You are on page 1of 13

Sex Roles, Vol. 39, Nos.

11/12, 1998

Assessin g the Current Validity of the Bem


Sex-Role Inventory
Cheryl L. Holt 1
Saint Louis University

Jon . B. Ellis
East Tenn essee State University

The Bem Sex-Role Inventory (BSRI) is a widely used in strum ent in m easuring
gen der role perception s. Recent concerns regardin g the validity of the adjectives
have arisen as changes in the roles of m en an d wom en have occurred in
Am erican society since the 1970 s. A partial replication of the m ethod that
Bem (1974) used to validate the m ascu line an d feminin e adjectives com prising
the in strum ent was con du cted. All bu t two of the adjectives were validated
using Bem s criteria. These findin gs suggest that the BSRI m ay still be a valid
in strum ent for assessin g gen der roles. However, evidence was revealed that
tradition al m ascu line an d femin ine gender role perception s m ay be weaken ing.
Fu ture validation of the BSRI is warran ted in light of these pattern s.
In the fie ld of psychology, much research is conducte d involving individuals
perceptions of gende r role s, and be havioral as well as attitudinal correlate s.
Gende r roles may be defined as e xpe ctations about what is appropriate
behavior for e ach se x (We iten, 1997, p. 325) . O ne can also add to this
definition the expe ctations which are he ld about appropriate personality
characte ristics for e ach se x. In 1974, Bem develope d the Be m Sex-Role
Inve ntory (BSRI), an instrume nt used to measure gende r role pe rceptions.
The BSRI is a wide ly use d instrume nt in psychology and othe r fie lds because it measure s masculine and fe minine gende r roles se parate ly, is able
to yie ld a measure of androgyny, and has ade quate psychome tric prope rtie s.
1

To whom corre sponde nce should be addre sse d at St. Louis University, De partme nt of
Psychology, Shannon Hall, Room 201, 221 N. Grand Boulevard, St. Louis, MO 63103; e-mail:
plagecl@ slu.edu
929
0360 0025/98/1200 0929$15.00/0

1998 Plenum Publishing Corporation

930

Holt and Ellis

Through increase d use of the instrume nt, the se psychome tric prope rties
and the correlate s of the BSRI have contribute d to its value .
Be cause the instrume nt is ove r two de cade s old, it was felt that a reassessment of the validity was in order. The psychome tric prope rties of the
scale were demonstrate d to be strong, thus making it an attractive instrument for use in psychological re search. It was publishe d in 1974, and is
still fre quently used in rese arch today. The prese nt authors have use d the
BSRI in se ve ral studie s (Holt & Ellis, 1998; Holt, Zinser, & Te nnyson,
1998) , but have e xpre sse d concerns ove r the current validity of the items
composing the scale .
The BSRI enable s re se arche rs to obtain a measure of psychological
androgyny, or high leve ls of both masculinity and fe mininity. This type of
individual according to Be m (1974) , is more adaptive in that the y are not
bound by behaviors associate d with traditional masculine or feminine gender roles, but may fee l comfortable e ngaging in behaviors appropriate for
either ge nde r. This give s these individuals increased flexibility and adaptability because their range of behaviors is not restricte d by a traditional
gende r role (Be m & Le nne y, 1976) . Research has supporte d the ide a that
androgyny corre lates with a numbe r of othe r positive attribute s, such as
highe r leve ls of identity formation in colle ge stude nts (Bem, 1974; Heilbrun,
1976; Orlofsky, 1977) . In addition, androgynous individuals have bee n demonstrate d to have more re asons for living than ge nde r-type d individuals
(Ellis & Range , 1988) . These findings sugge st that androgynous individuals
te nd to be more psychologically he althy and function more adaptive ly in
mode rn living. In contrast, rese arch sugge sts that individuals who are undiffere ntiate d in te rms of gende r role (low on both masculinity and fe mininity) te nd to be le ss adaptable ( Be m, 1974; Glaze r & Duse k, 1985;
Heilbrun, 1968; Latorre , 1978; O rlofsky, 1977) .
Be m ( 1974) re porte d high internal consiste ncy and test-retest reliability
of the BSRI. Coe fficient alphas compute d for masculinity and fe mininity
reveale d high re liability (Masculinity alpha = .86; Femininity alpha = .82) .
The BSRI test-retest reliability within a sample of 28 male s and 28 fe males
was de monstrate d to be highly re liable over a four wee k pe riod (Masculinity r = .90; Femininity r = .90; Androgyny r = .93) .
The psychome tric prope rties of the BSRI have , in addition, bee n e xamine d through many inve stigations (Bem, 1975; Be m & Lenne y, 1976;
Be m, Martyna, & Watson, 1976; Gaudre au, 1977; Gross, Batlis, Small, &
Erdwins, 1977; Pe dhazur & Te te nbaum, 1979; Strahan, 1975; Wakefie ld,
Sase k, Friedman, & Bowden, 1976; Walkup & Abbott, 1978; Waters, Wate rs, & Pincus, 1977; Wilson & Cook, 1984). Through the se research efforts
Be m de ve lope d the short form of the BSRI (Bem, 1981) , in which half of
the items from the original instrume nt were eliminate d to form a 30-ite m

Assessin g th e Cu rrent Validity of th e Bem

931

inve ntory. The inte rnal consiste ncy of the short form was highe r than that
of the original, as items which de monstrate d poor item-total correlations
with the masculinity and femininity scale s were eliminate d. The BSRI short
form corre late s highly (around .90) with the original form.
It was thought that although the BSRI is a well-constructe d instrument, the original adje ctive s repre se nting ste reotypically masculine and
fe minine ge nde r roles were se lected ove r 20 ye ars ago. This ope ne d up the
possibility that the BSRI may be outdate d in terms of the representations
of masculine and feminine ge nde r roles.
The re is, howe ve r, evide nce that colle ge stude nts pe rceptions of gender role s have not change d since the 1970 s (Stree t, Kimme l, & Kromrey,
1995) . O ne study examine d colle ge stude nts ge nde r role pe rceptions of
the ide al man, ide al woman, most men, most women and the mselves, by
comple ting the Se x Role Trait Inventory (SRTI) (Stree t & Mee k, 1980) .
Stude nts comple te d the SRTI for e ach of the five obje cts (ide al man, ideal
woman etc.). It was found that both the fe male and male stude nts rated
the ide al woman as androgynous, women pre fe rre d an androgynous ideal
man, and men preferred a masculine ideal man. Both female and male
stude nts rated most women and men in terms of stereotypical gende r role s.
It was conclude d that little has change d in the gende r role perceptions of
colle ge stude nts since the 1970 s (Stre et, Kimme l, & Kromre y, 1995) .
A re late d study conclude d similar findings whe n examining the ge nde r
role pe rceptions of colle ge faculty (Stre et, Kromre y, & Kimme l, 1995) . Using the same procedure as the afore mentione d study with colle ge stude nts,
it was found that male and fe male faculty preferred an androgynous ideal
woman and a masculine ide al man. Again, both male and fe male faculty
saw most men and women in te rms of the stere otypical gende r roles. Although these conclusions must be vie wed with caution and limite d generalizability, the se studie s presented e vide nce that traditional gende r roles
such as the masculine man and the feminine woman were still ope rating
in modern America.
Edwards and Ashworth (1977) replicate d Be m s sele ction of masculine
and fe minine ite ms in 1975 and in 1976, and found that only two of the
items (the masculine and feminine adje ctive s) were rate d as significantly
more de sirable for the intende d sex (in the e xpe cted direction) than for
the opposite se x. Howe ve r, these re se arche rs conducte d face-to-face interviews to colle ct the data which may have influe nced the results through
social desirability effe cts.
The pre se nt inve stigators atte mpte d to provide a more stringe nt e xamination of the BSRI adje ctives to determine whether they are curre ntly
valid re pre sentations of ge nde r role perceptions in te rms of masculinity and
fe mininity. Although there has be en evide nce that ge nde r role pe rceptions

932

Holt and Ellis

have not change d since the 1970 s (Stre et, Kimme l, & Kromre y, 1995;
Stree t, Kromre y, & Kimme l, 1995) these studie s were conducte d using the
SRTI, and it was felt that the validity of the adje ctive s on the original BSRI
nee de d to be examine d.

METHOD
Participants
O ne hundre d and thirty e ight individuals (68 men and 70 wome n) participate d in this study. The sample was pre dominate ly White , and middle
class. The age s of the participants range d from 18-52, with the ave rage age
being around 25 years. Individuals were aske d to volunte er from summer
te rm psychology course s at a Southe rn unive rsity. The y were offere d e xtra
course credit for their participation. This sample may have diffe re d from
Be m s (1974) sample of fifty male and fifty fe male Stanford Unive rsity unde rgraduate s. The most salie nt dime nsions on which these two sample s
most like ly differe d are in ge ographic re gion and in socioe conomic status.
Materials
A survey packe t was distribute d to participants including dire ctions and
the 40 (20 masculine and 20 fe minine ) adje ctive s from the original 60-ite m
BSRI. The 20 neutral items from the BSRI were not include d, as the present study was de signe d to assess change s in perceived ge nde r role s in terms
of the masculine and fe minine adje ctive s. It was thought that there would
be no re ason to e xpe ct that perceptions of the ne utral adje ctives would
have change d over time . A se ve n-point Likert scale was used, with the root
que stion, How desirable is it in American socie ty for a man (woman) to
posse ss each of the se characte ristics , (Bem, 1981, p.17) . The anchors include d e xtre mely de sirable (7) to not at all de sirable (1).
Procedu re
All of the data was colle cted in psychology classrooms and the procedure was standardize d for e ach class. The expe rimenter distribute d survey
packe ts and aske d that the participants comple te the questionnaire according to the instructions provide d. A statement on the cove r page of the que stionnaire packe ts inform e d the participan ts that the ir re sponse s would
remain anonymous and that participation is comple tely voluntary. It is im-

Assessin g th e Cu rrent Validity of th e Bem

933

portant to note that the participants rated how de sirable it is for either a
man or woman to posse ss the characte ristics, but no subje ct rate d the characte ristics for both sexe s. The procedure , instructions, and mate rials used
in the pre se nt study were ide ntical to the method that Bem ( 1974) used
to validate the adje ctives in the BSRI, with the exception that only the
twenty masculine and twenty fe minine adje ctives which appe ar on the original BSRI were examine d (we did not include any neutral items nor the
unuse d pe rsonality characte ristics comprising the original pool from which
the final ge nde r-type d or the neutral ite ms were sele cted). These othe r
items were not include d be cause the purpose of the pre se nt study was to
assess the validity of the masculine and fe minine adje ctives alre ady in use
on the BSRI, and not an atte mpt to ge nerate a new ge nde r role inve ntory.
Design
The participants were randomly assigne d to the two conditions (instructions to rate a man or a woman) by randomly distributing the questionnaire packe ts with the two differe nt instructions. A characte ristic was
validate d as fe minine if it was judge d by both male s and fe males to be
significantly more de sirable for a woman than for a man. A characte ristic
was validate d as masculine if it was judge d by both male s and female s to
be significantly more de sirable for a man than for a woman. If statistical
significance was not obtaine d, it would be conclude d that the characte ristic
in que stion was not an appropriate mode rn representation of masculinity
or fe mininity.

RESULTS
Two-taile d t-tests re ve aled that all of the masculine adje ctive s were
rated as significantly more desirable for a man than for a woman, p < .001
(se e Table s I & II). All but two of the feminine adje ctive s were rated as
significantly more desirable for a woman than a man, p < .001. O ne of
the se exceptions was the fe minine adje ctive loyal which was only marginally rated as more de sirable for a woman (M = 6.26, SD = 0.86) than
for a man (M = 5.94, SD = 1.32) , t (136) = 1.70, p = .09. The othe r
was childlike which was only marginally rated as more desirable for a
woman (M = 2.41, SD = 1.43) than for a man (M = 2.00, SD = 1.34) , t
(136) = 1.75, p = .08. A Cronbach alpha was compute d to asse ss the
inte rnal reliabilitie s of the BSRI in this sample for the masculinity scale
( a = .95) and for the femininity scale ( a = .92). These are highe r though

934

Holt and Ellis


Table I. Means and Standard De viations for BSRI Masculine Ite ms

Masculine Item
Acts as a Leader
Aggressive
Ambitious
Analytical
Asse rtive
Athletic
Competitive
Defends own Beliefs
Dominant
Forceful
Has Leadership Abilities
Independe nt
Individualistic
Makes Decisions Easily
Masculine
Self-Reliant
Self-Sufficient
Strong Personality

Desirability for Male s

Desirability for Female s

Respondent Gender

Respondent Gender

Male

Female Combined

Male

Fe male Combined

MN

5.83

5.65

5.71

4.35

3.75

3.93

SD

1.09

1.30

1.23

1.39

1.31

1.35

MN

4.63

4.26

4.39

2.40

2.67

2.59

SD

1.50

1.77

1.68

1.27

1.34

1.32

MN

5.96

6.28

6.17

5.10

4.52

4.69

SD

1.52

0.96

1.18

1.59

1.57

1.59

MN

5.04

4.76

4.86

3.75

3.33

3.46

SD

1.71

1.32

1.46

1.29

1.31

1.31

MN

5.33

5.52

5.46

4.10

3.58

3.74

SD

1.37

1.46

1.42

1.59

1.46

1.50

MN

5.46

5.41

5.43

4.80

4.29

4.44

SD

1.61

1.33

1.42

1.32

1.46

1.43

MN

5.63

5.39

5.47

3.95

3.63

3.72

SD

1.41

1.50

1.46

1.47

1.39

1.41

MN

5.63

5.89

5.80

4.65

4.50

4.54

SD

1.44

0.99

1.16

1.39

1.43

1.41

MN

4.50

4.67

4.61

2.45

2.27

2.32

SD

1.87

2.04

1.97

1.28

1.27

1.26

MN

3.96

3.74

3.81

2.45

2.60

2.56

SD

1.99

2.12

2.07

1.70

1.48

1.54

MN

6.00

5.98

5.99

4.55

3.98

4.15

SD

1.02

1.13

1.08

1.43

1.55

1.53

MN

6.04

6.22

6.16

4.40

4.58

4.53

SD

1.20

1.01

1.07

1.47

1.58

1.54

MN

5.67

5.39

5.49

4.00

4.75

4.53

SD

1.43

1.27

1.33

1.59

1.47

1.53

MN

5.71

5.54

5.60

4.65

4.69

4.68

SD

1.12

1.28

1.22

1.31

1.21

1.23

MN

5.96

6.13

6.07

1.90

1.48

1.60

SD

1.30

1.59

1.49

1.48

1.07

1.21

MN

5.88

6.00

5.96

4.40

4.40

4.40

SD

1.12

1.10

1.10

1.60

1.67

1.64

MN

5.92

6.28

6.16

4.50

4.85

4.75

SD

1.21

0.93

1.04

1.36

1.58

1.52

MN

6.04

5.91

5.96

5.05

4.75

4.84

SD

1.08

1.07

1.07

1.54

1.78

1.71

Assessin g th e Cu rrent Validity of th e Bem

935

Table I. Continued

Masculine Item
Willing to Take a Stand
Willing to Take Risks

Desirability for Male s

Desirability for Female s

Respondent Gender

Respondent Gender

Male

Female Combined

Male

Fe male Combined

MN

5.79

6.13

6.01

4.65

4.50

4.54

SD

1.38

1.00

1.15

1.69

1.52

1.56

MN

5.08

5.78

5.54

4.25

4.25

4.25

SD

1.71

1.19

1.42

1.74

1.36

1.47

Table II. Me ans and Standard Deviations for BSRI Feminine Items
Desirability for Male s
Respondent Gender
Feminine Item
Affectionate
Cheerful
Childlike
Compassionate
Does Not Use Harsh
Language
Eage r to Soothe Hurt
Fee lings
Feminine
Flatterable
Gentle
Gullible
Loves Children
Loyal

Male

Female Combined

Desirability for Female s


Respondent Gender
Male

Fe male Combined

MN

4.50

4.41

4.44

6.30

5.81

5.96

SD

1.50

1.81

1.70

0.86

0.91

0.92

MN

4.79

4.78

4.79

6.00

5.79

5.85

SD

1.79

1.53

1.61

1.03

1.01

1.01

MN

2.04

1.98

2.00

2.45

2.40

2.41

SD

1.16

1.44

1.34

1.28

1.50

1.43

MN

4.75

4.50

4.59

6.20

6.17

6.18

SD

1.48

1.92

1.77

0.70

0.91

0.85

MN

4.00

4.07

4.04

5.00

4.77

4.84

SD

2.02

1.61

1.75

1.75

1.86

1.82

MN

4.25

4.07

4.13

6.00

5.79

5.85

SD

1.73

1.90

1.83

1.08

1.07

1.07

MN

1.50

1.61

1.57

6.05

6.10

6.09

SD

1.06

1.36

1.26

1.47

1.19

1.27

MN

3.42

3.80

3.67

5.10

6.10

5.81

SD

1.32

1.72

1.59

1.41

7.31

6.19

MN

4.04

4.00

4.01

6.10

5.75

5.85

SD

1.92

1.94

1.92

0.85

1.04

1.00

MN

1.33

1.39

1.37

2.90

2.67

2.74

SD

1.05

0.86

0.92

1.86

1.60

1.67

MN

5.04

5.30

5.21

6.50

6.29

6.35

SD

1.63

1.49

1.53

0.69

0.92

0.86

MN

5.92

5.96

5.94

6.20

6.29

6.26

SD

1.28

1.35

1.32

1.01

0.80

0.86

936

Holt and Ellis


Table II. Continued

Feminine Item
Sensitive to the Nee ds
of Others
Shy
Soft Spoken
Sympathetic
Tender
Unde rstanding
Warm
Yielding

Desirability for Male s

Desirability for Female s

Respondent Gender

Respondent Gender

Male

Female Combined

Male

Fe male Combined

MN

4.58

4.48

4.51

5.85

6.02

5.97

SD

1.44

1.75

1.64

0.93

0.98

0.96

MN

2.04

2.13

2.10

3.30

2.96

3.06

SD

1.20

1.41

1.33

1.56

1.30

1.38

MN

2.58

2.28

2.39

4.60

3.94

4.13

SD

1.38

1.47

1.44

1.85

1.51

1.63

MN

4.46

4.59

4.54

5.75

5.94

5.88

SD

1.50

1.84

1.73

1.07

0.95

0.99

MN

4.00

3.98

3.99

5.60

5.98

5.87

SD

1.82

1.96

1.90

1.14

0.91

1.00

MN

5.13

4.85

4.94

6.10

6.17

6.15

SD

1.54

1.70

1.64

0.79

0.69

0.72

MN

4.54

4.63

4.60

6.10

6.04

6.06

SD

1.64

1.74

1.70

0.85

0.71

0.75

MN

3.29

3.11

3.17

4.65

4.58

4.60

SD

1.52

1.65

1.60

1.66

1.44

1.50

still comparable to Bem s (1974) inte rnal reliabilitie s for these scale s. An
e xam ination of the magnitud e of diffe re nce s in the diffe re nce score s
( change ) of the de sirability of the adje ctives for male s and fe male s from
Be m s 1974 sample to the present indicate s that the magnitude of difference in desirability has decrease d ove r this pe riod (se e Table s III & IV ).

DISCUSSION
The finding that all of the masculine and all but two of the feminine
adje ctive s were rated as significantly more desirable for a man or a woman,
respectively, sugge sts that the BSRI may still be a valid measure of ge nde r
role perceptions. These findings are not in accord with those of Edwards
and Ashworth (1977) , who found that only two of the adje ctives were rated
as significantly more de sirable for one ge nde r in the intende d direction. It
was not expe cted that the prese nt results would be in agre e ment with Edwards and Ashworth. The afore mentione d authors used face -to-face inter-

Assessin g th e Cu rrent Validity of th e Bem

937

Table III. T-Tests, Difference Score s, and Change in Social Desirability for BSRI Masculine
Ite ms From 1974 1997
Differe nce Score
Masculine Ite ms

T-Test

1974

1997

Change
1974 1997

Acts as a leader

8.13

3.35

1.79

1.56

Aggressive

6.98

2.60

1.80

0.80

Ambitious

6.23

2.64

1.48

1.16

Analytical

5.94

2.06

1.40

0.66

Asse rtive

6.92

2.37

1.72

0.65

Athletic

4.07

2.02

0.99

1.03

Competitive

7.15

2.76

1.75

1.01

Defends own be liefs

5.72

1.50

1.26

0.24

Dominant

8.10

3.63

2.29

1.34

Forceful

4.04

2.38

1.25

1.13

Has leadership abilities

8.17

2.50

1.84

0.66

Independe nt

7.22

3.06

1.63

1.43

Individualistic

3.93

2.06

0.96

1.10

Makes decisions easily

4.43

1.30

0.92

0.38

Masculine

19.32

5.25

4.47

0.78

Self-reliant

6.59

2.26

1.56

0.70

Self-sufficient

6.36

2.74

1.41

1.33

Strong personality

4.63

2.25

1.12

1.13

Willing to take a stand

6.32

1.98

1.47

0.51

Willing to take risks

5.25

2.00

1.29

0.71

Difference scores indicate differe nce s be tween the de sirability of the characteristic for males
minus the de sirability for females.
b
Change indicates the change in difference score from 1974 to 1997.
c
p < .05.
d
p < .01.
e
p < .001.

views to colle ct the gende r role pe rceptions of their participants, and the
pre sent study use d anonymous questionnaire s. By using inte rvie ws the participants may have responde d in a socially de sirable (and the refore, nonste re otypical) manne r. For e xample , the participants may not have wante d
to report in the presence of an e xpe rime nte r that they thought that it is
desirable for wome n to be yie lding and men to be ambitious.
O the r pote ntial e xplanations for the discrepancie s in findings may be
relate d to difference s in the de mographics of the participants. The Edwards
and Ashworth study use d inte rvie ws with stude nts in a campus library and
union building of a unive rsity. The participants in the prese nt study were

938

Holt and Ellis

Table IV T-Tests, Differe nce Scores, and Change in Social De sirability for BSRI Feminine
Ite ms From 1974 1997
Differe nce Score
Feminine Ite ms
Affectionate
Cheerful
Childlike
Compassionate
Does not use harsh language
Eage r to soothe hurt feelings
Feminine
Flatterable
Gentle
Gullible
Loves Children
Loyal
Sensitive to ne eds of others
Shy
Soft spoke n
Sympathetic
Tender
Unde rstanding
Warm
Yielding

T-Test

6.48
4.64 e
e

1.75
e
6.70
2.62
e
6.73

id

21.02
id
2.80
e

7.02
e
5.96
e

5.37
1.70

6.34
e
4.15
e

6.68
e
5.58
e

7.27
e
5.56
e

6.50
e
5.43
e

1974

1997

1.70
1.08

1.52
1.06

1.98
1.70

0.80
1.72

1.84
1.78
5.38
1.94
2.24
1.86
1.72
1.12
1.26
1.38
1.90
1.44
2.48
1.24
1.88
1.94

0.41
1.59
4.52
2.14
1.84
1.37
1.14
0.32
1.46
0.96
1.74
1.34
1.88
1.21
1.46
1.43

Change
1974 1997
0.18
0.02
1.43
0.19
1.18

0.02
0.86
0.20
0.40
0.49
0.58
0.80

0.20
0.42
0.16
0.10
0.60
0.03
0.42
0.51

Difference scores indicate differe nce s be tween the de sirability of the characteristic for males
minus the de sirability for females.
b
Change indicates the change in difference score from 1974 to 1997.
c
p < .05.
d
p < .01.
e
p < .001.

stude nts attending a relative ly small Southe rn unive rsity, and many commute d or were originally from ne arby are as if they did live on campus.
This ope ns up the possibility that the two sample s diffe red in both ge ographic region (which is highly like ly) and on de mographic characte ristics.
Although all but two of the adje ctives were validate d in te rms of statistical significance , an inte resting patte rn emerged upon e xamination of
the change (since 1974) in the magnitude of the difference scores for social
desirability of the adje ctives for men and wome n. The positive sign of the
change score s indicate s that in Be m s 1974 sample there was a large r magnitude of diffe re nce than in the present sample . For e xample , the magnitude of diffe re nce in de sirability for male s ve rsus fe males for the masculine

Assessin g th e Cu rrent Validity of th e Bem

939

adje ctive acts as a le ade r has de crease d since Bem s (1974) study. This
sugge sts that the ge nde r role ste re otyping in the prese nt sample was weaker
than the 1974 sample . This finding is important in that it sugge sts that the
validity of the BSRI may be de creasing ove r the ye ars. Give n the se patte rns,
it is possible that if the pre sent study were to be re plicate d twenty ye ars
from now, it would be conclude d that the BSRI is no longe r a valid me asure
of ge nde r role perceptions.
The pre se nt findings are only in partial agre ement with those of Stre et,
Kimme l, and Kromrey (1995) and those of Stre et, Kromrey, and Kimme l
(1995), who found that the ge nde r role perceptions of stude nts and faculty,
respectively, reflected gende r role ste re otyping on the SRTI. If the prese nt
participants did not e xhibit any gende r role ste reotyping, the masculine and
fe minine adje ctive s on the BSRI would not have be en validate d at the statistically significant le ve l. The fact that the adje ctives were validate d is evid e n c e t ha t th e p ar tic ip ant s e xp r e s se d s om e le ve l of ge n de r role
ste re otyping. Howe ver, the finding that the gende r role pe rceptions of the
pre sent stude nts refle cts less of the traditional masculine and feminine role s
than Be m s 1974 sample sugge sts that gende r role perceptions may have
unde rgone some change since that time , which is not in accord with the
findings of the afore mentione d inve stigators. Although the findings of the
pre sent study are in partial agre e ment it is difficult to gene ralize across
the se studie s in that the pre sent used the BSRI and the forme r used the
SRTI to asse ss gende r role perceptions.
Pe rhaps the findings of this study will lend increased confide nce to
researchers who have be en using the BSRI with re servations, or who may
have avoide d the instrume nt because of conce rns about validity. Howe ve r,
the se findings are to be viewed as te ntative in that it was e vide nce d that
gende r-role perceptions have change d over the years, but not enough to
invalidate the BSRI at this time. It was thought that with the change s in
the role s of both men and women that traditional masculine and feminine
gende r role ste re otype s may be weakening in mode rn American socie ty.
Although it may be conclude d that the BSRI is still a valid instrume nt in
the pre sent day, the re was evide nce of this weake ning of the traditional
masculine and fe minine gende r role s in the pre sent sample .
The pre se nt rese archers are not concerned that the ite ms childlike

and loyal were not fully validate d. Because the significance level was still
in the marginal re gion and the means were in the predicte d dire ction, it
is not felt that the y are e nough of a threat to the validity of the instrume nt
to justify their e xclusion. It is difficult to gene rate an e xplanation as to why
the se two adje ctive s were not fully validate d. This may be a reflection of
the weakening of traditional ge nde r role perceptions which was evide nced
in the pre se nt sample .

940

Holt and Ellis

The results of the pre se nt study should be viewed with caution because
of the limite d ge ne ralizability of studie s involving colle ge stude nt participants. It is possible that ge nde r role pe rceptions are more conse rvative (and
traditional) in the South than in othe r are as of the country; the BSRI may
be le ss valid in some re gions than in othe rs. This should be kept in mind
when vie wing the results of the pre sent study, and may be a dire ction for
future rese arch.
Future rese arch with the BSRI may also include a validation study
with a sample othe r than colle ge stude nts. It would be inte re sting to dete rmine whethe r there are gene rational diffe re nce s in ge nde r role pe rceptions. Future studie s could examine the perceptions of middle -age d and
olde r adults. It is possible that their ge nde r role perceptions may differ
from those of the younge r colle ge stude nt population. Pe rhaps the most
important direction for furthe r BSRI rese arch would be to re plicate the
validation proce dure and e xamine the magnitude of difference s in de sirability of the adje ctive s for men and women, to de te rmine whether the re
is furthe r change in ge nde r role perceptions. The pre sent re sults sugge st
that the nee d for validation of the BSRI should not e nd with this study,
but should continue to be inve stigate d as the gende r role perceptions in
American socie ty change over the coming ye ars. By the mid twenty-first
century the BSRI may be come comple te ly outdate d, re fle cting furthe r
change s in gende r role perceptions.

REFERENCES
Bem, S. L. (1974). The measureme nt of psychological androgyny. Journal of Consulting and
Clinical Psychology, 42, 155-162.
Bem, S. L. (1975) . Sex role adaptability: One conseque nce of psychological androgyny. Journal
of Personality and Social Psychology, 31, 634-643.
Bem, S. L. (1981) . Bem Sex Role Inventory: Professional manual. Palo Alto, CA: Consulting
Psychologists Press.
Bem, S. L., & Lenne y, E. (1976). Se x typing and the avoidance of cross-se x be havior. Journal
of Personality and Social Psychology, 33, 48-54.
Bem, S. L., Martyna, W., & Watson, C. (1976) . Sex typing and androgyny: Further explorations
of the expressive domain. Journal of Personality and Social Psychology, 34, 1016-1023.
Edwards, A. L., & Ashworth, C. D. (1977) . A replication study of item se lection for the Be m
Sex-Role Inventory. Applied Psychological Measurem ent, 1, 501-507.
Ellis, J. B., & Range , L. M. (1988) . Femininity and reasons for living. Educationa l and Psychological Research, 8, 19-24.
Gaudreau, P. (1977) . Factor analysis of the Be m Sex-Role Inve ntory. Journal of Consulting
and Clinical Psychology, 45, 299-302.
Glazer, C. A., & Duse k, J. B. ( 1985) . The relationship between sex-role orientation and resolution of Eriksonian developme ntal crises. Sex Roles, 13, 653-661.
Gross, R. W., Batlis, N. C., Small, A. C., & Erdwins, C. (1977) . Factor structure of the Be m
Sex-Role Inventory and the Pe rsonal Attributes Que stionnaire. Journal of Consulting and
Clinical Psychology, 47, 1122-1124.

Assessin g th e Cu rrent Validity of th e Bem

941

Heilbrun, A. B. ( 1968) . Sex-role identity in adolesce nt females: A theore tical paradox. Adolescence, 3, 79-88.
Holt, C. L., & Ellis, J. B. (1998) . Exam ining gender and gender role differences in attitudes
toward abortion and euthanasia. Manuscript in preparation.
Holt, C. L., Zinser, O., & Tennyson, J. A. ( 1998) . G ender differences in object recall: The effects
of gender-related stimuli and motivation. Manuscript in preparation.
Latorre , R. A. ( 1978) . Ge nder role and psychological adjustment. Archives of Sexual Behavior,
7, 89-96.
Orlofsky, J. ( 1977) . Sex-role orientation, identity formation, and self-esteem in college me n
and women. Sex Roles, 3, 561-575.
Pe dhazur, E. J., & Tetenbaum, T. J. (1979) . Be m Sex-Role Inventory: A theoretical and methodological critique. Journal of Personality and Social Psychology, 37, 996-1016.
Strahan, R. F. (1975) . Remarks on Bem s me asure ment of psychological androgyny: Alternative
methods and a supplementary analysis. Journal of Consulting and Clinical Psychology, 43,
568-571.
Street, S., Kimmel, E.B., & Kromrey, J. D. (1995) . Revisiting university student gende r role
perceptions. Sex Roles, 33, 183-201.
Street, S., Kromrey, J. D., & Kimme l, E. (1995) . University faculty ge nder roles pe rceptions.
Sex Roles, 32, 407-422.
Street, S., & Mee k, P. (1980) . Gree k and non-Greek student perceptions of sex roles. Journal
of the National Association for Wom en Deans, Adm inistrators, and Coun selors, 43, 10-14.
Wakefield, J. A., Sase k, J., Frie dman, A. F., & Bowden, J. D. (1976) . Androgyny and other
measures of masculinity-femininity. Journal of Consulting and Clinical Psychology, 44, 766770.
Walkup, H., & Abbott, R. D. (1978) . Cross-validation of item sele ction on the Bem Sex-Role
Inve ntory. Applied Psychological Measurem ent, 2, 63-71.
Waters, C. W., Waters, L. K., & Pincus, S. ( 1977) . Factor analysis of masculine and feminine
sex-type d items from the Bem Sex-Role Inventory. Psychological Reports, 40, 567-570.
We iten, W. (1997) . Psychology: Them es and variations, briefer version (3rd ed.). Pacific Grove ,
CA: Brooks/Cole Publishing Company.
Wilson, F. R., & Cook, E. P. (1984). Concurre nt validity of four androgyny instruments. Sex
Roles, 11, 813-837.

You might also like