Journal of Criminal Law and Criminology

Volume 61 | Issue 4 Article 10

1971

Deterrence and the Death Penalty: AReconsiderationHugo Adam Bedau

Follow this and additional works at: https://scholarlycommons.law.northwestern.edu/jclc

Part of the Criminal Law Commons, Criminology Commons, and the Criminology and CriminalJustice Commons

This Criminology is brought to you for free and open access by Northwestern University School of Law Scholarly Commons. It has been accepted forinclusion in Journal of Criminal Law and Criminology by an authorized editor of Northwestern University School of Law Scholarly Commons.

Recommended CitationHugo Adam Bedau, Deterrence and the Death Penalty: A Reconsideration, 61 J. Crim. L. Criminology & Police Sci. 539 (1970)

Tna JounxAL or CniMxAX LAw, CaloflLooY AMW Ponr o Scrzom Vol. 61, No. 4Copyri*

6t 0 1971 by Northwestern University School of Law Prined in U.S.A.

CRIMINOLOGYDETERRENCE AND THE DEATH PENALTY: A RECONSIDERATION

HUGO ADAM BEDAU,

This empirical reconsideration of Professor Van den Haag's analysis of the deterrent effect ofthe death penalty (published in the June, 1969 issue of this Journal) attacks the loose frameworkof the analysis as a prototype of research in this area. It sharpens the questions and cuts away themisleading conclusions surrounding the death penalty controversy and challenges criminologistsand the legislatures to determine empirically whether deterrence exists as a factor in the considera-tion.

Professor Van den Haag's recent article, OnDeterrence and the Death Penalty," raises a numberof points of that mixed (i.e., empirical-and-concep-tual-and-normative) character which typifies mostactual reasoning in social and political controversybut which (except when its purely formal aspectsare in question) tends to be ignored by philoso-phers. This discussion will pass by any number oftempting points in his critique in order to focus indetail only on those which affect his asserted majortopic-the issue of deterrence as it bears on the re-tention or abolition of the death penalty.

Van den Haag's main contentions appear to bethe following:

(1) Abolitionists of a utilitarian persuasion"claim that capital punishment is useless becauseit does not deter others .... )2

(2) There are some classes of criminals and somecircumstances for which "the death penalty is theonly possible deterrent." 1

(3) As things currently stand, "deterrence [ofcriminal homicide by the death penalty] has notbeen demonstrated statistically;" but it is errone-ous to assume that "non-deterrence" has beendemonstrated statistically. 4

(4) The death penalty is to be favored overimprisonment because "the added severity of thedeath penalty adds to deterrence, or may do so." 6

* Professor of Philosophy, Tufts University.'Van den Haag, On Deterrence and the Death Penalty,

60 J. Ctim. L. C. & P.S. 141 (1969). This is a "re-vised version" under the same title of an article whichfirst appeared in 78 E.Tcs 280 (1968). The authoris grateful to Professor Van den Haag for the provisionof a reprint of each version of the article.

2 60 J. CzRw. L. C. & P.S. 141 (1969).3 Id. at 145.4Id.r, Id. at 146.

(5) "Since it seems more important to sparevictims than to spare murderers, the burden ofproving that the greater severity inherent in irre-vocability adds nothing to deterrence lies on thosewho oppose capital punishment." I

The refutation of the foregoing assertionswill constitute the task of this article. The re-buttal arguments may be succinctly summarized asfollows: regarding (1), utilitarian abolitionistsdo not argue as Van den Haag claims, and theywould be in error if they did; his assertion in (2),that situations exist in which the death penalty isthe only possible deterrent, is misleading and, inthe interesting cases, is empirically insignificant;concerning (3), the heart of the dispute, Van denHaag is correct in affirming that deterrence has notbeen determined statistically, but he is incorrectin denying that non-deterrence has been demon-strated statistically; his suggestion, (4), that theadded severity of the death penalty contributesto its deterrent function, is unempirical and one-sided as well; finally, his contention regarding theburden of proof, (5), which he would impose en-tirely upon abolitionists, is a dodge and is based ona muddled analysis.

The reason for pursuing in some detail what atfirst might appear to be mere polemical contro-versy is not that Professor Van den Haag's essay isso persuasive nor that it is likely to be of unusualinfluence. The reason is that the issues he raises,even though they are familiar, have not beenadequately discussed, despite a dozen state,congressional, and foreign government investiga-tions into capital punishment in recent years. InMassachusetts, for example, several persons undersentence of death have been granted stays of execu-

6 Id. at 147.

HUGO ADAM BEDA U

tion pending the final report of a special legislativecommission to investigate the death penalty. Theexclusive mandate of this commission is to studythe question of deterrence.7 Its provisional conclu-sions, published late in 1968, though not in linewith Professor Van den Haag's views, are open tothe kind of criticism he makes. This suggests thathis reasoning may be representative of many whohave tried to understand the arguments and re-search studies brought forward by those who wouldabolish the death penalty, and therefore that hiserrors are worth exposure and correction.

I

The claim Van den Haag professes to find "mostpersuasive"- "capital punishment is useless be-cause it does not deter others"-is strange, and itis strange that he finds it so persuasive. Anyonewho would make this claim must assume that onlydeterrent efficacy is relevant to assessing the utilityof a punishment. In a footnote, Van den Haagimplicitly concedes that deterrence may not be theonly utilitarian consideration, when he asserts thatwhatever our penal "theory" may tell us, "deter-rence is ... the main actual function of legal pun-ishment if we disregard nonutilitarian ones." 8 Buthe does not pursue this qualification. It may beconceded that if 'the main actual function' meansthe main intended or professed function of a pun-ishment for those responsible for instituting it,deterrence is probably the main function of punish-ment. His definition of deterrence, however, re-mains vulnerable. According to Van den Haag, itis "a preconscious, general response to a severe butnot necessarily specifically and explicitly appre-hended or calculated threat." 9

This definition of deterrence has two merits andat least one fatal defect. First, it preserves the ideathat "a law can have no deterrent effect upon apotential criminal if he is unaware of its exist-ence." 1 0 Surely, this is a truism necessary to theestablishment of a definition of 'deterrence'.Second, by emphasizing threats, it avoids theerrors in defining deterrence as "the preventative7 See ch. 150, Mass. Acts & Resolves 929 (1969);

MASS. LEGISLATIVE REPOR.T, INTERhxK REPORT OF THESPEcIAL Comi sSIoN EsTABLISiED To MAKE ANINVESTIGATION AND ST n RELATIVE TO TnE EFFEc-TIVENESS OF CAPITAL PUNI5SMIET AS A DETERENTTO Cann (1968) (unpublished).8 Van den Haag, supra note 1, at 147 n. 11 (empha-

sis added).9 Id. at 146.10 Ball, The Deterrence Concept in Criminology and

Law, 46 J. CPu. L., C. & P.S. 347, 351 (1955).

effect which actual or theoretical punishment ofoffenders has upon potential offenders." n On sucha definition, one could not distinguish between thedeterrent effect of the death penalty and its moreinclusive preventive effects. Obviously, an executedcriminal is prevented from further crimes, but notby having been deterred from them.P

Only rarely will the preventive and the deterrenteffects of a given punishment be equivalent. Vanden Haag's definition, however, falls before a simi-lar objection upon consideration of the general,though by no means universal, desire of persons toavoid capture and punishment for the crimes theycommit. Some criminologists have thought thisdesire to be the primary outcome of severe punish-ments. If so, then the outcome can result whetheror not the deterrent function succeeds. Yet such adesire to avoid punishment is embraced by Van denHaag's rubric of "general response" and thereforecould count as evidence for the deterrent efficacy ofa punishment! Since Van den Haag's conception ofdeterrence does not discriminate between suchfundamentally different types of "general re-sponse" to the threat of punishment, it is too ill-formulated as a definition to be of any serious use.

Among the ideas to be incorporated into anydefinition of deterrence are a pair of truisms: ifsomeone has been deterred then he doesn't com-mit the crime, and conversely if someone doescommit a crime then he hasn't been deterred.Likewise, the key notion in deterrence is preven-tion by threat of punishment. Therefore, assume(Definition 1) that a given punishment (P) is adeterrent for a given person (A) with respect to agiven crime (C) at a given time () if and only ifA does not commit C at t because lie believes heruns some risk of P if he commits C and A pre-fers, ceteris paribus, not to suffer P for committingC. This definition does not presuppose that Preally is the punishment for C (a person could bedeterred through a mistaken belief); it does notpresuppose that A runs a high risk of incurring P(the degree of risk could be zero); or that A con-sciously thinks of P prior to t (the theory neededto account for the operation of A's beliefs and

n Id. at 347."Ball writes that "Capital punishment can be

totally effective as a deterrent ... The executedmurderer is no longer a threat to society. He has beenpermanently deterred." Id. at 353. This is an erroneousconclusion to reach, and when Ball goes on to use it toargue in favor of the deterrent efficacy of the deathpenalty, it reveals the menace which lies hidden in afaulty definition.

[Vol. 61

DETERRENCE AND THE DEATH PENALTY

preferences on his conduct is left open). Nor doesit presuppose that anyone ever suffers P (P couldbe a "perfect" deterrent), nor that only P couldhave deterred A from C (some sanction less severethan P might have worked as well). Finally, itdoes not presuppose that because P deters A att from C, therefore P would deter A at any othertime or anyone else at t. The definition insuresthat we cannot argue erroneously from the factthat A does not commit C to the conclusion thatP has succeeded as a deterrent: the definitioncontains conditions which prevent this. Further,the definition prevents the commission of the moresubtle converse error of arguing from the fact thatA has not been deterred by P to the conclusionthat A will (or must have) commit(ted) C. Boththese errors arise from supposing that "the educa-tive, moralizing and habituative effects of punish-ment", 3 which serve to prevent the bulk of thepublic from committing crime, are euphemisms for'deterrence' or operate by the same mechanismsthat deterrence does.

Definition 1 suggests a general functional ana-logue appropriate to express scientific measure-ments of differential deterrent eficacy of a givenpunishment for a given crime with respect to agiven population (Definition 2). Let us say that agiven punishment P deters a given population Hfrom a crime C to the degree D that the membersof H do not commit C because they believe thatthey run some risk of P if they commit C and,ceteris paribus, they prefer not to suffer P forcommitting C. If D = 0, then P has completelyfailed as a deterrent, whereas if D = 1, P hasproved to be a perfect deterrent. Given this defini-tion and the appropriate empirical results forvarious values of P, C, and H, it should be possibleto establish on inductive grounds the relativeeffectiveness of a given punishment (the value ofD) as a deterrent.

Definition 2 in turn suggests the followingcorollary for assertions of relative superior deter-rent efficacy of one punishment over another: agiven punishment PI is a superior deterrent toanother punishment P2 with respect to some crimeC and some population H if and only if: if themembers of H believe that they are liable to P 1upon committing C, then they commit C to thedegree di; whereas if the members of H believethat they are liable to P2 upon committing C, then

13Zimring and Hawkins, Deterrence and MarginalGroups, 5 JoumlA. or RE EARcn iN CR ANmD L QUENcy 100 (1968).

they commit C to the degree d2; and di < d2 . Thisformulation plainly allows the.P 1 may be a moreeffective deterrent than P2 for C1 and yet lesseffective as a deterrent than P2 for a different crimeC2 (with H constant), and so forth for other pos-sibilities. When speaking about deterrence in thesections which follow, these definitions and thiscorollary are presupposed.

Even if Van den Haag's notion of deterrence didnot need to be reformulated to incorporate theabove improvements, there would still be adecisive objection to his claim. Neither classic norcontemporary utilitarians have argued for oragainst the death penalty solely on the ground ofdeterrence, nor would their ethical theory entitlethem to do so. One measure of the non-deterrentutility of the death penalty derives from itselimination (through death of a known criminal)of future possible crimes from that source; anotherarises from the elimination of the criminal's proba-ble adverse influence upon others to emulate hisways; another lies in the generally lower budgetaryoutlays of tax monies needed to finance a systemof capital punishment as opposed to long-termimprisonment. There are still further consequencesapart from deterrence which the scrupulousutilitarian must weigh, along with the threepreviously mentioned. Therefore, it is incorrect toassume that a demonstrated failure of the deter-rent effect of the death penalty would generate aninference, on utilitarian assumptions, that "thedeath penalty is useless" and therefore ought tobe abolished. The problem for the utilitarian is tomake commensurable such diverse social utilitiesas those measured by deterrent efficacy, administra-tive costs, etc., and then to determine which penalpolicy in fact maximizes utility. Finally, inspectionof sample arguments actually used by abolitionists 4

will show that Van den Haag has attacked a strawman: there are few if any contemporary abolition-ists (and Van, den Haag names none) who arguesolely from professedly utilitarian assumptions,and there is none among the non-utilitarians whowould abolish the death penalty solely on groundsof its deterrent inefficacy.

II

Governments faced by incipient rebellion orthreatened by a coup d'etat may well conclude, asVan den Haag insists they should, that rebels (as

14See the several essays reprinted in H. BEDAu,THE DEATH PENALTY IN AMERIcA 166-70 (Rev. ed.1967).

HUGO ADAM BEDAU

well as traitors and spies) can be deterred, if atall, by the threat of death, since "swift victory" ofthe revolution "will invalidate [the deterrentefficacy] of a prison sentence." 15 But this does notreveal the importance of providing such deterrence,any more than the fact that a threat of expulsionis the severest deterrent available to universityauthorities reveals whether they should insist onexpelling campus rebels. Also, since severe penal-ties might have the effect of creating martyrs forthe cause, they could provoke attempts to over-throw the government to secure a kind of politicalsainthood. This possibility Van den Haag recog-nizes but claims in a footnote that it "hardlyimpairs the force of the argument." 16 From alogical point of view it impairs the argumentconsiderably; from an empirical point of view,since one is wholly without any reliable facts orhypotheses on politics in such extreme situations,the entire controversy remains quite speculative.

The one important class of criminals deterrable,if at all, by the death penalty consists, accordingto Van den Haag, of those already under "life"sentence or guilty of a crime punishable by "life".In a trivial sense, he is correct; a person alreadysuffering a given punishment, P, for a given crime,C1, could not be expected to be deterred by anti-cipating the re-infliction of P were he to commitC2. For if the dread of P did not deter him fromcommitting C1, how could the dread of P deterhim from committing C2 given that he is alreadyexperiencing P? This generalization seems toapplywheneverP = "life" imprisonment. Actually,the truth is a bit more complex, because in practice(as Van den Haag concedes, again in a footnote)so-called "life" imprisonment always has its ag-gravations (e.g., solitary confinement) and itsmitigations (parole eligibility). These make itlogically possible to deter a person already con-victed of criminal homicide and serving "life"imprisonment from comnmitting another suchcrime. The aggravations available are not, inpractice, likely to provide much added deterrenteffect; but exactly how likely or unlikely thiseffect is remains a matter for empirical investiga-tion, not idle guesswork. Van den Haag's seemingtruism, therefore, relies for its plausibility on thefalse assumption that "life" imprisonment is a

15 Van den Haag, supra note 1, at 145. The sameargument has been advanced earlier in Hook, 7 TnxNEw YoRK LAw FoRum 278-83 (1961). For therevised version of this argument, see H. BEDAu, supranote 14, at 150-51.

16 Van den Haag, supra note 1, at 145 n. 8.

uniform punishment not open to further deter-rence-relevant aggravations and mitigations.

Empirically, the objection to his point is thatpersons already serving a "life" sentence do not ingeneral constitute a source of genuine alarm tocustodial personnel. Being already incarcerated andintegrated into the reward structure of prisonlife, they do not seem to need the deterrent controlsallegedly necessary for other prisoners and thegeneral publicP There are convicts who are excep-tions to this generalization, but there is no knownway of identifying them in advance, and theirnumber has proved to be small. It would be ir-rational, therefore, to design a penal policy whichinvokes the death penalty for the apparent purposeof deterring such convicted offenders from furthercriminal homicide.18 Van den Haag cites no evi-dence that such policies accomplish their allegedpurpose, and a review of authorities reveals none.The real question which Van den Haag's argumentraises is: Is there any class of actual or potentialcriminals for which the death penalty exerts amarginally superior deterrent effect over everyless severe alternative? With reference to thisquestion there is no evidence at all, one way orthe other. Until a determination is made as towhether there is a "marginal group" for whom thedeath penalty serves as a superior deterrent, thereis no reason to indulge Van den Haag in his specu-lations 9

III

It is not clear why Van den Haag is so anxiousto discuss whether there is evidence that thedeath penalty is a deterrent, or whether, as hethinks, there is no evidence that it is not a deter-rent. For the issue over abolishing the deathpenalty, as all serious students of the subject haveknown for decades, is not whether (1) the deathpenalty is a deterrent, but whether (2) the deathpenalty is a superior deterrent to "life" imprison-ment, and consequently the evidential dispute isalso not over (1) but only over (2). As this author

17 See, e.g., Sellin, Prison Homicides, in CATAL.PuNIsmExNT 154-160 (T. Sellin ed. 1967).

1 Rhode Island (1852), North Dakota (1915), NewYork (1965), Vermont (1965), and New Mexico (1969),have all qualified their abolition of the death penaltyby enacting such a policy. See H. BEDAu, supra note14, at 12.

29 Zinring and Hawkins, supra note 13, at 104-05,explain that by a marginal group they mean "theentire class of persons who are objectively on the marginof a particular form of criminal behavior, or, in otherwords, the class of persons 'next most likely' to engagein criminal behavior in question."

[Vol. 61

DETERRENCE AND THE DEATH PENALTY

has argued elsewhere, 0 abolitionists have reasonto contest (1) only if they are against all punitivealternatives to the death penalty. Since fewabolitionists (and none cited by Van den Haag)take this extreme view, and since most are, infact, reconciled to a punitive alternative of "life"imprisonment, we may concentrate on (2) here. Itshould be noticed in passing, however, that if (1)could be demonstrated to be false, there would beno need for abolitionists to marshall evidenceagainst (2). Since the truth of (1) is a presupposi-tion of (2), the falsity of (1) would obviate (2)entirely. While it is true that some abolitionistsmay be faulted for writing as if the falsity of (1)followed from the falsity of (2), this is not acomplaint Van den Haag makes nor is it an errorof inference upon which the argument against thedeath penalty depends. Similar considerationsinveigh against certain pro-death penalty argu-ments. Proponents must do more than establish(1), they must also provide evidence in favor of(2); and they cannot infer from evidence whichestablishes (1) that (2) is true or even probable(unless, of course, that evidence would establish(2) independently). These considerations show ushow important it is to distinguish (1) and (2) andthe questions of evidence which each raises. Vanden Haag never directly discusses (2); he onlyobserves in passing that "the question is not onlywhether the death penalty deters but whether itdeters more than alternatives .... "21 Since heexplicitly argues over the evidential status of (1),it is unclear whether he chose to ignore (2) orwhether he thinks that his arguments regardingthe evidence for (1) also have consequences for(2). Perhaps Van den Haag thinks that if there isno evidence disconfirming (1), then there can beno evidence disconfirming (2); or perhaps he thinksthat none of the evidence disconfrming (2) alsodisconfirms (1). (If he thinks either, he is wrong.)Or perhaps he is careless, conceding on the onehand that (2) is important to the issue of abolitionof the death penalty, only to slide back into adiscussion exclusively about (1).

Van den Haag writes as if his chief contentionswere these two: first, we must not confuse (a) theassertion that there is no evidence that (1), with(b) the assertion that there is evidence that not-(1), i.e., evidence that (1) is false; and second,abolitionists have asserted (b) whereas all they are

2 0 H. BEDAu, supra note 14, at 260-61.Van den Haag, supra note 1, at 145.

entitled to assert is (a).2 I grant, as anyone must,that the distinction between (a) and (b) is legiti-mate and important. But since, as I have argued,(1) need not be at issue in the death penaltycontroversy, neither are (a) and (b). What is atissue, even though Van den Haag's discussionobscures the point, is whether abolitionists mustcontent themselves with asserting that there is noevidence against (2), or whether they may gofurther and assert that there is evidence that not-(2) (evidence that (2) is false). Whereas Van denHaag would presumably confine abolitionists tothe former, weaker assertion, it shall be arguedthat they may make the stronger, latter, assertion.

In order to see the issue fairly it is necessary tosee how (2) has so far been submitted to empiricaltest. First of all, the issue has been confined to thedeath penalty for criminal homicide; consequently,it is not (2) but a subsidiary proposition whichcritics of the death penalty have tested- (2a) thedeath penalty is a superior deterrent to "life" im-prisonment for the crime of criminal homicide. Thefalsification of (2a) does not entail the falsity of(2); the death penalty could still be a superiordeterrent to "life" imprisonment for the crime of

2 Van den Haag accuses Professor Thorasten Sellin,a criminologist "who has made a careful study of theavailable statistics," of appearing to "think that thislack of evidence for deterrence is evidence for the lackof deterrence." Id. That is, Van den Haag claims Sellinthinks that (a) is (b) ! Sellin's writings, see, e.g., note17 supra, do not support the contention that he"thinks" the one "is" the other. A review of his writ-ings, which span the years from 1953-1967, will reveala certain vacillation between the two manners ofstating his conclusion. His most recent statement isunqualified in the (b) form. See Sellin, supra note 17,at 138. Since Van den Haag also cited this author's,THE DEATH PENALTY IN AMERCA, supra note 14,though not in this connection, it should be added thatthe distinction between (a) and (b) was there made;but it was not insisted, as it is here, that the argumententitles'abolitionists to assert (b). See id. at 264-65.For the views of writers, all criminologists, who haverecently stated the same or a stronger conclusion, see,e.g., Chambliss, Types of Deviance and the Effectivenessof Legal Sanctions, 1967 Wis. L. REv. 703, 706 (1967)("Capital punishment does not act as an effectivedeterrent to murder"); Morris & Zimring, Deterrenceand Correction, 381 THE ANars 137, 143 (1969)("The capital punishment controversy has producedthe most reliable information on the general deterrenteffect of a criminal sanction. It now seems establishedand accepted that... the death penalty makes nodifference to the homicide rate..."); Reckless, TheUse of the Death Penalty, 15 Carm & DELIrNQ. 43, 52(1969) ("ITihe evidence indicates that [the deathpenalty for murder] has no discernible effects in theUnited States .... "); Doleschal, The Deterrent Effect ofLegal Punishment, 1 ThromAOrio REv. oN Cpa,AND DELiNQ. 1, 7 (1969) ("Capital punishment isineffective in deterring murder").

19701

HUGO ADAM BEDAU

burglary, etc. However, the disconfirmation of(2a) would be obviously a significant partial dis-confirmation of (2). Secondly, (2a) has not beentested directly but only indirectly. No one hasdevised a way to count or estimate directly thenumber of persons in a given population who havebeen deterred from criminal homicide by the fearof the penalty. The difficulties in doing so areplain enough. For instance, it would be possibleto infer from the countable numbers who have notbeen deterred (because they did commit a givencrime) that everyone else in the population wasdeterred, but only on the assumption that theonly reason why a person did not commit a givencrime is because he was deterred. Unfortunately forthis argument (though happily enough otherwise)this assumption is almost certainly false, as wehave noted above in section I. Other methodswhich might be devised to test (2a) directly haveproved equally unfeasible. Yet it would be absurdto insist that there can be no evidence for or against(2a) unless it is direct evidence for or against it.Because Van den Haag nowhere indicated what hethinks would count as evidence, direct or indirect,for or against (1), much less (2), his insistenceupon the distinction between (a) and (b) and hisrebuke to abolitionists is in danger of implicitlyrelying upon just this absurdity.

How, then, has the indirect argument for (2a)proceeded? During the past generation, at least sixdifferent hypotheses have been formulated, ascorollaries of (2a), as follows :2

(i) death penalty jurisdictions should have a2 The relevant research, regarding each of the six

hypotheses in the text, is as follows:(i) Schuessler, The Deterrent Influence of the Death

Penalty, 284 THE ANNALS 54, 57 (1952); Reckless,The Use of the Death Penalty-A Factual Statement,15 CRn AN DELINQ. 43,52 (1969) (Table No. 9).

(ii) Thorsten Sellin, The Death Penalty, reprintedin H. BEDAu, supra note 14, at 274-84; updated inSellin, supra note 17, at 135-38.

(iii) Sellin, supra note 17, at 34-38; reprinted inH. BEDAU, supra note 14, at 339-43.

(iv) See works cited in (i).(v) CANADA, MruWms Am PROCEEMINGS OF Evi-

DENCE, JOINT COMM=rTEE oF Tm SENATE ANDHousE oF CommoNs OF CAPITAL PuNmmINT ANDCOmOREAL PUNISHNT AND LoTTERIEs and THESTATE PoLicE AND =H DEATH PENA.TY, app. F, pt.I, at 718-35 (1955); The Death Penalty and PolceSafety, in H. BEDAU, supra note 14, at 284-301, andin CAPITAL PuNis;nirm, supra note 17, at 138-54.(Vi) MASSACHUSETTs, REPORT AND REcomENDA-

TIONS OF THE SPECIAL COiMtSSION [ON] mx DEATHPENALTY, 1958, in H. BEDAu, supra note 14, at 400;Sellin, Prison Hornocides, in CeiTA PuNIsmENT,supra note 17, at 154-60.

lower annual rate of criminal homicide thanabolition jurisdictions;

(ii) jurisdictions which abolished the deathpenalty should show an increased annualrate of criminal homicide after abolition;

(iii) jurisdictions which reintroduced the deathpenalty should show a decreased annual rateof criminal homicide after reintroduction;

(iv) given two contiguous jurisdictions differingchiefly in that one has the death penalty andthe other does not, the latter should show ahigher annual rate of criminal homicide;

(v) police officers on duty should suffer a higherannual rate of criminal assault and homicidein abolition jurisdictions than in death penaltyjurisdictions;

(vi) prisoners and prison personnel should suffera higher annual rate of criminal assault andhomicide from life-term prisoners in abolitionjurisdictions than in death penalty jurisdic-tions.

It could be objected to these six hypotheses thatthey are, as a set, insufficient to settle the questionposed by (2a) no matter what the evidence forthem may be-that the falsity of (i)-(vi) does notentail the falsity of (2a). Or it could be objectedthat each of (i)-(vi) has been too inadequatelytested or insufficiently disconfirmed to establishany disconfirmation of (2a), even though it isconceded that if (i)-(i) were highly disconfirmedthey would disconfirm (2a). Van den Haag's lineof attack is not entirely clear as between these twoalternatives. It appears that he should take theformer line of criticism in its most extreme version.How else could he argue his chief point, that theresearch used by abolitionists has so far failed toproduce any evidence against (1)-we may takehim to mean (2) or (2a)? Only if (i)-(vi) wereirrelevant to (2a) could it be fairly concluded fromthe evidential disconfirmation of (i)-(vi) thatthere is still no disconfirmation of (2a). And this isVan den Haag's central contention. The otherways to construe Van den Haag's reasoning aretoo implausible to be considered: he cannot thinkthat the evidence is indIfferent to or confirms(i)-(vi); nor can he think that there has been noattempt at all to disconfirm (2a); nor can he thinkthat the evidence which disconfirms (i)-(vi) is nottherewith also evidence which confirms the nega--tions of (i)-(vi). If any of these three were true itwould be a good reason for saying that there is"no evidence" against (2a); but each is patentlyfalse. If one inspects (i)- (vi) and (2a), it is difficultto see how one could argue that disconfirmation of

[Vol. 61

DETERRENCE AND THE DEATH PENALTY

the former does not constitute disconfirmation ofthe latter, even if it might be argued that verifica-tion of the former does not constitute verificationof the latter. Therefore, there is nothing to begained by further pursuit of this first line ofattack.

Elsewhere, Van den Haag seems to adopt thealternative criticism, albeit rather crudely, as whenhe argues (against (iv), seemingly, since he no-where formulated (i)-(vi)) that "the similar areasare not similar enough." 21 He fails to explain whythe rates of criminal homicide in Michigan and inIllinois from 1920 to 1960 are not relevant, butsimply alleges that the states aren't "similarenough." His criticism does, however, tacitlyconcede that if the jurisdictions were "similarenough," then it would be logically possible toargue from the evidence against (iv) to the dis-confirmation of (2a). And this seems to be inkeeping with the nature of the case. Thus it is thissecond line of attack which needs closer examina-tion.

Van den Haag's own position and objectionsapart, what is likely to strike the neutral observerwho studies the ways in which (i)- (vi) have beentested and declared disconfirmed is that theirdisconfirmation, and afortiori, the disconfirmationof (2a), is imperfect for two related reasons. First,all the tests rely upon unproved empirical assump-tions; second, it is not known whether there is anystatistical significance to the results of the tests. Itis important to make these concessions, andabolitionists and other disbelievers in the deterrentefficacy of the death penalty have not always doneSO.

It is not possible here to review all the evidenceand reach a judgment on the empirical status of(i)-(vi). But it is possible and desirable to illus-trate how the two qualifications cited above mustbe understood, and then to assess their effect onthe empirical status of (2a). The absence ofstatistical significance may be illustrated byreference to hypothesis (v). According to thepublished studies, the annual rate of assaults uponon-duty policemen in abolition jurisdictions islower than in death penalty jurisdictions05 Butthe studies do not answer whether the difference isstatistically significant because the data were notsubmitted to tests of statistical significance. Nor is

21 Van den Haag, supra note 1, at 146.25 A rate of 1.2 attacks per 100,000 population in

abolition jurisdictions as opposed to 1.3 per 100,000population in death penalty jurisdictions.

there any known method by which the data couldbe subjected to any such tests. This is, of course,no reason to suppose that the evidence is reallynot evidence after all, or that though it is evidenceagainst (i) it is not evidence against (2a). Statis-.tical significance is, after all, only a measure of thestrength of evidence, not a sine qua non of eviden-tial status.

The qualification concerning unproved assump-tions is more important, and is worth examiningsomewhat more fully (though, again, only illustra-tively). Consider hypothesis (i). Is one entitled toinfer that (i) is disconfirmed because in fact astudy of the annual homicide rates (as measuredby vital statistics showing cause of death) un-questionably indicates that the rate in all abolitionstates is consistently lower than in all death penaltystates? To make this inference one must assumethat (Ai) homicides as measured by vital statisticsare in a generally constant ratio to criminalhomicides, (A2) the years for which the evidencehas been gathered are representative and notatypical, (As) however much fluctuations in thehomicide rate owe to other factors, there is anon-negligible proportion which is a function ofthe severity of the penalty, and (A4) the deterrenteffect of a penalty is not significantly weakened byits infrequent imposition. There are, of course,other assumptions, but these are central andsufficiently representative here. Assumption A, iseffectively unmeasurable because the concept of acriminal homicide is the concept of a homicidewhich deserves to be criminally prosecuted. Never-theless, A, has been accepted by criminologists forover a generation. A2 is confirmable, on the otherhand, and bit by bit, a year at a time, seems to bebeing confirmed. Assumption As is rather moreinteresting. To the degree to which it is admittedor insisted that other factors than the severity ofthe penalty affect the rate of homicide, to thatdegree As becomes increasingly dubious; but at thesame time testing (2a) by (i) becomes increasinglyunimportant. The urgency of testing (2a) restsupon the assumption that it is the deterrentefficacy of penalties which is the chief factor in therate of crimes, and it is absurd to hold that assump-tion and at the same time doubt As. On the otherhand, A4 is almost certainly false (and has beenbelieved so by Bentham and other social theoristsfor nearly two hundred years). The falsity of A4,however, is not of fatal harm to the disconfirma-

26 For a discussion surrounding this point see, H.BEDAu, supra note 14, at 56-74.

1970]

HUGO ADAM BEDAU

tion of (i) because it is not known how infrequentlya severe penalty such as death or life imprison-ment may be imposed without decreasing itsdeterrent efficacy. The available information onthis point leads one to doubt that for the generalpopulation the frequency with which the deathsentence is imposed makes any significant differ-ence to the volume of criminal homicide.

These four assumptions and the way in whichthey bear upon interpretation and evaluation of theevidence against (i), and therefore the discon-firmation of (2a), are typical of what one finds asone examines the work of criminologists as itrelates to the rest of these corollaries of (2a). Is itreasonable, in the light of these considerations, toinfer that there is no evidence against (i)-(vi), orthat although there may be evidence against(i)- (vi), there is none against (2a)? Probably not.Short of unindentified and probably unobtainable(Ccrucial experiments," it is impossible to marshallevidence for (2a) or for (i)-(vi) except by meansof certain additional assumptions such as A1-A 4.To reason otherwise is to rely on nothing morethan the fact that it is logically possible to grantthe evidence against (i)-(vi) and yet deny that(2a) is false; or it is to insist that the assumptionswhich the inference relies upon are not plausibleassumptions at all (or though plausible are them-selves not confirmed) and that no other assump-tions can be brought forward which will both beimmune to objections and still preserve the linkagebetween the evidence, (i)-(vi), and (2a). Thedanger now is that one will repudiate assumptionssuch as A1-A4 so as to guarantee the failure ofefforts to disconfirm (2a) via disconfirmation of(i)-(vi); or else that one will place the standardsof evidence too high before one accepts the dis-confirmation. In either case one has begun toengage in the familiar but discreditable practice of"protecting the hypothesis" by making it ineffect immune to any kind of disconfirmation.

In sum, then, the abolitionist's argument re-garding deterrence has the following structure:an empirical proposition not directly testable,(2), has a significant corollary, (2a), which in turnsuggests a number of corollaries, (i)-(vi), each ofwhich is testable with varying degrees of indirect-

See R. DANN, Tim DETErRNT ErIECT Or CAPITALPuismma (1935); Savitz, A Study in Capital Punish-ment, 49 J. Gn. L. C. & P. S. 338, 338-41 (1958)(reprinted in H. BEDAU, supra note 14, at 315-32);Graves, A Doctor Looks at Capital Punishment, 10 Mix-.ARTs & ScrENc.Es 137, 137-41 (1956) (reprinted in H.BEDAu, supra note 14, at 322-32).

ness. Each of (i)-(vi) has been tested. To acceptthe results as evidence disconfirming (i)-(vi) andas therefore disconfirming (2a), it is necessary tomake certain assumptions, of which A1-A 4 aretypical. These assumptions in turn are not alltestable much less directly confirmed; some ofthem, in their most plausible formulation, mayeven be false (but not in that formulation neces-sary to the inference, however). Since this struc-ture of indirect testing, corollary hypotheses, un-proved assumptions, is typical of the circumstanceswhich face us when we wish to consider the evidencefor or against any complex empirical hypothesissuch as (2), I conclude that while (2) has by nomeans been disproved (whatever that might mean),it is equally clear that (2) has been disconfirmed,rather than confirmed or left untouched, by theinductive arguments surveyed.

An attempt has been made to review and ap-praise the chief "statistical" arguments, as Van denHaag calls them, marshalled during the pastfifteen years or so in this country by those criticalof the death penalty. But in order to assess thesearguments more adequately, it is helpful to keepin mind two other considerations. First, most ofthe criminologists sceptical of (1) are led to thisattitude not by the route we have examined-theargument against (2)-but by a general theory ofthe causation of crimes of personal violence.Given their confidence in that theory, and theevidence for it, they tend not to credit seriouslythe idea that the death penalty deters (very much),much less the idea that it is a superior deterrent toa severe alternative such as "life" imprisonment(which may not deter very much, either).? Theinterested reader should consult in particularProfessor Marvin Wolfgang's monograph on thissubject. Second, very little of the empiricalresearch purporting to establish the presence orabsence of deterrent efficacy of a given punishmentis entirely reliable, because almost no effort hasbeen made to isolate the relevant variables.Surely, it is platitudinously true that some per-sons in some situations considering some crimescan be deterred from committing them by somepenalties. To go beyond this, however, and sup-plant these variables with a series of well-con-firmed functional hypotheses about the deterrent

28 See, for an excellent critique of a recent study indeterrence, Zimring and Hawkins, supra note 13, at111-14.

29 M. Wolfgang, PATTERNS or CRarIAL HoniDcmE(1958).

[Vol. 61

DETERRENCE AND THE DEATH PENALTY

effect of current legal sanctions is not possibletoday.'

Even if one cannot argue, as Van den Haagdoes, that there is no evidence against the claimthat the death penalty is a better deterrent thanlife imprisonment, this does not yet settle thereliability of the evidence. Van den Haag could,after all, give up his extreme initial position andretreat to the concession that although there isevidence against the superior deterrent efficacy ofthe death penalty, still, the evidence is not verygood, indeed, not good enough to make reasonablethe policy of abolishing the death penalty. Thereply, so far as there is one, short of furtherempirical studies (which undoubtedly are de-sirable), is twofold: the evidence against (i)-(vi) isuniformly confirmatory; and this evidence is inturn made intelligible by the chief current socio-logical theory of the causation of crimes of personalviolence. Finally, there do not seem to be any goodempirical reasons in favor of keeping the deathpenalty, as a deterrent or for any other reason, apoint to be amplified in the next section.

IV

Van den Haag rests considerable weight on theclaims that "the added severity of the deathpenalty adds to deterrence, or may do so;" andthat "the generalized threat of the death penaltymay be a deterrent, and the more so, the moregenerally applied." These claims are open tocriticism on at least three grounds.

First, as the modal auxiliaries signal, Van denHaag has not really committed himself to anyaffirmative empirical claim, but only to a truism.It is always logically possible, no matter whatthe evidence, that a given penalty which is exhypothesi more severe than an alternative, may bea better deterrent under some conditions notoften realized and be proven so by evidence notever detectable. For this reason, there is no pos-sible way to prove that Van den Haag's claimsare false, no possible preponderance of evidenceagainst his conclusions which must, logically, forcehim to give them up. One would have hoped thosewho believe in the deterrent superiority of thedeath penalty could, at this late date, offer their

30 For a general review, see Doleschal, The DeterrentEJec of Legal Punishment: A ReviB w of the Literature, 1INFoRmAo REviEw oN C~mn A" DELmQ. 1, 1-17(1969), and the many research studies cited therein,especially the survey by Morris and Zimring, supra note22, at 137-46.

critics something more persuasive than logicalpossibilities. As it is, Van den Haag's appeal topossible evidence comes perilously close to anargument from ignorance: the possible evidenceone might gather is used to offset the actualevidence that has been gathered.

Second, Van den Haag rightly regards his con-clusion above as merely an instance of the generalprinciple that, ceteris paribus, The Greater theSeverity the Greater the Deterrence, a 'plausible"idea, as he says. Yet the advantage on behalf ofthe death penalty produced by this principle is afunction entirely of the evidence for the principleitself. But no evidence at all is offered to makethis plausible principle into a confirmed hypothesisof contemporary criminological theory of specialrelevance to crimes of personal violence. Untilevidence concerning specific crimes, specificpenalties, and specific criminal populations isbrought forward to show that in general TheGreater the Severity the Greater the Deterrence,the risk of being stupified by the merely plausibleis run. Besides, without any evidence for thisprinciple there will be a complete standoff withthe abolitionist (who, of course, can play the samegame), because he has his own equally plausiblefirst principle: The Greater the Severity of Punish-ment the Greater the Brutality Provoked Through-out Society. When at last, exhausted and frus-trated by mere plausibilities, one once again turnsto study the evidence, he will find that the currentliterature on deterrence in criminology does notencourage a belief in Van den Haag's principle."

Third, Van den Haag has not given any reasonwhy, in the quest for deterrent efficacy, one shouldfasten, as he does, on the severity of the punish-ments in question, rather than, as Bentham longago counselled, on all the relevant factors, notablythe ease, speed, and reliability with which thepunishment can be inflicted. Van den Haag cannothope to convince anyone who has studied thematter that the death penalty and '"lie" im-prisonment differ only in their severity and that inall other respects affecting deterrent efficacy theyare equivalent; and if he believes this himself itwould be interesting to have seen his evidence forit. The only thing to be said in favor of fasteningexclusively upon the question of severity in theappraisal of punishments for their relative deter-rent efficacy is this: to augment the severity of apunishment usually imposes little if any added

31Se authorities cited notes 22 and 30 supra.

19701

HUGO ADAM BBDAU

direct cost to operate the penal system; it evenmay be cheaper. This is bound to please the harriedtaxpayer, and at the same time gratify the demandon government to "do something" about crime.Beyond that, emphasizing the severity of punish-ments as the main, or indeed the sole, variablerelevant to deterrent efficacy is unbelievablysuperficial.

V

Van den Haag's final point concerning wherethe burden of proof lies is based, he admits, onplaying off a certainty (the death of the personsexecuted) against a risk (that innocent persons,otherwise the would-be victims of those deterrableonly by the death penalty, would be killed).n Thisis not analogous, as he seems to think it is, withthe general nature of gambling, investment, andother risk-taking enterprises. In none of them isdeath deliberately inflicted, as it is, for instance,when carrot seedlings are weeded out to enablethose remaining to grow larger (a eugenic analogy,by the way, which might be more useful to Vanden Haag's purpose). In none, is it necessary tosacrifice a present loss in the hope of securing afuture net gain; there is only the risk of a loss inthat hope. Moreover, in gambling ventures onerecoups what he risked if he wins, whereas inexecutions society must lose something (the livesof persons executed) no matter if it loses or wins(the lives of innocents protected). Van den Haag'sattempt to locate the burden of proof by appealto principles of gambling is a failure.

Far more significantly, Van den Haag framesthe issue in such a way that the abolitionist hasno chance of discharging the burden of proof oncehe accepts it. For what evidence could be mar-shalled to prove what Van den Haag wants proved,that "the greater severity inherent in irrevo-cability [of the death penalty] ... adds nothingto deterrence"? The evidence alluded to at theend of section IV does tend to show that thisgeneralization (the negation of Van den Haag'sown principle) is indeed true, but it does not proveits unqualified validity. It must be concluded there-fore, that either Van den Haag is wrong in hisargument which shows the locus of burden ofproof to lie on the abolitionist, or one must accept

2The same objection has been previously raised inFeinberg, Review of the Death P enlty in America, 76ETHics 63 (1965).

less than proof in order to discharge this burden(in which case, the very argument Van den Haagadvances shows that the burden of proof now lieson those who would retain the death penalty).

"Burden of proof" in areas outside judicialprecincts, where evidentiary questions are at stake,tends to be a rhetorical phrase and nothing more.Anyone interested in the truth of a matter will notdefer gathering evidence pending a determinationof where the burden of proof lies. For those whodo think there is a question of burden of proof, asVan den Haag does, they should consider this:Advocacy of the death penalty is advocacy of arule of penal law which empowers the state todeliberately take human life and in general tothreaten the public with the taking of life. Ceterisparibus, one would think anyone favoring such arule would be ready to offer considerable evidencefor its necessity and efficacy. Surely, some showingof necessity, some evidentiary proof, is to be ex-pected to satisfy the sceptical. Exactly when and inwhat circumstances have the apologists for capitalpunishment offered evidence to support their con-tentions? Where is that evidence recorded for usto inspect, comparable to the evidence cited insection III against the superior deterrent efficacyof the death penalty? Van den Haag conspicuouslycited no such evidence, and so it is with all otherproponents of the death penalty. The insistancethat the burden of proof lies on abolitionists,therefore, is nothing but the rhetorical demand ofevery defender of the status quo who insists uponevidence from those who would effect change, whilereserving throughout the right to dictate criteriaand standards of proof and refusing to offer evi-dence for his own view.

The death penalty is a sufficiently momentousmatter and of sufficient controversy that theadmittedly imperfect evidence assembled over thepast generation by those friendly to abolitionshould by now be countered by evidence tending tosupport the opposite, retentionist, position. Itremains a somewhat sad curiosity that nothing ofthe sort has happened; no one has ever publishedresearch tending to show, however inconclusively,that the death penalty after all is a deterrent anda superior deterrent to 'life" imprisonment. Amongscholars at least, if not among legislators and otherpoliticians, the perennial appeal to burden of proofreally ought to give way to offering of proof bythose interested enough to argue the issue.

[Vol. 61