawned byfaddish theorieswhich careful research mightbetter have refuted at their inception.

It is probably far too much to hope that we haveseen the last of the studies "stimulated" by Shel-don's notions about physique and temperament, orby the overly simplified but widely popular two-factor theory of job motivation (Herzberg, Maus-ner, & Snyderman, 1959).

A second common delusion seems to arise out ofthe early recognition that gathering data from realpeople emitting real behaviors in the day-to-dayworld proves often to be difficult, unwieldy, andjust plain unrewarding. Thus many retreat intothe relative security of experimental or psycho-metric laboratories where new laboratory or testbehaviors may be concocted to be observed, meas-ured, and subjected to an endless array of internalanalyses. These usually lead to elaborate theoriesor behavioral taxonomies, entirely consistent withinthemselves but lacking the acid test of contactwith reality. Last year, McNemar (1964) sum-marized once more for us the evidence showing thepathetic record of factor analytically derived testsfor predicting day-to-day behavior. A former pro-fessor at Minnesota used to saywhen describinga lost soul"He disappeared into the Jungle ofFactor Analysisnever to be heard from again."Psychologists who choose to partake of the advan-tages of the more rigorous controls possible in thepsychometric or experimental laboratories mustalso accept responsibility for assuring the day-to-day behavioral relevance of the behavioral observa-tions they undertake.

A third unfortunate delusion rationalizes certainpractices on the grounds that they are intrinsicallygood for humanity and that they need not, there-fore, meet the usual standards demanded by scien-tific verification. In this regard, Astin (1961) hasdone an effective job of analyzing the functionalautonomy of psychotherapy and offers a numberof reasons why it continues to survive in spite ofa lack of evidence about its effectiveness. In in-dustrial psychology, a most widespread currentfashion is the extensive use by firms of group-process or sensitivity-training programs; the effec-tiveness of such programs is still proclaimed solelyon the basis of testimonials., and a primary ra-

FADS, FASHIONS, AND FOLDEROT, 347

tionale for their inadequate evaluation is that theyare a form of therapy and must, therefore, be goodand worthwhile.

Finally, yet another pair of delusions, represent-ing polar opposites of one another, were discussedby Cronbach (1957) in his American Psycho-logical Association Presidential Address. Oneextreme, shown chiefly by the experimentalists,treats individual differences as merely bothersomevariationto be reduced by adequate controls ortreated as error variance in the search for GeneralLaws. Such assumptions cannot help but lead toan oversimplified image of man, for the simplifica-tion is introduced at the very beginning. We can-not expect a science of human behavior to advancefar until the moderating effects of individual vari-ation on the functional relationships being studiedare taken fully into account. People do, after all,differ greatly from one another and they differeven more from monkeys, white rats, or pigeons.It should not really be too heretical to suggestthat many of the lawful relations governing thebehavior of lower organisms may be inapplicableto the human species and, moreover, that lawsdescribing the behavior of certain selected humansubjectssuch as psychology sophomoresmayupon examination prove only weakly applicable tomany other individuals. It should be incumbentupon the experimentalist or the theorist either toincorporate a consideration of individual differ-ences into his research and theorizing or to defineexplicitly the individual parameters or populationcharacteristics within which he expects his laws tobe applicable.

The other extreme, actually extending consid-erably beyond the correlational psychology dis-cussed by Cronbach, is just as delusory and evenmore detrimental to the eventual development ofpsychology than the one just discussed. Differ-ences between individuals are regarded as so per-vasive that it is assumed no laws can be stated.The likely outcome of a strong commitment tothis point of view must ultimately be an admissionthat the methods of science cannot be applied tothe study of human behavior. Yet, this outcomeis not often openly recognized or honestly acceptedby those believing in the ultimate uniqueness ofeach individual. Instead, they speak of "newapproaches," less "mechanistic emphases," and amore "humanistic endeavor."

Cronbach, nearly a decade ago, sounded an

urgent call for his fellow psychologists to castaside the delusions represented by these two ex-tremes. Unfortunately, today we seem no closerto achieving this end than we were then.

THE SECKETS WE KEEPWe might better label this game "Dear God,

Please Don't Tell Anyone." As the name implies,it incorporates all the things we do to accomplishthe aim of looking better in public than we reallyare.

The most common variant is, of course, the tend-ency to bury negative results. I only recentlybecame aware of the massive size of this greatgraveyard for dead studies when a colleague ex-pressed gratification that only a third of his studies"turned out"as he put it.

Recently, a second variant of this secrecy gamewas discovered, quite inadvertently, by Wolins(1962) when he wrote to 37 authors to ask. for theraw data on which they had based recent journalarticles. Wolins found that of 32 who replied, 21reported their data to be either misplaced, lost, orinadvertently destroyed. Finally, after some ne-gotiation, Wolins was able to complete seven re-analyses on the data supplied from 5 authors. Ofthe seven, he found gross errors in threeerrors sogreat as to clearly change the outcome of the re-sults already reported. Thus, if we are to acceptthese results from Wolins' sampling, we mightexpect that as many as one-third of the studies inour journals contain gross miscalculations. Infact, this variant of the secrecy game might wellbe labeled "I Wonder Where the Yellow (data)Went." In commenting on Wolins' finding, Fried-lander (1964), impressed by the strong commit-ments psychologists hold for their theories, testsand methods, suggests that "Hope springs eternaland is evidently expressed through subjectivearithmetic"a possibility which is probably tooclose to the truth to be taken lightly.

Another extremely vexing and entirely unneces-sary type of secrecy is clearly apparent to anyonewho takes but a moment to page through one ofour current data-oriented psychological journals.I chose a recent issue of the Journal oj Personalityand Social Psychology. It was very difficult to findsuch mundane statistics as means or standard de-viations. Instead, the pages abounded with analy-sis of variance tables, charts, F ratios, and even ttests in the absence of their corresponding means

348 AMERICAN PSYCHOLOGIST

and SDs. The net effect of this is to make verydifficult and often impossible any further analysesthat a reader might want to undertake. The impli-cation of this, it seems to me, is that many authorshave actually failed to bother computing such sta-tistics as means or SDs and that, further, theyprobably have not examined their data with suffi-cient care to appreciate in any degree what theymay really portray.

Other examples of the secrecy game abound.They include such practices as dropping subjectsfrom the analysesa practice discussed at somelength in the critical review of a sampling of dis-sonance studies by Chapanis and Chapanis (1964),experimenter-biasing factors, incomplete descrip-tions of methodology, failure to carry out or toreport cross-validation studies, and the more gen-eral problem of failure to carry out or to reportreplication studies.

I believe you will agree that these tactics ofsecrecy can be nothing but severely damaging toany hopes of advancing psychology as a science.It seems likely that such practices are rather widelyapplied in psychology by psychologists. 1 sug-gest that we vow here and now to keep thesesecrecy games secret from our colleagues in theother sciences!

THE QUESTIONS WE ASKThere are many titles that might be appropriate

for this last game that I shall discuss. One mightbe, "Who's on First?"or better yet, "WhatGame Are We In?"or a rather common versionin these days of large Federal support for research,"While You're Up, Get Me a Grant." My majorpoint here is quite simply that the other games weplay, the pets we keep, our delusions, our secrets,and the Great Name Game interact to cause us tolose sight of the essence of the problems that needto be solved and the questions that need answers.The questions that get asked are dictatedall toooftenby investigators' pet theories or methods,or by the need to gain "visibility" among one'scolleagues. One of my respondentsa youngerbut undoubtedly wiser head than Isummed it upnicely.0 He said:Psychologists seem to be afraid to ask really importantquestions. The whole Zeitgeist seems to encourage researchefforts that earn big grants, crank out publications fre-quently and regularly, self-perpetuate themselves, don't

r > J . P. Campbell, personal communication, 1965.

entail much difficulty in getting subjects, don't require theresearchers to move from behind their desks or out oftheir laboratories except to accept speaking engagements,and serve to protect the scientist f rom all the forces thatcan knock him out of the secure "visible circle."

Another of my respondents, a verbal behaviorresearcher, illustrated the dilemma by mentioninga fellow researcher who phrased his research ques-tion as: "How do the principles of classical andinstrumental conditioning explain the learning oflanguage?" This sort of question is clearly illus-trative of the tendency to defer too readily to ex-isting popular points of view and to allow them todistort the direction of research activities. Itwould be better simply to ask "What is learned?"rather than making the premature assumptionsthat (a) language is learned in the sense that theterm learning is usually used or (b) all learning isof only two types.

An even more serious and, unfortunately, prob-ably more common form of the question-askinggame is the game of "Ha! Sure Slipped That OnePast You, Didn't [?" Here, the investigatorshrewdly fails to state the question he is trying toanswer, gathers data to provide answers to simplerquestions, and then behaves as if his research hasbeen relevant to other unstated but more impor-tant and more interesting problems. The vastmajority of studies devoted to measuring employeeattitudes have committed this error. It is notrick to develop questionnaires to gather systemati-cally the opinions of workers about their jobs. Itis quite something else, however, suddenly to be-gin talking about measures of employee motivationand to suggest that the employee responses havedirect relevance to what they may actually do onthe job. The literature on response set and re-sponse style is another clear case of new questionsbeing designed to fit existing answers. Showinghigh correlations between scores on empiricallydeveloped scoring keys and the numbers of itemskeyed True or some other item index should not betaken as having any bearing on the empirical va-lidities of these keys. Yet for over a decade, out-literature has been burdened with all sorts of setand style studies characterized by seemingly end-less factor analyses and silly arguments betweenpersons committed to acquiescence and those com-mitted to social desirability.

Thus, we all are far too eager to ask such ques-tions as: "What problems can be easily answered?"

FADS, FASHIONS, AND FOLDEKOL 349

"What else can I do with my test?" "What prob-lems or questions does my theory lead to?""What aspects of behavior can I study with mycomputer or with my apparatus?" or "What prob-lems can I find that I can fit this method to?"

Certainly, as psychologistsas scientists pre-sumably interested in the subject matter of humanbehaviorwe should be able to do better thanthis!

THE CAUSES

You may have inferred by now that I feel somesense of pessimism about the current state of psy-chology. Based on what you have heard so far,such an inference is probably appropriate. I dobelieve that the games I have described offer littlethat can be beneficial for psychology in the longrun. The behaviors underlying the games repre-sent enormous and essentially wasteful expendi-tures of our own research energy.

Even so, my mood is not basically pessimistic.In fact, we should be able to emerge from thissoul searching with a constructive sense of discon-tent rather than one of destructive despair. Thedescription of our condition carries with it a num-ber of implications for corrective action. More-over, we may infer from the condition some possi-ble causes, and listing them should also suggestpossible correctives.

To this end, let me consider briefly what I be-lieve to be the major causes of psychology's fads,fashions, and folderol.

The most important, I believe, is related quitedirectly to the relative insecurity of being a scien-tist, a problem that is particularly acute in psy-chology where we must cope with such complexphenomena as those involved in the study of hu-man behavior. The scientist's stance includes theconstant need to doubt his own work. Moreover,the long-range significance of his work cannot oftenbe forecast, and rarely can the scientistleast ofall, perhaps, the psychologistpreplan his inspira-tions and his ideas. It is little wonder, then, thatmany seek, through their theories, methodologies,or other of the games we have discussed, to organ-ize, systematize, and regularize their creative out-put. When viewed against the backdrop of publi-cation pressures prevailing in academia, the lure oflarge-scale support from Federal agencies, and thepresumed necessity to become "visible" amongone's colleagues, the insecurities of undertaking

research on important questions in possibly un-tapped and unfamiliar areas become even moreapparent. Stern (1964) has recently very effec-tively stated the case for the desirability of mov-ing from research into equally fulfilling careers ofteaching and administration. But we cannot for-get that the value system of science places researchand publication at the peak, and it should, there-fore, be no surprise that the less able researchersin psychologylearning early that no great break-through is in the offingsimply seek to eat theircake and have it too, by playing the games andthe song and dance of scientific research, usuallyconvincing even themselves that the games are "forreal" and that their activities really "make a dif-ference."

The perpetuation of this state of affairs is re-lated to our present system of graduate education.Many psychology graduate students today findthemselves under the tutelage of a faculty memberwho has bought the system wholeheartedly. Suchstudents live for a period of from 3 to 8 years inan environment that enforces and reinforces thelearning of a particular approach, a narrow pointof view or a set of pet methodologies which cometo define for them the things they will pursue aspsychologists.

THE REMEDY

But here I amsounding pessimistic and noxiousagain, and getting farther out on the limb than Ireally want to be.

In order to convince you of my good intentionsand my hope for the future, I had better get onwith some constructive suggestions. My suggestedremedyif it can be called that, for indeed it maybe more painful than the diseasecan be sum-marized in five imperative statements:

1. Give up constraining commitments to theo-ries, methods, and apparatus!

2. Adopt methods of multiple working hypoth-eses !

3. Put more eclecticism into graduate educa-tion !

4. Press for new values and less pretense in theacademic environments of our universities!

5. Get to the editors of our psychological jour-nals!

Let me elaborate briefly on each of these recom-mendations.

350 AMERICAN PSYCHOLOGIST

First, 1 advocate a more careful and studiedchoice of research questions. As should be ap-parent, I believe research energy should be di-rected toward questions that contain as few as pos-sible of any prior unproven assumptions about thenature of man. We must be constantly alert tothe narrowing of research perspectives due to priortheoretical or methodological commitments. I amcalling for less premature theorizingparticularlythat which leads to vaguely stated "wide-band"theories that are often essentially incapable of dis-proof.

"I am not advocating the abandonment of deduc-tion in psychology; in fact, psychology needsstronger and more specific deductions rather thanthe weak and fuzzy ones so typical of so manycurrent theories. What I am advocating is themore systematic study of lawful relationships be-fore interpretations are attempted. When explana-tion is attempted, the data should be sufficient toallow hypotheses to be stated with the clarity andprecision to render them directly capable of dis-proof. As the philosopher Karl Popper has said,there is no such thing as proof in science; scienceadvances only by disproofs.

This leads directly to my second recommenda-tion which is to state and systematically test mul-tiple hypotheses. Platt recently advocated this ap-proach which he calls Strong Inference (Platt,1965). The approach entails devising multiplehypotheses to explain observed phenomena, de-vising crucial experiments each of which may ex-clude or disprove one or more of the hypotheses,and continuing with the retained hypotheses todevelop further stibtests or sequential hypothesesto refine the possibilities that remain. This processdoes not seem new; in fact it is not. It simplyentails developing ideas or leads, stating alternativepossibilities, testing their plausibility, and proceed-ing to develop predictive and explanatory evidenceconcerning the phenomena under investigation.One might say that the research emphasis is oneof "studying hypotheses" as opposed to "sub-stantiating theories." The difference seems slight,but it is really quite important. However, inpsychology, the approach is little used, for, as wehave said, the commitments arc more often to atheory than to the process of finding out.

The method of multiple hypotheses takes ongreatly added power when combined with greatercare in the analysis and reporting of research re-

sults. Instead of serving as the sole statisticaltest of hypotheses, the statistical null hypothesisshould always be supplemented by estimates ofstrength of association. The psychologist owes itto himself to determine not only whether an as-sociation exists between two variablesan associa-tion which may often be so small as to be trivial-but also to determine the probable magnitude ofthe association. As Hays (1964) has suggested,if psychologists are content to adopt conventions(such as .OS or .01) for deciding on statisticalsignificance, they should also adopt conventionsconcerning the strength of association which maybe sufficiently large to regard as worthy of furtherinvestigation. Obviously, such conventions can-not be the same for all areas and for all researchquestions, but it should be clear that an emphasison magnitude estimation will demand that re-searchers give much more careful thought thanthey now do to defining ahead of time the actualmagnitudes that will be regarded as possessingeither theoretical or practical consequence.

By now, it is apparent why my fifth recom-mendation has to do with our journals. It willrequire a new kind of surveillance from both theeditors and their consultants if we are to imple-ment the greater care in research conception andin data analysis and reporting that I am advocat-ing. When and if null-hypothesis testing is ac-corded a lower position in the status hierarchy andcomes to be supplemented by emphases on StrongInference and magnitude estimation, I would pre-dict that the bulk of published material will, fora time, greatly diminish. That which does appear,however, will be guaranteed to be of considerablygreater consequence for furthering our understand-ing of behavior.

One of the possible loopholes in the method ofStrong Inference, it should be clear, is the greatdifficulty of designing and carrying out crucialexperiments. Recently Hafner and Presswood(1965) described how faulty experiments had ledphysicists astray for several decades as they soughtto explain the phenomenon of beta decay. Wemust broaden our conception of multiple hy-potheses to include as one quite plausible hypoth-esis the possibility of poorly conceived or poorlyconducted experiments. This, of course, simplyspeaks to the need for more replication in psy-chology of crucial experiments, a practice whichundoubtedly would become more widespread if

FADS, FASHIONS, AND FOLIIEROT, 351

psychologists possessed fewer of their own theo-relical pets and stronger motivation to examinesystematically whole sets of contending hypothesesand alternative explanations.

My third and fourth recommendations need notbe elaborated extensively. Both are intended tofoster less pretense in the conduct of psychologicalresearch by enabling those scholars who may beill fitted for the research enterprise to gain re-wards in other endeavors. The change in theacademic atmosphere would need to take the formof according more status to good teaching and togood administration. Perhaps this change wouldbe most rapidly fostered if the scientific gamesI have described would be more readily recognizedfor what they are and appropriately devalued inthe scheme of things within academia.

Obviously, greater eclecticism in graduate educa-tion is crucial to the successful outcome of myother suggestions. It is difficult to know how thiscan be implemented. But, at least, the goals seemclear. We desire to teach the core of psychology'sknowledge and methods, its subject matter and itsquestions, the statistical methods and their ap-propriate applicationsbut most of all, throughselection or training or both, we should seek toturn out persons with intense curiosity about thevast array of psychological questions and problemsoccurring everywhere in the world around us, witha willingness to ask open questions unhamperedby the prior constraints of a particular point ofview or method. Let us hope that graduate edu-cation, in the years ahead, will become moreeclectic and that even the Great Men in our fieldmay adopt a sense of humility when transmittingknowledge to the fledglings of our science.

THE OUTCOME: UTOPIAHow do I envision the eventual outcome if all

these recommendations were to come to pass?What would the psychologizing of the future looklike and what would psychologists be up to?

Chief among the outcomes, I expect, would bea marked lessening of tensions and disputes amongthe Great Men of our field. I would hope thatwe might once again witness the emergence of anhonest community of scholars all engaged in thezestful enterprise of trying to describe, under-stand, predict, and control human behavior.

Certainly our journals would be more meaty andless burdensome. There would be more honesty

in publishing the fruits of one's labors. Negativeresultsthe disproof of theoretical formulationsand the casting aside of working hypotheseswould be a more important part of the journals'contents. In consequence, the journals would con-tribute more meaningfully to the broad effort toachieve understanding, and we should expect towitness a sharp decline in the number of discon-nected little studies bearing little or no relation toeach other.

Moreover, I expect that many present schismsin psychology would be welded. The academic-professional bipolarity described by Tryon (1963)would be lessened, for the advantages to both ofclose association between basic researchers andthose practicing the art of psychology should be-come more apparent. The researchers wouldthereby establish and maintain contact with thereal world and real problems of human behavior,and the professional practitioners would be morefully alert to the need for assessing their methodsby generating and testing alternate deductions andhypotheses growing out of them.

Thus, in the long run we might hope for fewerdisputes, a spirit of more open cooperation, greaterinnovation in the generation and testing of work-ing hypotheses, greater care and precision in thedevelopment of theoretical formulations, and in-creased rigor in specifying the magnitude of out-comes such that they have both practical andtheoretical importance.

Does this sound like Utopia? Indeed it does.But is it too much to expect of a science now wellinto its second 100 years? I think not. Let usget on then with the process of change and ofreconsolidation.

REFERENCES

ASTIN, A. W. The functional autonomy of psychotherapy.American Psychologist,, 1961. 16, 75-78.

BERNE, E. Games people play. New York: Grove Press,1964.

BINDER, A. Further considerations on testing the nullhypothesis and the strategy and tactics of investigatingtheoretical models. Psychological Review, 1963, 70, J07-11S.

BRODY, N. Anxiety and the variability of word associates.Journal of Abnormal and Social Psychology, 1964, 68,331-334.

CHAMBERLIN, T. C. The method of multiple workinghypotheses. Science, 1965, 148, 754-759.

CHAPANIS, N., & CHAP AMIS, A. Cognitive dissonance: Fiveyears later. Psychological Bulletin, 1964, 61, 1-22.

352 AMERICAN PSYCHOLOGIST

CKONBACII, L. J. The two disciplines of scientific psychol-ogy. American Psychologist, 1957, 12, 671-684.

FEIGEL, H. Philosophical embarrassments of psychology.Psychologishe Reitrage, 1962, 6, 340-364.

FRIEDLANDEK, F. '-Type I and Type II bias. AmericanPsychologist, 1964, 19, 198-199.

GANZEK, V. J., & SARASON, I. G. Interrelationships amonghostility, experimental conditions, and verbal behavior.Journal of Abnormal and Social Psychology, 1964, 68,79-84.

GRANT, D. A. Testing the null hypothesis and the strategyand tactics of investigating theoretical models. Psycho-logical Review, 1962, 69, S4--61.

GUILFORD, J. P. Psychological measurement a hundredand twenty-five years later. Invited address presentedat American Psychological Association, Chicago, Septem-ber 1960.

HAFNER, E. M., & PRKSSWOOD, S. Strong inference andweak interactions. Science, 1965, 149, 503-510.

HAYS, W. L. Statistics for psychologists. New York:Holt, Rinehart & Winston, 1964.

HERZUERG, F., MAUSNER, B., & SNYDERMAN, B. The moti-vation to work. New York: Wiley, 1959.

MADDEN, J. M., & BOURDON, R. D. Effects of variationsin rating scale format on judgment. Journal of AppliedPsychology, 1964, 48, 147-151.

MAIER, N. R. F. Maicr's law. American Psychologist,1960, 15, 208-212.

McNEMAR, Q. Lost our intelligence? Why? AmericanPsychologist, 1964, 19, 871-882.

MEDNICK, M., MEDNICK, S. A., & JUNG, C. C. Continualassociation as a function of level of creativity and typeof verbal stimulus. Journal of Abnormal and SocialPsychology, 1964, 69, 511-515.

MEEHL, P. E., & HATHAWAY, S. R. The K factor as asuppressor variable in the MMPI. Journal of AppliedPsychology, 1946, 30, 525-564.

NUNNALLY, J. The place of statistics in psychology. Edu-cational and Psychological Measurement, 1960, 20, 641-650.

PI.ATT, J. R. Strong inference. Science, 1964, 146, 347-352.

ROZEBOOM, W. W. The fallacy of the null-hypothesis sig-nificance test. Psychological Bulletin, 1960, 57, 416-428.

SOMMER, R. On writing little papers. American Psycholo-gist, 1959, 14, 235-237.

STERN, C. Thoughts on research. Science, 1964, 148, 772-773.

TRYON, R. C. Psychology in flux: The academic-profes-sional bipolarily. American Psychologist, J963, 18, 134-143.

WOLINS, L. Responsibility for raw data. American Psy-chologist, 1962, 17, 657-658.