Journal of Personality Assessment

ISSN: 0022-3891 (Print) 1532-7752 (Online) Journal homepage: http://www.tandfonline.com/loi/hjpa20

The Derivation of Gender-Role Scales GM and GF for MMPI-2 and Their Relationship to Scale 5 (Mf) Cynthia D. Peterson & W. Grant Dahlstrom To cite this article: Cynthia D. Peterson & W. Grant Dahlstrom (1992) The Derivation of GenderRole Scales GM and GF for MMPI-2 and Their Relationship to Scale 5 (Mf), Journal of Personality Assessment, 59:3, 486-499, DOI: 10.1207/s15327752jpa5903_5 To link to this article: http://dx.doi.org/10.1207/s15327752jpa5903_5

Published online: 10 Jun 2010.

Submit your article to this journal

Article views: 2

View related articles

Citing articles: 9 View citing articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=hjpa20 Download by: [Deakin University Library]

Date: 07 November 2015, At: 03:41

JOLXNAL OF PERSONALITY ASSESSMENT, 1992, 596.33, 486-499 Copprigiit lt 1993, Lawrence Erlbanm Associates, h c .

Downloaded by [Deakin University Library] at 03:41 07 November 2015

The Derivation of Gender-R GM and GP 90 elationsh Cynthia D. Peterson johns Hepkirss L h i z m i t y 5 c h d of Medicim

Separare scales for mascuiine and demini~~e gender roies (GM arid GF, respectively) were de\~elopedfor the Minnesota Muitiphasic Persosaiiry inventorv-Z (MMPI-2) based on the iten endorsements of men and women in the restandardiza:ion sampte. Each scale recects the pattern of answer. of e majority of ?he members of -, the respec:ioe sexes. i here are no i:ems ir, common Serween the twc scales, and they correiate - .I3 with each other for both men and w a n e s . Distributionai, temporal s:abili:y, and internal consistency charac:erisdcs were ana!vzed, as well as their itern overiap, and correlatiorrs with the basic prof& scales. These separa:e unipoiar scaies were con:ras:ed with Scaie 5 (the Mascu!inirv-Ferninin~ry scale, MfJ, the traditional measure of these constracts in :he Minnesota Mukiphasic Personality Inventory (MMPI). The conjoin:- use oi SM and GF to forrr, gender roie groups Is, recommended to suppienen: and clarify );he azbiguitv of midlevel scores o n Scale 5.

In spite of che passage of nearly 20 years. mascukinity and iernininity cocdme to be, in ConstantinopEe's iI973) succinct appraisel, among the 'hnuddiest concepts in the psychoiogist's vocabulary" (p.39C). One oithc pwsistenr diEiiiculties Lies in the conceptualization of these attributes: Are they p o l z opposites aiong a single c~ntirruum,or are they n o r e accura~eiyportrayed as separate coexisting traits o r syndromes? Far example, the origins of the basic measure of these concepts in the Minnesota h';uXtiphasi: Personality Inventory (MMPQ, Scale 5 (the Ma~culi~~t:;-F;em:ninity scale?M f i H a t h a u y ?1956),are rooted in early research e n the characreristics studied by Terrnar: and Miies (1936). These investigators posed them as opposite ends of a single contimum: ranging from extreme

Downloaded by [Deakin University Library] at 03:41 07 November 2015

GM,GF

.4ND SCALE 5

487

femininity, threugh neither clearly feminme nor rnasculirre, to extreme masculinity. Their .4ttitude-Interest .Analysis Test (Terman & Miles, 13381 was z model for Scaie 5. A similar view of masculinity and iernininity was incorporated in:o the -n.original versions of the Strong Vocational Inrerest FjEank (S!,7B; Strong, ir-t,O: the masculinits-feminini:~ key for the men's farm and the femini~itymascu!initv key ior the women's form. (Camphell, 1966, omitted these scaies in the later versions of the SVIB.) Gough (1959, 1388) included 3 similzr bipolar scale, Femininity (Fe). in his California Psychological Icventors i@FI!. Bern (1974) broke from this conceptualization of mascuiinirv and fenininks by introducing separate scales for each, but she ased the same norion of a single underlying con:inuum in her recommendations to subrract. one score from the other to determine a subject's gender-roIe tvpe. The early research on which these bipolar scales were based characterized masculinity and femininity as patterns or consteliatlons of personalicy features: masculinity as instrumental behaviors, such as self-confidence, independence, leadership, and assertiveness: and femininity as expressi\-e behaviors, such as empathy, gentieness, ractfulnes~~ and high levels of comrnunicaricn skiib. Bakar: (1956) termed these patterns ugenq for the mascuhe coniipration (,an originacing source of an inctividual's own determination) and corramunisn (a readiness for experience, warmth, and cioseness) far the feminine pattern. These concepts have also czrried irnpiications for direction of sexual interes: (i.e., heterosexud vs. homosexual) but not strength of sexual matlvatlon, p o t e n c y , or virility (Terman &Miles, 19%). For e x a m p i e , the initial research on Scaie 5 of the MMPI was based on efforts ro identifv male in\.ersicn (feminizacionj and sexual orientation, Records from male psychiatrnc patients seeking treatment for ego-dystonic homosexual tendencies were contrasted with ::sample of males given the MMPI as they were being inducted into the Army at 2 military base eariy in World War 11. The items differentiari~g between the two samples cons:ituted the nucleus of Scale 5. (Hathaway, 1956, attempted to End sufficient numbers of sema2iv inverted female patients for delreiopment of z Femininity-Masculinity iFmi scale bur was unsuccessful.) Additional items to complete Scaie 5 were found by simpIy tallying the differences ir, item endorsements between samples of adult mer: and women. The two approaches to item selection (clinical cases i7s. normal and men-in-general vs. women-in-general' yielded essendab the same set of items. Constantinopk (I 973) raised serious doubts that masculinity and femlnkitp were appropriately conceptualized as simply two ends of a single bipo!ar continuum. She proposed viewing cbese traits as two separate patterns of personality organization, differing in degree in both mzles and females. Foilowing Constantinopie's lead, a number of invesrigators (e.g., Baucorr,, 1946; Bern, 197%;Spence, Helmreich, & Stapp? 1974, 1975) have been persmded to carry out investigations aimed at generating two separate midimensionai scales

Downloaded by [Deakin University Library] at 03:41 07 November 2015

488

PETERSON AND DAHLSTROM

to replace previous brpolar scales of mascuiin~t~i-fem:ninity.As independen: +--, ,,dts, each has aeaning in the personalities of both males znci females. Bern's and Spence et al.'s work has proceeded f r o 3 sets ot quae arbitrary assumptrons abour the nature of nascuimicy or feeminmev e~therin terms of stereotvped bei:efs :hat peopk hoici ahout these personal:rv atrr:b.ites cf menin-genera! and wcmen-:n-general (Basow, 1980) or h terms of traits deemed vaiuable or ciesIrabie for indivJriuais of each sex. Baucom (19'76) established a set of scales for massuiinity and' femininity based empirically on wha: men and women actmalkp say about themselves in their answers to a personaiiry invenmrp.

Saucom's (1376)research based on the CP: is the most similar line of invescigation in this area to our research ieading to separate gender-roie scaies far the Minnesaca Mnltiphasic Personality Inventory-.2 (MMPi-2. Using records from maie and female college undergraduares who had &en the CPI as part of 2. course on psychological assessment, Baucom carried out hers aarraiyses to identify items that tvp:fie:i either male or female su'sJects. That is, the items that he was searching for were those that were endorsed by a dear majority (ac least 70%) of members of one sex (either male or female) and answereri i3 that same direcrion by a: least I\"% fewer members of the ocher sex, Use of these crireria assured that any item seiecced clearly characterized 2. giver. sex bur did not necessariiQ have to be characterisric of d l or nearly all members of that sex. The two resulting scales were Masca!ir:ity !MSC: and Femininity (FhlX::;Baxom, 1996;Gough. i958:. A similar strategy has beerr employed in our investigation employing the ieerr! p o d of che MMPI-2 to develop separate gender-role scales for the new XMPH, tiawever, the data base xias provided by samples of adult men and women from communities across tbe

U.S. Baucoa (1916) was also one of the first investigators to recomnersd employing separate gender-role measures in a circumplex, capitalizing on the use of two uncorrelated dimensions to form a typology of gender roles. That is, he used his m o C?1 scales to define four quadrants: stereotyped mascuiine (high d (high femininity, icw rr?,ascu!ini:s, low femininity)? s ~ e r e o t ~ p efeminine msasculinirf), androgynous (high on both measures), and undifferentiated (low or. 'both\. Ln snch a circumplex iliusxrated in Figure 1 , the traditional scales p s t discussed can be envisioned as falL Sipoiar Ma~ccTin:tv-Fe~iniII;~t~i ing along a diagonal running from stereotvped mascuiine tto stereotyped feminine. A similar use of the two gender-roie scaie.: developed for the M M R - 2 is envisioned ro supplement the interpretation of Scale 5. 1f the sepzate genderrole measures on the MMPh-2 are used tc form rhese same gender-role :ypes, tberi Scale 5 sholrlcf be related to the= along a s:milar diagonalfroin extreme rnascuiinity to excreme femininlry.

G.V CGF AND SCALE 5

489

High

Ksscaliniry

Downloaded by [Deakin University Library] at 03:41 07 November 2015

Hascsline Stereotype ( tiJf/LP)

High Fe~ininity

Low Feminini t

L'ndifferentiated Fattterri ;'LM.,/LF>

Femlcine Stereotype (LY;EfF)

Low Masculinity

h4ETHQDS AND RESLITS Item endorsement data were used from the 2.6X men and women of rhe rescandardization sample for the MMPI-2 (Butcher, Bahistron, G r a h a n , lellegen, 5r Kaernrner, 19891. These subjects were recrairsci from seven different states across :.he U.S. They ranged in age from under 23 to over 80 and in education from less than high school to several years ?ostcoiiege. Although most were either R'hite or Siack, Xative ilrnerican, Asim-rlmerican. and Hispanic-American subjects were also included. They were paid s n o x i m i amount for their participation; ail testing was anonymous. T a h e acswers to each iten? were d i e d separateiv for each sex in Terms oi the percentage answering "true" or "csnno: say." (See Appendix I :r: the hfMP1-Z: h4antcai for cS,dmir.isreringand Scoring; 8utcher et a:., 1989~ The cr::eria of item separatmn between mer: and women In Eerm5 of t k r self-descii?trons used in these analyses were the same as :hose nrrhzed 5v B a x o m iIQ76) for hla CP: scales: at least 7 8 0 of exher men c?r women answermg :he rten one u-av ;erther true or fake! and at least 10% fewer menbeis of the opposite sex znswertng :he item in that same dlrectron. iThs method assures that each :tern chosen reflec:~

Downloaded by [Deakin University Library] at 03:41 07 November 2015

a characteristic of the majoriry, alrhough not necessarily the totaiity, o i that sex.) Al! items meeting these criteria were grouped intc twc new scales: the Gender Role-kbasculine $ 3 4 ) for the 41 items endorsed by a dear majoriry of men, an2 the Gender Role-Feminine (GF) for rhe 46 Items endorsed by a dear majoriq of women (Peterson, !99i ). There are no irems in coxmorr becween the G.M and CF scaies. The item numbers comprising these two scales are listed in Yabies I and 2, together with rhe direction 91 scoring on the hfMPI-2. Means and standard deviations for the two scales are also s h o w . Two corrections from the listing of these scales. in the MM?l-2: M a m a ! for Administering ridScoring (Eiutcher et a!., 1989) should be nored: Erem 331 on GM is scored when answered false and Item 733 on CF is scored w h e ~answered rrue. Readers wishing ro score records hased on the original MMPI can use shortened versions of GM and GF, referred t,a as GM-S and GF-S, by - c developing keys rrom the item ?is:s and scoring direcrions provided ir. Tables 1 and 2. The means and standard deviarions for GA%-Sand G&S for maies and females in the res~andarclizatior,sampie are atso provided in the tables. The product-morner,: ccrrrelatior: of GM with GM-S ic .?8 for both sexes; the correiations of GF and G F S art .9i and .?2 for men and women, respectively. Externai carrelares of the GM and GF sraies should therefore apply ro the GM-S and G F S as &I. TABLE i

Item Compos~tior;of the GM Scale on the MhFI-2 and of the GM-S on the Original W I

Note. For tine GM scaie: far maies, N: = 27.87, 5 3 = 4.87.for females. M = 28.83, S 3 = 6.5 1. Fo;.the G.M-S (shortened versionj: for males, M = 33.90, SD = 4.35, ibr iemaies, K = 25.1'3, SL; = 5.74. aT. L he directloc of scoring for item 331 has beep corrected from the k i n g in :he MM?l-2. Manta! for .4drriirrh:e::ng and Smring (Butcher er ai., 1989)."Item 47 Lag siig:xlr: different wording irox that in h':mI-2.

Downloaded by [Deakin University Library] at 03:41 07 November 2015

Gh? GF, AXD SCALE 5

491

In Table 3: che intercorrelations are shown between t-he ariginzl Scaie 5 and

the t w new ~ gender..role measures (GM and CR separately for males and females. For each sex, GM anci GF are correlated - .IG. Scaie 5 is cnrreiated to a greater extenr with GF thar. with GiM. The expianation b r this difieierence in magxirude of correiation lies with the degree of item owrlap (in either the same or opposire direction oi scoring) rhat occurs between these scales (see Table 4':. That is, 15 k e n s for men or 16. items far women ir: the GF scale are scxed in the same direction as Scale 5 , whereas only 9 items of the GM scale appear on Scais

TABLE 3 Raw Score hterfo~erahom.4mong GM, GF, and Scale 5 for Males and Femaies m the h W I - 2 Restandardization Sam~le

Gh4 GF Scak 5

- .:+**

-,j-,?* :?* - .>I. -.

.-74*+ ~

.%**

.43**

Note. Coefficients above the diagonal are for men. beiou the d~agonalfar women. males; n -- 1,462 for femaies. *g < .0?!. **p < .OC3l.

7,

= 1, !33 far

492

PETERSQN AND DAEILSTRQM

Downloaded by [Deakin University Library] at 03:41 07 November 2015

TABLE 4 Item Overiap of GM and GF Wlth the Baslc Scaies of the MMPI-2

5, which are aIi score2 in :he opposite direction. The corresponding item overiap oi G M and GF with the other scales in the bask h4h4FI-2 profiie are a h listed in Table 4,m Z their correkitims are shown in Yabie 5 .

Score Distributions As shown iri F>gures2 and 3 , the distribctmn of scores for males on GF and for females on GM ere biiaterally symme:ric an5 approx:mare?v nornai In form. However, the d~strlbudonsfor femaies cn GI: and males on GM are skewed negativeiy. Inspectior, of the distribufor! of GM raw scares for males and fenales (see Figure 2) cc?nfirmst h e intuitively obvious hypothesis: males tend to obtair. higher scores on CM thar: do females. However, the females' scgres on this scaie are spread o \ w E greater range than the maies' scores; some sccred aimost the maximum possible score. On che GF sce'le: the scores from the n d e ssbjects are more clearly separated from those of the [elemale subjec:~. The i -score norms for adults or: these two scaies are provided In Appendix A, Table 6! of Butcher et al. (1989). 1

"

7

Age Trends It 1s alsc important :o nme that the responses tc these rwo gender-role scaies

vary over the agt range of s&jects :n the restandard~zadonsample. Tables 6 ancl 7 irst the means ancl standard deviatmns of scores on the GM and GF scsies

TABLE 5 Raw Score Correlanons of G M and GF Scales \\-&I the Basic Scdes o; the XkPI-2 M&"

h!vP:-2

G\!

hie.

Downloaded by [Deakin University Library] at 03:41 07 November 2015

+.rldcspoase;'partner ratings coliected on the subjects ir: the restandardization sample showed that :he GF scare is correlated with 'hypercritical behavior :n rnaies? as we3 as with reiigiosity and avoidance of swearing, with bossiness and poor temper control. For- both maies and females, GF is also related to misuse oi nonprescription drsgs znd aicnhol. In females, GM reflects a lack of worrying, honesty, and an enterprising wiliingness tc expiore new things. For both males and females! GM is correiate;! a:tb high self-confidence, persistence coward one's goal, and a lack of feeling of seif-reference. However, conjoin? use of the G?d and. GF scaies enables the mvestigator t o ciistingllish among indivrduak wi:h an androgynous pattern ci characreriscics from those who have little of either gender-masculine or gender-feminine characteristics.

DISCUSSION The current scales and Baucorn's (1976) CPI scaies empicv onh- items that &Yerentlare statlst~callvbetweec the sexes, regardless of :tex fa7oraSd:ti. in

Downloaded by [Deakin University Library] at 03:41 07 November 2015

coztrast. both the Bern Sex-Role Inventor:; IBSRi; Bern, 1974) anci the Personal Atrributes Q~estionnaire(PAR,Spence et al., 1994,3975) were developed b y drawing from a pooi of judged-favorable characteristics. Items or: the BSRI a x those that were ccnsidered to lie more appropriate fen men or for women. The P A Q includes items judged to be more characteristic of cne sex or the other. Ir: summary, although items were selected for inciusiort o n the BSRI an2 PAR based on retings ~ia~proprizreness anci trpicakiry, :he newly de\dopeci CM and GF scales and Baucom9sscaies for the CP! contain kerns shown KO differentiate empirkal'lv between *ihesexes. CM and CF are highly correlated wirh ail three of t h e just-mentioned sex-r& inventones (Peterson7 1992.z),but they are most . especially correia:ed with Baucom's CPI scaies c:ue to the similar me:hoci s f derivation, The newiy formed independent gender-role scaies are intended tc serve as sxpplements te Scale 5 of the PLIMPI-2. These new scales may serve as measure.menrs of masculinity and femininity within the cor.text of a widely used inventory-the MMPT-2. Scaie 5 of the Mh4PI-2, iike its predecessor on the . . original MMi3Z7 asssmes that rnascul:nirr; and femininity lie on a sipoiar dimensicn, This placement o n opposi:e ends of the continwm portrays mascuiinity and femininity as mutuaily exclusive traits, Wkea using this type of instrument, ir is impossi'!ie to differer,riate between ancirogy-nous and undifferen~atedindi\ii&;als because both :end TO score somewhere in the middle, whereas sex-type6 :ndividsals fail at one of &e two extreme ends of the scale. Independerrt, unipolar mascuiinity and feminixicy scaies for the MMPI-2 help solve the mdciie-ground dilemma that exists for Scaie 5. By using GM and GF, the foliowing fourfoib gender-role typaiog-y ma:; be cunstmcred sirnilzr to that employed by Baucom (1982) with his MSC and F?\?X scales derived from the CP!: androgynous. (high masculinity, high feminkryjl stereotyped masculine (high nascuiiniry, isw feemininity), stereot-yped feminine (iow rnascuhni:y, high femininity), and undifferentiated (Iow mascuiinir:;, iow femininity), By offering ~ of the MMP1-2, independent measures of masculinity and fernininky t : users the conjakt use of these scales wiil fdrclst: inportant distinctions in regard to rhe patterning of gender roles in an individual that would be overiooked ir, the application of Scale 5 alone. A summary of additiond internal (MMFI-2) and externai correlates b r the b u r gender-& types as defined by this circump'rex may be f m n d in Butcher et al. (1989). These fiind:ngs wfii be reported in greater de:aii in $ a s r e publications (Peterscn, 1992a, 199%). 7

REFERENCES Sakan. D :1955). Tiu.duaii:? 3! irumar, existence. Chicago: Rand McNalIv. Basow, S. A. (138Oj. Sex-roic stereotypes: 'Pras!irior:s cnd altemariws. Mont.erel, CA: Erooks./Coie.

Downloaded by [Deakin University Library] at 03:41 07 November 2015

GM, GF,AND SCALE 5

W. Grant Dahlstrom Departrne~tof Psvchologg U~iversityoi Xorth Carolina

CRP 3270,Davie 3aH C h a p ! Hill? NC 27599-311C Received Februarv 14, i992

4%

The derivation of gender-role scales GM and GF for MMPI-2 and their relationship to Scale 5 (Mf).

Separate scales for masculine and feminine gender roles (GM and GF, respectively) were developed for the Minnesota Multiphasic Personality Inventory-2...
839KB Sizes 0 Downloads 0 Views