lOe Spanisb joori~il nl Psicliolog

2(11)2, \k,1 5, No. l 29

Ii

Co,yrigbi 21)1)2 b) 1 be Spjn sli bu ial nl Psyclioiogy 1138-7416

Nuil Sex Differences in General Intelligence: Evidence from the WAIS-IIJ

Roberto Colorn, Luis E Garca, Manuel Juan-Bspinosa~ and Francisco i. Abad
Autonomous University of Madrid

Tbere is an increasing number of siudies clairning Ihar ibe sex difibrences in general intelligence are real. TOe empircal evidence is based on lEe srimmalion of lOe siandardized sex differences o several cognitive balleries. 1-toweven dic sejentifie constroct of general tbility rests on dic correlaiions amono test icores, ratler iban on iheir surnmation. Tbe taller (ability o general) is an arbilrary variable, nol a sejentitie consnic. General ability is on a funetion of any particular cognitive tes, bu a source of variance evidenced by tOe correlation between severa! diverse tesis. cadi of wliicb redecs general abilily (g) lo sorne exen, bol alto group faclors and les spccficiiy. Because diere are imporiant educational econornic, and social cotisequences nf a group diflerence in general abiliiy, it is especially germane lo evaluate tOe possibilily of an average tez difference in iii proxy measures. soel as IQ. TOe Spanisb itandardization of lOe WAISIII is analyzed o dic presen sody. TOe sample was made o~, of 703 fernales and 666 males, aged 15-94. drawn as a represenlalive saniple of ilie population in teniis of educadonal level and geographical location. Altbougl a male advantage of 3.6 IQ points it observed, dic diffcrence is in ability in general, nol in general abilily (sfl. Given bat llie main ingredien of tOe srong astociation between IQ and a broad coge nf social correlates is g, and given thai ihere is no sex difcrence in g. lOen lOe average IQ scxdiderence favoring males nius be atiriboled to specific group facors and tes specifieity. Key ,,onlv: ge,wral inrlligeoec,sex diffe,vnc-e.s, copairi,e abilitis, psycioogieal assessrnent, practical alidirv

Un nmero creciente de estudios sostiene que existen diferencias entre los sexos en inteligencia general. Las pruebas empiricas se basan en la suma de tas diferencias estandarizadas entre los sexos en diversas bateras cognitivas. Sin embargo, el constructo cientfico de inteligencia general se basa en la correlacin entre las puntuaciones obtenidas en los tests, no en su suma. La suma de puntuaciones (inteligencia en general) constituye una variable arbitraria, no un constructo cientitico. La inteligencia general no es funcin de un determinado test, sino que constituye una fuente de varianza puesta de manifiesto por la correlacin entre diversos tests, cada uno de los cuales refleja inteligencia general (g), factores de grupo y especificidad del propio test. Puesto que existen importantes consecuencias educativas, econmicas y sociales dc las diferencias de grupo en inteligencia general, resulta especialmente pertinente valorar la posibilidad de que exista una diferencia promedio entre sexos en medidas come el Cl. En este estudio se emplea la adaptacin espaola del WAIS-lll. La muestra est formada por 703 mujeres y 666 varones de entre 15 y 94 aos de edad, representativa de la poblacin en nivel educativo y localizacin geogrfica. Aunque se obsen,a una ventaja promedio de los varones de 3.6 puntos de Cl, la diferencia se debe a la inteligencia en general, no a la inteligencia general (g). Dado que el principal ingrediente de la fuerte asociacin que existe entre el Cl y un amplio conjunto de correlatos sociales es g, y que no existe una diferencia segn el sexo en g, entonces la diferencia promedio de Cl que favorece a los varones debe atribuirse a los factores de grupo y a la especificidad de los tests. Palabras efave: inteligencia general, diferencias de sexo, aptitudes cognifivas, evaluacin psicolgica,
validez prctica

llie rcsearcb referred to o Ibis arride was sopported by a gran fonded by tOe .Spanisb Ministerio dc Educacin y Cultura, Gran N lB92-0066. Wc are g-atelul lo TEA, SA, mr providiog dic database analyzed in tbe presen slady, witl, special thanks to Dr Nicols Seisdedos, Correspondence coceroi ng ibis anide slould be addressed lo Roberto Colom, Facultad de Psicologa. Universiad Autnoma de Madrid. Ciudad lii ver aria de CanroNanco. 28(149 Madrid <Spain). Email: roberlocolon @ utirn.e5

29

3(i)

(it)l)M. (AR(lA, JUANLSIINOSA. ANt) AAl)

II s isual y stated il a sex cl1 erenees iR cenc-al i dell Cciiia alultv tic nonexsteiit (lrody. 1992; GoLon, [998: Halpera. 992 JuanEspinosa. 1997; Nel sser ci al.. 1996). 1 lowever. A nkev (1992. 995>. R usbion (1992). and 1 xnn (1994.. 1998. [999) noted a paradox: Males. en ilicaverace. bave laiger bruj os chao lema Les and bruj o sic is pos iiivel y i55 <e it Ccl 5V it b inielligence (lensen. 1998; Mackinesb. 1998; Resinen & Ankev, >996). llence, it woulcl be expected thai males wetild buye Li bglier average level o> intellnence Iban lemales. In un extensive revjew, Lvii o (1994) calen [atal a mean sex di flerence i n general 1uteligenee of 3.8 IQ poinis laveriog mal es, precise [y t be advaotace. thai cuj be preclcied irom males Lamer brans. Ibis predction 5 bseci oh a nican correLiioo OF .35 between o vivo brujo sic (measured liv fML<I, see Rosbioi & Ankcy. 1996) and IQ atid a sex djfference of .28 SI) iii adill brujo sic (utopsjerl br;ios). bence u preclictec male-lemale dilference u IQ of .35 7 .78 SD .27 SL) 7 15 4 IQ poinus. b~ve ycurs latee Lyon (1999. p. 10) sated 1bar males rIo huye fi jgher nican 1 Qs iban lema les by Lipproxniiiely 4 IQ 1)01 ffs, cenineiisurate wiib iheir laicer aveva 4e bra u s[se. Fbi s cune 1 u s ion bolds. xvbeiier general jote LII cence i s ceIii tec us ibe su iii of ibe ve ita 1 co ipre beni on. reso og. uod sputiul group factois. as fi cd intel Ii ceoce or leaseniog a[iiiuy. or us Spearmaii s y iicaswed from tbc fi st. liii itipLil coiiipoiicni (ir as tbe global IQ obia ned fion standard intelligenee und apij tude tests. so 1og us tbi 5 fuIii lIs ihe condjtjons stipulaied by Jensen. Lyon (1999) coosjdei-erL 20 frrtber dura seis oj Sex di ifereoces iliLil. coiTespoiicI xv ib tIte 1994 estimares for geocral inlelligence (Lyon. 1994). Tbe quesuon of wbether Of ini ibere is u ses dilferejice in general iniellgenee is especially germajie br psycliological assessmen e ib os, br 1 u stauce. clic pracleal val idi iy of measules of general Intel Iigeoce i s usual [y odicatec by u sIgoi fican predctve usebol coiTelarjoj xvib soiiie educational. economie. or social crilerjon. I-Iighly g-loaded tesc seores (IQ, see below) sbow a greaier unjversal prac cal val diiv tban any otber psycbologieal cotsti-oci. IQ tredicis perh.miaiice o every kind of bebavior ibar culIs br Icain ng. decision. unc jodenen. fue validjlv of IQ is un iOcieasiiig niojiOlOiijc functjon of ibe level of cogoitive Complexity lo tbe ejiterion. Tbe sialj sijeul removal of y Irom any psyeboiietric tesc or bat iery. Lea v ng on >} groop fziCi(Y5. proc> uces a nec> i ej ble practica 1 sal Idi 1 y xv bei tbe.y tic used jo a represe o ial 1ve populution (Gordon. 1997; (iotcfredsoo, 1986.1 99/u. 1997b). Ihe geiierul tacior (y) iban can be exiruereel trono a correlation munix beween several cogjiive tesis predics sebol usije acl 1 eveoenc. becau se y i 5 ntnj jisj e te 1 earii jng novel material, grasp jng coneepis, meantgs. unc ..<i fon fi. Fuillermore. y 5 ilie nain Cogitve correlate Liod bes single jirectctor of success iii ob trainiig Luid ob perlorinaiice. Nleia-unulyses o> buodneds of tesl val jdaiioo stiles buye sboxvn tbat ihe validjtv of a Iiighlv yIouded tes xvjrl preveil valicLiiy br a puniicolr ob jo puni~eular orgujtiLIIjoliLiI setiiog is geocrulizable to il oiber jok aid seliings (Ree & Far es. 1991; Scbni idi, IIujter, Cuierbridee. & Gol]? 988).

lictor js sijl nelleeted ji otlier line;.t<L seejal a s see jtI uctul ciii s 5 ijeL i Li5 ci op ji iii g on. <u sc]eol ebrojije xveilkre status. eljdneglec, poveriv, aceiclen
001 eoiii es.

[le

roieess. del inqucnev. or erjlie. Lliese relicieiisbps are reii jitcle;ieodeilv of sc>ejLil el;ss el onjeil. lliese siiejiI ccinrcLies buye in jnverse iiiojloioojc jelitioii ie IQ lo ilie [iopitlLliiciii.sliex<itie. Oil .I\LtiLiCe. 1W tillicis tbe 1ierceiiage cii ocelirrence iii Llie Lewest c1uLiriiLe ob ibe total clisirjLoijen of IQ is iii ibe bjnlcsi c1iiarije (Ilernsiejo & Murnax~ 1994; II t. 1995; Mackjnicis b. 1998; Neisser ci al.. 1996>. Tbe eclucuijona. ecooiijc. Luid social eosec1ticiices o> Li groel cijferejice jo 1<) avise fnont ixvo elfecis: (;) ibe stLItIsiic;iL cbLracterjstics of tite icnial corve. ajd (Ii) itie jj moni proli;ble iliresboul of ilie level of .iliiljiv nec1i.nec.l <ir centaij 5(iciiiL LittLhiiinienis. Wen xvii 0< )iiiiLil disrilbu jojis cii 1 Q Ive cli fferent i tcuos. LIIrbeugb 1 le en ves 1 Lirg.e lv ovenl;1i oiie Liiioiber. ; gi veo cutofi[o ni <iii che 1 Q seale eait make a xciv ajee dj blerejice beixveeo ibe pio~iori i mis o> tI <e 1 ower SeO!) 04 gleOf Li d t be Ji ybey seoning g\ <~~ ib; it liii le 1 oxv ilie eritolf 1)01 ni.. be fi iii ben tbe cjsiaice of ibe eoioffpoiiir (ren tIte mean of ilie beber seon ng grol p, iLe 1 uger xvill be tic greu~i clj blerejice bctweeji ib e jrc p <it j oit el ee It grou p t Li it Ial Ls ztboye or Lelow ib e cLhtof seene (ieiseii, 1980). Guiiiig scnres oit tIte LQ seLile thi.it bu II al crjt CLII Il eslo Ics resuf lo cfis liLil les behveeji (Lic rcpori 1< iiis of i Le Li jgl ev oc! 1 ower seorog gnoop s ch it Ial to cl jI mciii silCiLil and occupi 1 OiiLiI eLliegoles (1-lonier. 1 983. 1986; II onier & llonter. 1984; 1-1 oner & Schmdi. >990; MeHeory. HougLi.Tocjuai. 1-1.osoi, & Ashworth. 199(0. Cojsider a nican group dilference of 3.8 IQ poiits. Assume titar udmsSjon to a bigll y selective iral niog course 5 lasec oj Li cuiiiog LQ seone of [20. WLai percenrage oleacl grouii ial Ls huye dic cotoff secre? (See Fjg Ore 1 .) Fcmn (le gnoup xviLi jiteaj IQ of [00, tite ccmnespondiiig 2 secre is 1.33. TLie Ltreu of ilie oorjii il c tose luliii e aboye 1.332 is 9. 1 8 <4< For be giro i xv di a OcLlit 1 Q of 96.2. aji ~ cii 1 20 s eq Lii xlent ti) a 2 scctne of 1 .59. [te Lnea of ilie onnd curve fulji g iliow 1 .59; is 5.59%. Tterefcne. uj excess of ijijircix1 uiiaiely 3% of le liig Len se(iri uy yre ip lId Ls iboye i Le ci.bol selii-e.

(30

100

120

Inteiligeoce Quot<erit

Figure 1. Two venial d sjbuticijis:

itse epcseis two cote seotes al

lOO oid Nl

96.

Fe Ii cuje
=

([(3

80) <<iii X (1(i)

1 21)).

N ri

L MEX 1)1 EFL?RLiNCES IN (EN ERAL INTILL.LL(ENGE

31

It sitoijd be jiotec ilt sioc es of ses cjilerctees iii general ubi 1 icy Lvc lcej eoofociiicLecl ly ipiojier defi titios ajic iiieasciremeois cii cejerul iLjljiv bsecL oj dic sinple SI/ion Idi fl o 1 S o liies seojes liii iii severa1 L)Lit i cii cs ch .i i cli len it ilcir greof iaciors (fon exuii~Lc, see repciris liy L.yiii: II uiicri & Luto, 997 lyon. 1994. 1998. 999>. Tbe anaLyscs yle 1(1 a mean sex di flerenee o ibe total sccre. bol sucl resulis are arLiitrury, el liiiiii2c geieuaLiiy, unc el Little seientj lic (unc practical) ijrces (Jeosen. [998). Scme iceen anal yses huye puioiecl ooi itegligible sex cjifereoce o general intel Li ge ncc rIetiucd as g afrer u bioud sari city of cogo i rive ujc sehel as e larieries <Al uj a, Ccloo. A buc, & Joaibspinos. 2000: CoLom, JoaoEspiiiosa. Aliad, & Garca, 2000; cusco. 1998). A Iclciugl ihe titeaj siuncLird sed sex difierertee was ji stione Liizreeuiieni wiib ibe che nepoi~ied by Lyon (1994, 1998, 1999: Llatiori & Lyio. 1997). ibe dilleretice xvas uuc artuiburable Lo y. It musc he eui~iliasized bat ihe si ople so ni of van oos.s ubrcsr scores is o 1 oo seicorifie un puLictical nceiest. becacise it eauuut be ecinsidened Li proper uicisure uf gejerul bility. Tbc conecpt of general aliility, delincct <5 ,g. rescs on che ecnnelatois Litnciitg <es secires raiber ih ciii ibeir sciiot. [Le luirer (uliiLicy ir geerl) is un arbjiiarv viriabLe. ci sejejitilie cosircer. The empirical fact thai al 1 mental dii 1lies are pcisitively ecrrel ared cu Ls fon un nial vi e iuxeooiiy cii cocal abi [lijes based cii sonie lcr of eorielatj oit iou)ysi s. be di itensiujs leunc iii ibe lieior Liiiulvsjs ol ilie eueLuroos uiiong a vLtnicty oi meiiul abiliy neasureiertts co be anangec bierajebicalls aeecrdioy tu ibeir gcoeraliy (Gand. >993. 1997). Tbe y factor is Ilie osr general of al 1 od is coiuiot te il mental abilcies. lbe y factor is a conuco seurce of individual diffcretces itt ah r.agnhioe tesis. The knowicdge and skiLIs cappcd by res perforniance merely prcvide a vebicle fon tite measoremeor of y (Jeosco, 1992). Nol every veljele is a lijie uieasure of ibe cuosruc. [le consiroer nost be elicired o uy difieren 1 wuys. It s orir u fo oc co oi aoy puni cular vebce, bu a suorCe of variLnee evidenced by (Le Corielaijon betxveet severul diverse wsis. cad of xvheL refleci.s y te sume exict, bur uy ulso jeflec groui IaClors itic les speebciiy. Nc pone tesc of y exisis, Ilie sclutioo is ro obrair a conposiie secire ircuin severa 1 bigh [y clverse y loidec tescs. [~licgrcuten dic ncoiilieis of tcsts tliit eruten tc che coixipusite seone, che mone i be un yait ecl siio ces of van unce une avegcd riel, lii tite Liest-scaildLicbed test batteries 75 <~ or itiore cii clic van toce iii ilie cctm[iosj te sectres cciosi sis 01 y. Ibis is typieal br mest idi y idea1 IQ tesis (scel as che WeelsLen). Eacb tes serme refleccs ibe [cccl of y unc Ihe progenies of ihe tes self (Gaud. 1978: Acusen, 1998), lo shcirt, Ley questuiit iii tite rcseircli o coyiiitixe sex di lere jice5 5 xvii etht r, 00 .1 verLige. le ttt les Luid iiial es cli [lcr iii y. TLis c 1ocsiiut 5 iecLiiicllx tite iiirisi clii fucol te Liitswer unc Itas lidej tIte [CLisjiivesijcaiecL. Ibis aiiicLe exLijoes sex cli Ifereoces iii te iii s cii u 1 eg lud ir ex tiactec Irc iii che 8p~oi 5 Li siLoicILodi sai hu cml fe WA8111. Wc ovesi1 eieci ~v lcben ttere is aiy sex cli llezeitee iii geeL intel 1 igeitee cLebijecL rs y.

Meiliod
Piiic pu; <Ir u,d Aleas, res

Ube Weebs[er Act It Intel ligeoce Sea Le (WA 18-111: Wcchs lcr. [997) wLts scuoclird sed jo Spain iii [998. Tbe stuoclarclia on sap Le ecns srecl of 703 feuaL es aocI 666 nico. agec 15-94 (N<6 N>~ [63; N,024~ = 53 N 408 NM <,~ y. = 237; 71 ev 1 36). drux o as a ir preseo lat ve su ni pe oi Ihe 1iopci latiou o forus cf uclucalicuu tI level (ueucleiie level acut 300: acurLernie Level emir III) 499 md ueacleuiijc Level two[ocation 524: 1 bice geecnaphcal aod acurlec (NN = 348. [ese N~ ~ 299. M~~< 1 359. N~,0<< 363). Ube SpaoisLi siudarcLacot of ihe WA 18111 ijelcdes [4 xveLIkuoxvo suihiesis: voeubt[ary. sinijlarities. uriiiiielc, di gi r span. inforuuiiu. compreheiisioui. leiterm u uiben series. fmCI ore con p[eu co, cccli ny, b ecL cLesign. hiurriCes, pic une ariunge iticur, sy nnbci 1 seancL. oid ubjecc asse bi y.
ji

U ve,,,, 272: N2~54

jets yetmns
=

Aujolv.ses TLc ecocl rif coreLaicc veei.ons is espeeialLy tppndt[inute fuir couijiarog Ibe veClors deftcr by the y Icaclijes of a varieiv ci tesis ajd tite staiicLardieci ncuu gncup differenees (1) jo ilcse tests (Colon el U.: 2000, Acosen. 1998). Phis echocl mus coniply xvith severul ecuditicos (see Jensen lcr uorc detuils): Tbe sunples mus be large md rcprcseuittive, ilie nt,uber of tesis nalyzed nius be large cnougb. ibe tesis nust be diverse, the tesis relabi uy eoeffjcenis nius be tken lote acecun, and ihe values correspondiog ce ihe ceogruence coeffcieocs among che fucicius of ioenest ebtined from ilie greujs conpured sheuld be higher Iban .90. Ihe ecogruenee cceffieeu( ~ is un mdcx of fuccr sni[aniy. lf ihe groups sboxv ideociy of che y factor, conbi iii ng ibe cxvo veeicrs cao mercase che cli ubj Li ly rif che vectrn rif y cadiugs (Jejsej, [998). The siacist cal iest uf ilie b y pcibcsi s conecnuilng mean grou[i clifieretces is ftc correlution beixceen ihe vector of time tcsts y foucings aod dic vector cf stundardizcd ocian cli ffecoces bctwcco che gruups co cad rif lite icsts (d). tukiitg clic tests reti atu ticy cueftieieuts jote aceutxt. Ube o ic lucid lun tesirte ile bypotlcss depeuds un ilie uuuagoiludes uf clic gnotup riulicreoce aevruss tescs char dilfer o cheiv y1oaci ogs. [be Speanuau nao 1<orden ecrreLati 00 (rs) of (be cohim vector of sublesis y Ioadigs witl <le vector cf ihe sex di fieneoces (d) en ihe s obiesis oclicuies ilie deg ree tui xvl cli y is re aiccl lo dic aoL cincien of dic sex cli ifenejces co ihie vunidcis stbtesrs. Li ihe correLuijon 5 jc stlisicuLLv si giifietuur. ibeo ilie stuzclrcli ecl sex di fieneoces !) are 001 el atec tu general o te LI igenee clefi nec as y. Wc percirmec Iienrcluietl Iiccur aiiuLysis (Sebmjcl Le jiuL ji i i ii 5 Ionituat 1cmii) seliLtia te Iv nr mu Les tui ci feutaLes, lu le Sc hu clLel iii ruaj sfu ruiatj u mu (Se u iii cl & Lel. 1 957). Lic Ii Li dr riuclen laetui is Lije Li II oxvecl to uccii Liii 1 ion LiS u itLic Ii

32 Table 1

GOLOM. GARCA. JUAN-ESPINOSA. AND ABAD

Descripiie Data aoci ScnjdcI,-di2cd Ma, Di/fr reocs (ci)

WAIS-lLL subiests
M 1. \ocbuluny 38.12

Males
SD

Fcm ales

N 665 666 666 titO 666 666 665 665 661 662 666 665 661 664

Al 36.28 16.65 it,19

14.16

SD

N 702 703 702 703 703 703 702 702 702 701 703 701 702 702

ci

2. Siunilaniijes 3. AnitLoetic 4. Digii Spuo 5. Infonmation 6. Cooiprchen.siuu 7. Leitcr-Nuohec 8, Pienune Ceutipiecicn 9. Codiuug 10, I3Lock tiesigo II. Macnices 12. Pieture Annangemen [3. Syuiboi Scuneb 14. Objeet Asscmbly

17.40 13,37 [5.46 17.32 18,54 9.88 18.12 64.58 40.17 16,33 12.93 29.75 30.48

[3.44 6.93 3.92 4,75 5.88 6.17 3.58 5.17 24,94 [4.68 6.28 3.89 11,86 10.64

4.00 6.62 3.62

4,44

14.85 17.56 9.01 17,48 60.26 35.23 14.61 [.80 27,44 28.82

5.99
6.11

3,51 5,35 27.11 14.47 6.40 5.81 12.06 10.42

.13 .11 .58 .28 .42 .16 .25 .12 .17 .34 .27 .19 .19 .16

Nola Thc ci values were compuced by dividiog che nican dilference betwcco Ibe gnoups uy uheir peoled wthio-gnoup standard deviation,

uf time coneaiion <meng time observed variables as pessible, whereas ihe Lewev order faccers are neduced te residual faccors uncorrelated cucher ce eueh echen or te che Ligier order faetons. Thenefore, caeh factor vepresencs ihe iudependent cetitribucion cf the factor o questien (Carrolt 1993; Lueho, 1992), Besides Ihe method uf ecrveLaced veetcrs, stiLl unocher meched fon examining che sex dfference iii psychemetric g is ce repvesenc che sex difference en each elche stubtests tu terms cf a peiuit-biseniaL cerrelacien ameng subtests seeres and che sex variable, and include chese cerrelaticmus wiihin ihe fel uuacrix cml subtesc intercernelaciens fon lacten auaLysis.

The resul wil revea] clic factor Joading of ses un cad of ihe faecens thai emerges from time aoulysis, including g. The factor lcading cf sex is equivalen le time poin-bisenial cetrelacien between g unc che sex variable.

Results The desetiptive dala are sbown lo Table 1. Time standandized sex differeices (rl) used fer subsequeoc scudy of che nelationslips xviim che g leadiogs are alse shown,

Table 2 Correlution Mairix of ihze WAIS-IIl Subiess (Ma/e Corrc/aions al dic Top Ha1J Lenta/e Carrelarjaus al dic Bcuton I-Ialfl. Retabullies along iba Diagona[ Subuesis 1. Vocabutany 2. Sinuilariics 3. Aniihmetje
4. Digl Spao 5. Infornacion

10

II

12

13

[4 .547 .557 .583 .568 .525 .507 .601 .636 .631 .746 691 .696 .676 .68

6, Comprehetsion 7, Leccen-Nunber 8. Picture cernmleiioo 9. Codiog 10. l3Iock desigo

II. Matrices

12. Picture arraogc,tieot [3. Symbot seareh 14. ciNcel asseoubLy

.95 .747 .768 .89 .601 .604 .581 .596 .731 .704 .719 .703 .643 .624 .624 .617 .622 .604 .595 .611 .667 .692 .626.622 .592 .576 .581) .565

.643.536 .619 .541 .88 .582 .637 .89 .632 .567 .546 .520 .680 .764 .558 .579 .581 .592 .596 .594 .692 .640 .600 .574 .565 .536 .505 .515

.710 .741 .670.701 .668 .580 .505 .486 .93 .653 .698 .85 .619 .573 .615 .562 .552 .496 .597 .496 627 563 .639 .573 .548.490 .556 .489

.611 .613,582 .607 .599 .633 .555 .604 .665 .61)7 .579 .635 .753 .506 .582 .526 .579 593.503 .615 .565 .553 .507 .498 .91 .582 .669.624 .653.91 .624 .683 .705 .659 .82 .648 .681 .686 .684 .94 .711 .700 .751) .776 .663 .674 .659 .671 .683 .617 .807 668 .593 .616 .618 .743

.653 .608 .586 .654 .635 .565 .703.584 .608 .569 .565 .584 .634 .596 .511 .572 .583.497 .661 .669 .663 .728 .687 .636 .657 .680 .761 .755 .713.678 .94 757 .727 .754 86 .672 .708 .666 .77 .697 .660 .625

NULL SEN DLFEEL{ENCES IN GENERAL INTELLIGENCE

33

PabLe 3 Male atd Perna/e g Factor Loadings Exiracied afYer a Hierarchical Factor Analysis (Sehrnid-Leirnan Tran.sforrnaiion); Average g Lnadings are a/so Prese,ted
g Loadings WAIS-111 subiesis

Avenage g Loadings Males Females .77 .76 75 .78 75 69 86 75 .78 76 .83 .76 .75 .69 58.5 .77 .76 .76 .76 .74 .70 .84 .74 77 .76 82 78 .76 .71 .77 .76 .76 .74 .73 .71 .83 .74 .77 .76 81 79 .77 .73 5786

1. Voeabulary 2. SimiLanities 3. Ancbmetic

4. Digic Spao 5. L,formatjon

6. Compreheosion 7. Letien-Numben
8. 9. 10. 11. Picuore cenpLetiun Ccding I3loek desigo Matrices

12. Piciune arrangemen [3. Synbol scanch [4. Objcci assembiy

% Variance

TabLe 2 presencs the correlacien matrix, separaceLy for males aod females. Subtests reliabilities are also loctuded. A Sehmid-Leima llerarchical factor aoayss was cempiuced separatcLy fer females aod males. The g factor was represented by che higler erder factor. Pable 3 shows the g Leadings fer males and females. The cengriucrce coelficien competed frem dic g factor leadings o TabLe 3 lev males aod feunales was .999. Hece, che g factor is the sume, irrespeccive of sex, se ihe average g loadings can be computed te appLy the meihod of cerrelaced veeters. The Spcaroan rank-order correlacion becweeo time vector
.6

cf g Loadings and ihe vector elche siandardized sex difierences (rl) was r5 = .059 (p = .840), a value suggesting a nul sex differerce in g (Pearsen r = .008, p = .979, partiaL ceirelation coolrolLing for y = .0981, p = .750). Figure 2 slews (he scaUer diagram ccrrespoding te dic cerrelatien between che g md ci veccors. Tlis resuLts ti che failure te rejeet the nuIl hypothesis of no sex differeoces lo geocral intelligence defined as g. The poinc-biserial correlatiens becween sez and the WAIS-III subcests were also cemputed. These correlacicos were included withi che ful] matrix of subtests correlacieos for factor aoaLysis. Time restulting g loadirug of sez was .159.

o ello Anithrr

5 Information

4 4

Alad Design
o

3Compnehension Object Assembly

Digit Span o Symbo> Search o Piciure Cenuipleilon o

70 72 74 o Picture Arrangement o Coding oVecabutary 76

Matoes o

o Leifor-Number

.2 -

o d 1 68
g ieadings

o Similauies .76

80

82

84

86

Figure 2. A seatier diagran of ihe eorrclaciou of.840; ihe siandardized sex differences wiih cojelation che WAIS-l11 suhiescs fon plocced as a fuoction of 59,p = Pearson r = .008, p = .979; (U) Pantial eencnoLling subcescs neLiabilicies ihe subesis = g leadiugs. Speannaut r = .O = O98l.p 750

34 1) seos sien

COLOM. GARCA. JUAN-ESPINOSA, ANt) ABAD 1ocs faveri ng niales, ooi sri fur frco ihe 3.8 1 Q pciis 1 Q pci reponed by Lyuun (1994, 1999). Me uuipontant i55LiC 5 tliit ihe mechen of ceirelutec veetois ccotradiccs clic ermoelusiej thai eciulc be cienived fre m clic si upLe se un uu ion rif ibe staodarciLecl nican gnoep di fcreoces (ci). Because of ilie greacer seleotifie adequaey umf ihe ncihod of cenreluted vectcns te iest che o ol hy~mothesi s ceocennng sex differeuces lo general intelligenee defined us g, we cao crine Lude thai ihere is no sex d ifievenee o general
ji te llige nec.

Jenseo (1998) Lip[iled Ihe niciboc of ccmrreLuced veetcrs fon time cumpuvison of che g-Loadiogs o several cegnitive batcenies (W1SC-R, WAIS, GAPB, ASVAB. od ibe Bnjiisb Abilicy Seales, BAS) wiih ihe siandardized sex diffeieice en che seales iuclcded iii chuse baiteries. Plie coielaoius fecod by Jeosen (1998, p. 539) xveve .346. .036,024, .127, aod .103, vespeetively. He alse competed che g Loadiog uf sex in chose cognitive batieties, cbtaioing che values .094, .006, .255, .180, unc .001, respecciveLy. Jeoseos (;m. 540) mato coneluslon wasSThe mechod of conrelated vecicns slews thai in no case is chere a cornelation beiween subcesi.s g-loadings od che mean sex dffetenees un che varices sebtests 1...] ibe g-Lcudings of che sex diffencoces uve al quite small Cccm ec al. (2000) feuod a negligible sex dilfevenee o g alter che Lavgesc sumpLe en which a sex dfference iii g hud ever been tested (N = [0,475). Tbe Pearson r of che celurno vector of stibtescs g leadiogs with ihe vector cf che sex diffevenccs (~ un time subiesis was .122 (p = .721). Wich che veetov of veLabiLity coelfucicnts purcialled ouc, che g aod ci vectors were eciTelaced .051. Plie Speuvman rank evder cerrelation wus .000 (ji = .999). Ferchernicre, time g loadiuig

TLe mill sex diftreoce lu g sciggesis ihac: (a) Ube factor

(g) thai is pneseui lu uucnLy ul measures (uf cegtilive abiLty (unc cht accoc itis frmr uuicie that hall. uf Ilie tocl conmino

of sex was .216. Therefeve, ibeir findings ate eotireLy

eensisient with chose tusing quite diffeteoi batienies uod sebjeec samples. Aluja et al. (2000) feetd un average g-Ioding of sex cf 179 using cwe samples of 670 aod 887 young adcleseecs.

Thus, che value wus censiscen wicli previces fiodiogs.

Considering al che availuble empinical evidence, incuding thac we neport in che presenc stedy after ihe Spansh standardizacion sumple of ihe WAIS-ilI, che average coruclution betweeui g and ci is .09, whereas che uvenage gIoading of che sex variable is .02. Thereforc, it is clear tiat ibe scuodurdized sex diffevenee iii typieal IQ tesis cuuooc be altributed ce general ittelligence defined as g (Aleja et al., 2000, Colom cit <1., 2000. Jensen, [998). It is excremely impcrtaot te gathev cemulutive evidenee mcm diffetenc buiteries and subjcec samples because, as Carroll (1997, p. 31) has scated, 4 [...] is Likely ce be presetl, o some degvee, jo nearLy al meuseres of cegoitive ability. Ferihermere, it is an impercaoc factor, because en che average, oven many .studies cf cogoitive ubility tesis, it is fcund te censtitute mere chao bali uf ihe cecal cemoo factor variance in a tesiS New pieces of evidence oust be considered, becuese chere is un increasing unumber of .scudies claiming thai che sex difference in general inceLligenee is real (Aikey, [992, 1995; Lyno, 1994. [998. [999: Reshton. [992). The presen scudy shows cbac ihe scuppnsed sex cliffercoee is a differenee o intelhigeoce in general, bu nec o general inteLligence (Aleja et al., 2000; Colem ci al., 2000 Jenseo. 998). The mean ci thac can be calculated from ihe Table 1 of che presen study is 0.241, Phis value truoslates jote 3.6

factor variajce o a tes) dres uucmc diffev between sexes; (b) uuen-g factors und/or tes specifiecy une respensible len tie ubserved ecguiltive sex difejenees; and (e) ube paiadox-.the fiodings of urgen nale brujo, time usscciation cmi brujo sise wjilu IQ. and clic absenee era sex differeuice lu ovevail IQis noc relevan ce che uob]em of wheiher un uiot is diete a sex differcnce lo generL oceL 1 igeuce, beeaese ilere ir lo sex diffe-e;tce Hz genera/ itehhigence. Phe pructical importunee of ihe tiul scx dilfereoce in g s clireciLy related te its social correlaces. A di fference beisveen ibe uncuos uf cwe po~melacien gnoeps has a quite differeoc kiod of eciusequcoce iban dees ihe vely sume sise dfference wheti ebaiued becween twe iodividuuls en dic sume seale. Fon groeps, the niost importun ceoseqeenee of a group diffenenee lu nicuos is of a siucistical natene, Pbe cuoseqeenees of populatien dflercuices in IQ aje of greuten inpertajce than are mcst ocher measurable eharucienisties thai show comparable pupulation dilfereoces (eeusiderog al time impoitaoc sociul coyreLaces cf IQ). Because che perceocage ci individeaLs xvho fal o a giveo SI) ruoge deereases se napidLy as une nieves away freun ibe nican und iowand ejiher cuil of che nenumal cli sin bution, it beecmes obvees clat che pcmputirmos aje dis~rcpcrtoiateIy represeuut.ecL jo ihe epper utd 1 ewer tui [5 (Fi gere 1). Pci t be e xteut tliac chere are cli ficrec seleetion thneshcLds fon che level of IQ requited ion cercain leveis of educucienal attannen. un fon ado issioo loto col Leges, nicceputicmos. rin special sed irai ning prognanus, pnpo[atico greups thai differ o mean

IQ will be represented eneqeuLLy o ibe selection ouicene.

Phis is a dineec conseqeenee uf ibe coneluileo betweeo

IQ od ihese scmcial Lv si enificunt variables xvichi o each

popelacion. 61 vej thai clic iiial o ogreccnt of ilie u s snciui ion heisveen IQ aod chese suejal cuirelates is g, atd given tlat iheve is no sex difference in g, it mesc be cencleded thai che uvenage IQ sex djffereoee is aitt-jbutubLe (by defaeit) tu group faecnrs <od/en te tes speeifici ty. Pbereferc. che funetionul dilferenee betweeo che sexes o che real seitines where g is feoetiuning uiest be expected te be negligible. Phis eviclejuce nzusl be ectsideved lo tbe praetcal assesstnetit of intelligence.

NULL SEX DIFEERENCES IN GENERAL INTELLIGENCE References Aleja, A., Colun, R., Abad. Fi.. & Juan-Espinosa. M. (2000). Ses diffenenccs jo general lonelligenee defined as g anoog young adoleseencs. Personalirv aoci huduid,al Diffrrences, 28. 813-820. Ankey, C. (1992). Sex differeoces lo nelative bralo sise: Phc msueasure of womau, toe? Intelligence, /6.329-336. Akey, C. ([995). Sex differcnces lo brain sise aod nencal abilicies: Cumencs ciii R. Lyon aod D. Kimura. Fersonality aoci Individual Differenees, 18, 423-424. Brody, N. (1992). Inie/ligence ( 2d ed.). Sao Diego, CA: Acadcmic Press. Cannel. J.B. ([993). Human cagoUve abilifies. A surcey offactor aua/ycic siudies. Cambnldge, MA: Cambnidge Univensity Press. Carnoil, iB. (1997). Psycbemecnies. lotelligence, and publie percepilun. Inie//igence, 24, 25-52. Cacceil, R,B, (1978). The seienlc use of factor ana/ysis. New York: Pleneun. Colom, R. (1998). Psicologa de las diferencias individuales. Teora y prctica. Madrid: Pirmide. Colom, R., Jean-Espinosa, M., Abad, EJ.. & Garca, LE (2000). Negligible sex diffenences lo general loteltigenee. Inielligence, 28, 57-68. Gordon, RA. (1997). Eveyday life as an incelligenee tes: Effeecs of inielligenee and inielligenee concex. Inel/igence, 24, 203320.

35

Renten, A,, & Scbmidi, E. (1990). Methods cf meta-anaiysis: Correecing error and blas lo nesearcb funding. Newbery Park. CA: Sage. Jeosen, A. (1980). Bias o uental iesuing. London: Mecheen. Jensen, A. (1992). Vebieles of g. Psvcho/ogical Scienee,3, 275-277. Jensen, A. (1998). TIte gfactor London: Praeger Jean-Espinosa, M. ([997). Ceografa de la inreligencia humana, Madrid: Pirmide. Loeblin, J.C. (1992). Laten variable inocics: An inroduction to factor, path, aoci siructural analvsis (2nd cd.). Hilsdale, Ni: Erbaun. Lyon, R. (1994). Sex differences o lotelLigenee aod brain sise: A paradox resolved. Persona/uy aoci Individual Differences, 17,
257-271.

Lyon, R. (1998). Sex differences lo incelligence: Data from a Scotiisb sandardizalion of che WAIS-R. Persona/uy aoci Individual DJiferences, 24, 289-290. Lyon, R. (1999). Sex differeoces jo intelligeoce and brain sise: A developmendal theury. Intelligenee, 27, 1-12. Mackintosh, N.J. (1998). IQ and huiran inte/ligence. Oxford, UK:
Oxford Univensity Press.

Gectfredson, L. (Ed.) (1986). The g factor jo employmeot LSpeeiaL lssue]. Jaurnal of Vocational Behavior, 29 (3). Goicfredson, L. (1997a). Why g natiers: The complexiiy of everyday [fe. Intelligence, 24, 79-132. Gottfredson, L. (1997b). Foneword lo Intelligeoce aod Social PoIicy. Inelligence, 24, [-[2. Halpen, D. ([992). Sa dfferenees in cognhtive abilities. llilisdale, NJ: Erlbaem. Haden, 1<., & Lyon, R. (1997). Male-femaLe differences un ihe Japanese WAIS-R, Personality and Individual Differences, 23,
53 1-533.

Henrosrein, R., & Merray, Ch. (1994). 77w beIl curve. Inelligeuce and class siruciure u American /zfe. New York: Ence Press. Hen, E. (1995). Will be we so uarr enough? A cogo tice analysis of ihe crnning workforce. New York: Russell Sage Foendacion. Renten, JE. (1983). Overview of validitv generalization for tIte US. Emploirne; Serrice (USES Tes Researeh Repon N~ 43). Washington, DC: US. Depantuneor of Labor, EmpLoyment and Traioiog Administration. Hunden, JE. (1986). Cogoitive abiLicy, cognitive aptitudes, job knowLedge. and job performance. Joarnal of Vocalional Behavior, 29, 340-362. Hunien, JE., & Honren, RE (1984). Validiy and uiiLity of alcemative pnedietons of job perfurmance. P.ryehologieal Ru/lerin, 96, 7298.

McHenny, J.J., Hough, L.M., Poquan, J.L.. Hansen, MA., & Ashwunih, 5. (1990). Projecc A validity reselts: The relatiooship becween pnedicior and cnitenien demahos. Personnel Psychology 43, 335-354. Neissen, U.. Boedeo, G., Beuehard, 11, Boykuo, A., Bredy, N.. Ceci, 5., HaLpeno, D.. Loellin, J., Penloff, R., Sternberg, R., & Urbina, 5. (1996). hnielligence: Knowns md uukowns. American Psychologis4 51, 77-101. Ree, Mi., & EarLes, JA. (1991). Predicling cnaining success: No much mure Ihan g. Personnel Psychology, 44, 321-332. Rtishlon, iP. (1992). Cranial capacity relaced lo sex, nank. and nace un a siracified sample of 6,325 US. military personnel. Intelligence, 16. 401-414. Reshion, iP & Aokey, C.D. (1996). Brain sise aud cognitive abiLity: Connelaileos wicb age, sex, social cLass, and race. Psychonomic Bulleilo aoci Review, 3, 2 1-36. Sehmidt, PL., Hender, lE.. Oecerbridge, AN., & Goff, 5. (1988). bloc relation of expenienee and abticy with job penfonmance: Pes of ibree hypotheses. Jaurnal of App/ied Psychology, 73,
46-57.

Selunuid, 1., & Leinan, J.M. (1957). Tle developunen of hierarchical factor solecleos. Psyclzometrika. 22, 53-61. Wechsler, D. (1997). Wechsler Adul Intelligence Scale (3ud ed.). Sao Antonio. TX: The Psyehologicai Corpuration. (Spanish adaptation: WAiS-iiI: Escala WechsLen para adultos. Madrid: TEA, 1998). Received September 8, 2000 Revision received January 1, 2001 Accepted February 2, 2001