‘Sex Ror, Vol 46, No 78, April 2002 (6 2602)
Gender Awareness Inventory-VA: A Measure
of Ideology, Sensitivity, and Knowledge Related
to Women Veterans’ Health Care
Dawn M. Salgado,!? Dawne S. Vogt,'” Lynda A. King,'> and Daniel W. King'*
Within the Veterans Health Administration, it has become increasingly important to assess
health-care workers’ attitudes toward and beliefs about female patients, sensitivity to the
lunigue needs of female patients, and knowledge about women veterans and the programs
and services available to them, The current study descr
‘Avareness Inve
the development of the Gender
ry-VA, an instrument that assesses 3 components: enderrole ideology,
fender sensitivity, and knowledge amoag health-care stall. Data were drawn from various
Samples of Veterans Health Administration employees ina large geographical region (overall
1N > 1, 100; nationally, this population includes roughly 37% minorities). Ina series of psycho-
metric inquires, evidence for reliability and validity was established, and preliminary evidence
was provided forthe instrument's underlying factor structure.
KEY WORDS: gender, employee
Over the past two decades, a steady stream
of research has revealed problems in the provi-
sion of health care to women. Findings suggest that
‘many negative health outcomes among female pa
tients may be accounted for, in part, by gender in-
equities in preventative care, diagnosis, and weat-
ment (Gallant, Coons, & Morokoff, 1994; Matthews
et al., 1997; Rossouw et al, 1995; Stanton & Gallant,
1995), As a result, multidisciplinary studies such as
the Women’s Health Initiative (see Matthews et al.,
3097; Rossouw et al., 3995; Sbumaker et al., 1998,
for background and specific information) have been
conducted to document gender differences within
the health-care system, investigate the effects of
‘National Cente for Postuaumasie Sess Disorder, VA Botton
Healtheare System, Boston. Massachusetts
preset aadress Depattinent of Psychology University of Rhode
Mand, Kingron, Rhode Ila
Department of Papen, Boston University School of Medicine,
orton, Massachusetts
To whom correspondence should be addressed st Nation
Centr for Posttraumatic tres Disorder (1168-2), 150 South
Huntington Avene, Boson, MasachisettsU2130emailine®
world atdcom.
tude; women’ health female patents
genderspecific treatment and practices on health-
related outcomes, and examine possible causes of
gender-based inequities (Bernstein & Kane, 1981;
Tobin et al, 1987; Van Wijk, Van Vilet, & Kolk,
1996),
‘Within the United States’ largest health-care sys-
tem, the Veterans Health Administration, gender is-
sues in health-care delivery are particularly germane
‘because the vast majority of its patients historically
hhave been and continue to be men. According to
Miller, King, Wolfe, and King (1999), Skinner and
Furey (1998), and Wolfe, Daley, Stern, and Wilson
(2996), among others, one possible explanation of
gender inequities in veterans’ health care is a sheer
lack of familiarity and experience with—and conse-
quent deficits in responsiveness to—female patients,
Perhaps as a result, recent findings indicate that one
in five women do 201 feel welcome receiving health-
care services within the Veterans Health Adminis-
tration (Kressin et al., 1999; Skinner, 2000). A lack
of familiarity with female patients on the part of
health-care workers is likely to become an even more
salient issue as projections suggest that the number of
‘Women veterans with access to veterans health-care
247
‘oxoamsaapcoant.6 20 Funan Pattie Capen248
services will increase by 40% from 1990 10 2020
(Furey, 2000; Klein, 2001).
‘Accordingly, the current study involved the de-
velopment of a reliable and valid measure of health-
care workers’ gender awareness, conceptualized it
terms of three subsidiary componenis: gender-role
“ideology, the extent to which a healthcare worker
does of does not rely on negative sereotypes about
female patients to make judgments about aspects of
their cares gender sensitivity, the degree to which =
health-care worker is aware of and sympathetic t0
the unique needs and requirements of female pa
tients; and knowledge, the extent to which a health-
care worker possesses accurate information regarding,
women veteransand general aspects oftheir ease. Our
model of gender awareness is an original conceptual-
ization (see Miller et al, 1999) based on an integration
‘and synthesis of important findings from a number of
disciplines (eg. health and social psychology, social
cognition, public health, medical sociology, and other
relevant medical esearch oth inside and outside the
veterans health-care system) related to quality health
‘care and disparities in the care of female and male
patients
Previous researchers (€., Bell, Micke, & Kasa,
1998; Gross, Zyzanski, Borawski, Cebul, & Stange,
1998: Hall, 'Irish, Roter, Ehrlich, & Miller, 1994,
Heesacker et al., 1999; Ross & Duff, 1983: Willer
& Grossman, 1995) have suggested that characteris
ties of the patient-provider relationship are associ-
ated wit important treatment judgments by health
care workers as wel a service utilization and patient
satisfaction. Moreover, characteristics of the patient
provider relationship predict differential ievelsot slf-
disclosure by the patient, differential adherence to
preseribed medical pracedures by the patient, and dif-
ferential diagnosis of symptoms by the provider (eg,
Bell etal, 1998: Floyd, 1997; Kerssens, Bensing, &
Andella, 1997; Roter & Hall, 1997). Within tis con-
text, a health-care worker's gender-role ideology may
bequite salient: thatis, ender inequities the health-
care system may be influenced by the health-care
worker's use of negative gender-hased stereotypes
Badgeret al 1999:Fowers, Applegate, Tredinnick, &
Slusher, 1996; Heesacker eta, 1999; Redman, Webb,
Hennrikus, Gordon, &: Sanson-Fisher, 1991; Turpin,
Darey, Weaver, & Kruse, 1992: Willer & Grossman,
1995)” Previous findings suggest that patient stereo-
types are typically more negative for female patients
than male patients, possibly due to @ more androcen-
tie focus within the medical profession (Heesacker
etal, 1999; Miller etal, 1999). Infact, the recognition
Salgado, Vogt, King, and King
of the role that negative stereotypes play in health-
caro service delivery has resulted in the development
of ethical guidelines to raise providers’ awareness of
the harmful effects of these judgments and biases in
the treatment of patients (American Psychological
Association, 1992; Heesacker et al., 1999; Miller etal.,
1999)
Another factor that may impact the quality of
‘are provided io female patients is the degree 1o which
health-care workers age sensitive to the unique needs
and requirements of female patients. Researchers
concerned with health-care utilization, treatment, and
‘outcomes have focused on the health-care system’s
lack of gender-specific sensitivity toward health-care
delivery for women (Falik & Collins, 1996; Rynne,
1985). As the majority of normative information
‘within the medical field (e.,, preventative practices,
reported symptomatology, and treatment strategies)
continues to depend on primarily male patient pop:
ulations, disparities can be seen in such phenomena
as the underdiagnosis of certain medical conditions
(eg, chronic heart disease) and less effective treat-
ments for female patients (Altman, 1991; Ayanian &
Epstein, 1991). Other gender sensitivity concerns are
sociocultural, for example, health-care workers’ need
to recognize women’s disproportionate burden for
family caregiving responsibilities (Miller etal, 1999)
‘and the consequences of lower socioeconomic status
‘on health-care seeking and health outcomes (Cotton,
1999; Weiss & Ashton, 1994)
ur third component of gender awareness is
knowledge. Knowiedge about a patient population—
in this ease, women Veterans—and available services
‘and special programs within the bealtb-care sysiem
should jogically lead to more effective patiemt assess-
‘ment and treatment and, hence, sounder health out
comes, Because women make up only 5% of veier-
fans with access to health-care services inthe Veterans
Health Administration, or approximately 1.4 million
of 2 total veteran population of 25.5 million, there
is an obvious lack of exposure to female patients. In
turn, there is likely 10 be fess Knowledge about the
services available to female patients on the part of
health-care workers, which necessarily may impact,
the quality of cate delivered (Cotton, 1999; “Hearing
‘on Departiaent of Veterans Affairs 2000),
‘The present psychometric study complements
existing research on gender inequities within the
health-care system by focusing op the health-care
staffs level of gender-ole ideology, gender sensitiv.
ity, and knowledge as they pertain to the delivery of
health care to women veterans. The developrieat ofGender Awareness Inventory-VA,
an instrument 1 assess staff gender awareness within
the Veterans Health Administration seems both ap-
propriate and timely given the increasing numbers of
‘women who are accessing this system for care and the
already identified problemsin the provision of care to
this population of patients. Although measures are al
ready available to assess more global gender-related
attitudes (e¢, Sex-Role Egalitarianism Scale [Beere,
King, Beere, & King, 984], Atitudes Toward Women
Scale |Spence, Helntrech, & Stepp, 1973), the Gen.
der Awareness Inventory-VA is intended to provide
' face and content valid measute of gender-related
construc that is specific to the health-care seting
Previous research suggests that general attitudes of
ten differ somewhat {rom attitudes about individu.
asin particular contexts or settings (Kleiapenaing &
Hagendoorn, 1991). Thus, thisinsirumentis unique in
ins ability to assess attitudes about female patients in
the heallh-care setting Another strength ofthe Gen
dex Awareness laventory-VA lis in its integration of
literature related to the care provided to female pa-
tients, As a result t should be @ valid measure of
characteristics that actuslly impact the provision of
care
“This five-part study was undertaken to develop
a reliable and valid measure of gender awareness in
the Veterans Health Administration. Part 1 of the
stucly was the instrument development phase; it ia
volved defining the gender awareness components in
detail and operationalizing these definitions via item
generation and refinement, Part 2 involved the de-
termination ofthe psychometric properties of the re-
fined instrument and an assessment of the intercor-
refations among its component measures, Patt 3 was
an examination of the evidence for convergent and
‘discriminant validity. Part 4 was an evaluation of ad-
ditional psychometric evidence for eriterion-related
validity. Finally, Pact S was an exploratory inquiry
into the factor structure thac underlies pattern of re-
sponses tothe three affective components of gender
awareness: genderrole ideology, gender Sensitivity,
and knowledge.
PART 1 INSTRUMENT DEVELOPMENT
Defining the Gender Awareness Components
We used a rational approach to test construe-
tion (Jackson, 1971; Nunnally, 1978). Our fust seep
in the development of the gender awareness mea-
sure was to define clearly the three core components
29
that comprise the global construct and elaborate on
their content domains. As noted pseviousy, our def
intoas of the ideology, sensitivity, and knowledge
components were drawn from literature in the areas
‘of health and socal psychology. social cognition, pub-
licealth,ané medica sociology regarding important
fectorsrelatedto gender disparities within healthcare
settings (see prior explication, Milicr ea, 1999), Fu
ther, these definitions were informed by a series of
structured interviews with health-care personnel om
a diverse range of services. The forme] definitions
follow.
Heatihcare-elated genderrole ideology repre-
sents a healthcare worker's attitude toward female
patients. It is the degree to which the health-care
worker does or doesnot employ negative stereotypes
about femele patients to make judgments about their
‘are, Health-care-telated gender ideotogy reflect im
plicit and explicit compatisons between the genders
regarding beliefs about essential differences between
ren and women (e8, intelligence, personality, and
‘emotionality). Health-care-elated gender ideology is
germane 10 a number of different arenas, including
judgments about the appropriateness or legitimacy
of female patients’ use of health services, and jad,
rents about the manner in which female patients
present themselves and communicate with health-
care workers
Health-care-related gender sensitivity i the de
gree to which a health-care worker is aware of and
sympathetic to the nceds or requiremeats of female
patients These needs may be anatomically based and
gender-specific. They may also arise from the current
sociocultural or economic contexts in which women
receive their health care and may concern the alioca-
tion of resources to female patients’ medical necds;
proper awareness of female-specific disease mani-
festation; privacy, personal safety, courtesy and re
spect in the health-care setting, of parenting and
other family demands that are placed on women in
ener
The knowledge component of gender awareness
is defined in terms ofthe extent to which the Veterans
Health Administration employee possesses accurate
information relatedtowomen's health-care needsand
the programs and services available to women veter-
ans This includes knowledge about the populations
‘of women currently and previously serving in the US.
‘Armed Forees and seeking healtt-care as military vet-
exans, eligibility requirements for routine and spe-
cial services, and health initiatives and programs for
women veterans.250
Item Generation and Refinement
Guided by the above definitions, and with on-
going reference to the literatures on female patients
health care in general and women veterans in partic-
ular, we generated an initial pool of items to reflect
the content of each of the three core components of
gender awareness. Ideology items focused mainly on
the assessment of attitudes and beliefs about essen-
differences between the genders as patients and
also explicit or implici comparisons between men and
Women within the health-care setting (Bernstein &
Kane, 1981; Lopez, 1989). Sensitivity items focused
‘on the awareness of gender-specific conditions and
tunderidentified conditions in female patients, the mi-
nority status of female patients within the Veterans
Health Administration, and sociocultural and eco-
nomic realities women face in accessing health care
‘Belle, 1990; Clancy & Massion, 1992; McBride, 1990;
O'Leary & Hegelson, 1997). Additional items were
included on topics that have been identified as rel
evant to female patients’ health care such as inter
personal violence and childhood abuse, The initial
item pool for the knowiedge dimension came from
factual information extracted primarily from recent
government documents about women veterans eligi
bility criteria for health-care services, available pro-
grams for women veterans, and their utilization statis+
tics (Miller etal, 1999).
‘Once the items were written, we progressed
through series of steps to ensure that they were clear,
unambiguous, and appropriate to the general popu:
lation of health-care workers. First, two research as-
sistants and five expert judges familiar with gender
issues in health care sorted each item into one of the
three defined categories and rated the items accord-
ing to how well they represented each component,
lems were revised or discarded if they were not at-
tributed to the target component or if there was some
level of disagreement as to how well the item was as-
sociated with the specific component. The readability
ofeach item was then assessed, and, where possible,
words, phrasing, and sentence length were changed
to simplify language. The readability level of know!
ledge items was somewhat higher than the readabil-
iny of items in the other two components, given that
some technical terms typically used in the health-
care setting were unavoidable. The Flesch-Kincaid
Grade Level score (Flesch, 1949) was 8.7 across all
item statements
Next, the list of candidate items was presented
to 19 health-care workers from a variety of job wypes
Salgado, Vogt, King, and King
(professional and nonprofessional) to ascertain clar-
ity of presentation, These individuals were asked 10
identify any words or terms they did not understand.
‘The items were reviewed and edited once again as,
necessary. In the end, this multistage item genera.
tion and refinement process resulted in a pool of
83 items: 24 ideology items, 22 sensitivity items, and
37 knowledge items. These items together comprised
‘our preliminary version of the Gender Awareness
Inventory-VA.
Initial Kem Analysis
Participants
‘A sample of 619 health-care workers was drawn
{com two lage Veterans Health Administration med-
ical centers in the noxtheast using proportional szat
ifed random sampling by ste, employee gender, and
employee's amount of patient contact. Patient eon-
tact was composed of three categories: direct patient
contact as a routine part ofthe job (eg. physicians,
nurses, social workers) direct patient contact but not
health-care contact as a part of the job (@.recep-
tionists admissions clerks); and no patient contct as
‘formal partof the job (eg ,payroll personnel, phys
ical plant). Although direct providers of health care
say play a particularly important role in shaping the
experience of female (and male) patients, employees
{rom across the spectrum of hospital services also con
tribute to the milieu and quality of the care environ-
rent (Miler etal, 1999) The overall response rate
across the two facilities was 60% for a total sample
size of 374 (60% worsen, 40% men). A summary of
respondent characteristis is shown in the fist col
umn of Table I. Although our participants were not
asked in this or other parts of the study to provide
information on race/ethnicity or identifying informs
tion including service and job itl, human resources
data indicate that the distribution of employees is ap-
proximately 63% White and 37% various minority
groups National figures for the distibusion of Vet
tans Health Administration employees with regard
to gender closely mirrored our sample characteristics
for each part of the study. In addition, our inclusion
ofemployeesin each ofthe three patient-contact cat-
egories likely ensured that our samples were quite
diverse in terms of social class
Sor addiona ables odin rom the present sudy
‘by type of patient contact, please conta Lyada King,Gender Awareness Inventory-VA.
ble Demographics and Respondent Caracas for APs of he Stu
251
Testromest’ Computation ol Convergent and Factor
evelopment pryehometrc_deeiminant_Catenon-eated analyte
Vase (earl) properties (Part2) vary Part3)_vahdty (Pat 4) structure (Par 5)
Ae aroup Ce I 3)
<0 0 @ 0) 1) 10)
2036 36 (10) 76) 908) 516) 210)
31-40 73.09) 65 (22) BOs), san 7
4-30 14037) 10936) 301) 3263) 1900)
516 3928) 92/30) 4) 205) 1513
0 236 (10) 716) 40) 607) 36)
Gender ne3Tl(%) 2 300(%) 14(%6) n= 80(%) w= 475(%)
Female 224 (ea) 176(39) 63) 5(@) 300(53)
Male 147 (60) mm4(81) aa) uO) 73)
Previows service inthe military n=305(%) ALIS) ne) ATE)
Yes 99,30) RGN 209) 1195)
No mc 6708) 3909)
Yes tization of VA mesh) ma26(%) wae I2%)
Health Care within
the post year
Yer 50154) 48049) 158) 15462) 6332)
No 8 (6) 2665) 16032) 98) 3848)
Patent contact n= 383 (%) 05 (%) 14%) maith) memaTS(%)
‘Contactar drat health are 213658) 365055) oo) *6(50) 254 65)
Contzet button dtect heath eae 67(18) 709) 36a) 2c)
No official patent catct 8324) 708) 2G) Be)
Length of employment 10369 (%) naLit() =I) ATIC
02 years au sea) 14) 50) 7010)
BSyeart S515) 003) 208) #709) a)
5-10 sears 935) mn) 310) 1102) 9600)
11S year, ous, aan ms) 157) 909)
133) 560) 290) 2) 1338)
Highest evel afedueaton nas) 16%) aS) ATT)
ft rade or leet. 10) 010) 10)
Some high schoo! 1 30) 8a)
High choo! praduate 30) roan 4610)
Some callge wy 180) C8)
‘Vocational ev techni taining, 703) 30) ssa
Four-year cllege graduate 2.08) 38020) 7509)
Some graduate or profesional school 518) 565) 2)
[Gracateo proesional degree 310) 2) 7009)
This eategory was adden pobsequentramples
Procedure
Allpartcipants received a packet that costained
a coves letter that explained the voluntary neture of
the task and the anonymity oftheir responses, are
sur postcard to track participation, and the prelim
inary version of the Gender Awareness Inventory-
VA. The questionnaire was divided into three main
secuons: demographics (incleding gender, ag, previ
‘ous military serece, Veterans Health Administration
health-care utilization, and level of education) and
smiployment information (including amount of pa
tient contact and length of employment) sensitivity
and ideology items, each employing a 5point Likert
response format, with possible responses from 1
strongly disagree to 5 = strongly agree, and knowl.
edge items in a standard multiple-choice format.
‘A multistage Mangione (1995) method was used
in this and all succeeding parts of the study to
maximize response rates while retaining anonymity
for respondents, First, an inital letter to introduce
the project was mailed approximately 1 week prior to
the mailing ofthe questionnaire itsell Next, the ques
tionnaire was mailed accompanied by a preaddressed
and number-coded tracking card fos the respondent
to mail separately from but simultaneous with the
‘questionnaire. If this response card was not recsived
within a 2-week period, individuals were contacted @252
third time via a reminder postcard, followed by a re-
‘peat mailing of the full questionnaire packet approx-
imately 2 weeks later. The data collection sequence
‘ended with the mailing of a final reminder postcard
to individuals who had still not returned a tracking
card.
Analyses
Various characteristics were computed for the
ideology and sensitivity items: frequency distribu
tions, means, standard deviations, and corrected item-
total correlations, that is, the correlation of each item
With the total of all other items within that compo:
neat measure, Items for which the item-total corre
lation was low (<30) were discarded, and coefficient
alphas were then calculated on the remaining item
sets Probabilities of endorsement of knowledge items
for percent correct, as well as the frequency af re-
sponses to each of the multiple-choice options, were
computed and reviewed. To select the best knowledge
items at this stage, particular attention was given 10
‘optimizing content validity, that is, the breadth and
representativeness of information about women vet-
erans and their health care. Similar to other know
edge or achievement-oriented test, this component
‘as intentionally multifaceted; it represents & variety
‘of content areas concerning characteristics of women
veterans, guidelines and eligibility requirements for
care, and specific services offered to female patients.
Therefore, internal consistency reliability estimates
were expected to be relatively low for the know!-
edge component in this and other parts of the study
(see Aiken, 1994; Anastasi, 1982; or Nunnally, 1978,
for further discussion of the preeminence of con:
tent validity over scale homogeneity in achievement
‘tiented testing).
Resultsand Discussion
In our intial analyses of data from the prelimi-
nary Gender Awareness Inventory-VA, the average
item-total correlations were .49 and .33 for ideology
and sensitivity items, respectively. After items with
low item-iotal correlations were discarded (values,
<.30), the internal consistency reliability estimate for
idcology was .88 (n = 339), and the internal consis
tency reliability estimate for sensitivity was 75 (n =
346). An additional nine items were developed for
the sensitivity measure at this point, :o improve its
reliability inthe next round of data collection, In ad-
Salgado, Vogt, King, and King
dition, knowledge items that were overly redundant
visei-vis their content were removed, The internal
‘consistency rliability estimate for knowledge was 67
(v= 371), and as expected, somewhat lower than it
‘was or the other components of gender awareness. At
the close of Part 1 of this study, the Gender Aware
ress Inventory-VA consisted of a total of 69 items:
20 ideology items, 29 sensitivity items (including the
‘now items), and 20 knowledge items Table Il shows
2 sample of five representative items from each of
the three component measures along with associated
means and standard deviations for the total sample
and stratified by gender ®
PART2, COMPUTATION OF
PSYCHOMETRIC PROPERTIES
In Part 2 of the study we used the revised version
of the Gender Awareness Inventory-VA. The objec:
tive was to derive new reliability estimates and other
scale and item properties on the refined component
measures and to examine their intercorrelations.
Method
Participants and Procedure
In this part of the study a sample of employees
‘was drawn from the same two Veterans Health Ad-
‘ministration medical centers, with appropriate strate-
gies employed to assure that no participants from
Part 1 were solicited for Part 2. Individuals were se-
lected using proportional stratified random sampling
bysite, gender, and type of patient contact. Of625 per-
sons who received questionnaire packets, 307 (59%
women, 41% men) returned the instrument, for a 49%
response rate. The second columa of Table I presents
demographic characteristics for these respondents,
Analyses
Estimates of internal consistency reliability for
the ideology and sensitivity measures were computed.
In addition, means, medians, ranges, and standard
deviations were calculated for all three measures
‘Next, independent samples ¢-tests were computed to
to request «copy ofthe complete Gender Awareness Inventory
Va, please contact Lynda King,Gender Awareness Inventory-VA. 253
“Table I, Sample Gender Awareness Inventary-VA Remt
Wore Me
ology items a) MD a MSD
‘Compared to men, women's physical complaint are or Ow Ta 093 ioe 1a? 105
‘ore likely o be eaused by mectal robles.
Women are no mire likely han en to use health ym az7 1st 13317 hsp ame gas Las
‘serves they don't realy need
(Compared to men, women more often gotothe doctor «30 «18101 175-65 O98 L9H LOL
for reasons that have ite to 0 mith eat
‘Women eset more to pain than men. soa 107 1 468 628 208-80
‘Women ae no more likly than mento ignore thsi 30% 136 1 345137 e Sas
doctors advice
Sensi test
Womenveteraasartishorptalshoudhaveaccewto «3014470455964 10H
‘care by experts in womer's health,
eis too expansive fr tis hoptal to provide health ms im 108 165103 18a
Services that ony women nee,
‘This bospical shouldbe more alert tosgnsofabuse at ——«303,=«SSMTOT TE GODS 3g)
home among female patients
spin souldofer Gulcatetomomerswhen ey «50H BUS 1287S 2H MaMa
‘ome fora medal vis
Tne VAsdouldnotbeempected oprovidespeciatheath 50ST 1001734]? LT
‘ecvce for women,
Knowledge items nm Cones % x Comet % on Comrect%
{Tine of there services rouioly avalae vo SSCS me
female VA patients? (a) abortions; ()ntertity
sercer (¢) menopause management (Wb and
2-Fliiity of VA healt Denes dove nordepend.om: = «HMA 1S wD
(2) come; () gender ()rciation expanse
() service connected dab
43 Today’ female soldiers have sutere cichood oT SD
ewlabige a ales the eilan rate (lower
than (0) about the sre (6) greater fan,
4 Women have servedin US Miliary efor offcally 5061777817:
sad unofcially in all of Ameriea's major coats
fine: (a) the American Revolution (B) the Civil War;
{@) World Wer (a) Woets War 1.
Since 1965,the proportion of women nthe milzry ««3W6_=«25D_—««HSS1TH:CMS DAS
ns (a) decrenae (0) increased (9 sla the a
TRespomse scaler 1 strongly doagree
srroniyegree
Emenee 2 (03 (4 (5.0)
test respondent gender differences for each compo:
neat of gender awareness, Finally, Pearson product
‘moment correlation coefficients were computed t0
index the associations amang scores on the three com-
ponents of gender awareness.
Reso and Discussion
‘The internal consistency reliability estimate for
the 20-item ideology measure was 87 (n = 286) with
an average item-total cortelation of $2. The internal
consistency reliability estimate for the 29item sen-
lsagree but not sronahy
T= no opinion orwndnles; A mapree bulnor rangi 5
sitivity measure was 83 (1 = 282) with an average
item-total correlation of 46. Thus, both measures
attained good levels of item homogeneity or preci
sion of measurement. The internal consistency reli
ability estimate for the 20-item knowledge measure
was .67 (n= 306) with an average item-total corre-
lation of 26. As discussed previously, this estimate
‘was expected to be somewhat lower than the other
gender awareness components given its multifaceted
content.
‘Table III contains descriptive statistics and inter-
correlations for the total sample as well as stratifiedSalgado, Vogt, Kin
ions Among the Thee Gender Awareness Componen!
Desriptiven Trtercorelationst
Man spi uaeeee 3
Toalampe
T Keeologs wo mm B 4137
2Semitwiy E195 20/30
Knowledge = 6
Women
1 Kéeoloey mms oe soryeraT 66
ZSenstvny 178115387 soar
Knowledge = 2).
Men
Teolgy 24 8 sere eps
2 Sensi mm ue arias
3. Knot 12 15
"Upcoecedcorelatonsae intedbeTor the sash and corrected onelatons are Tver he slash
“peat
by gender. The distributions of scores on ideology and
sensitivity are slightly negatively skewed, and most
respondents portray moderate to high levels of gen
der awareness. Yet there is a fair degeee of disper.
sion or individual differences in the distribation of
these scores, and a portion of respondents scared at
the lower end of each dimension. Regarding kaow!-
ledge, the average number of items correct was dose
tothe middle of the range of possible scores (12 out of
20, or approximately 60%). Also, the range and stax-
ard deviation suggest that this measure is abie to
capture successfully individual differences in know-
edge about women veterans and their healthcare.
‘Results from the independent samples -tsts re
vealed that female health-care workers scored sig-
nificantly higher on both the ideology and sensiiv~
ity components of gender awareness, 1(296) = ~3.70,
p< 05, and 1(296) = -1.99, p «05, respectively,
‘whereas no difference was found on the knowledge
‘component, (298) = 1.20, ns. These findings suggest
that female health-care workers tend to have more
favorable attitudes and beliefs about female patients
and are also more sensitive to their unique needs than
male workers in the Veterans Health Administration
Setting. This ideology finding is consistent with other
research on gender-role stereotyping, which suggests
that women possesssomewhat more postive attitudes,
‘on gender issucs (King & King. 1997)
‘As shown in Table III, the measures of ide
ology, sensitivity, and knowledge were modestly to
moderately correlated, as indicated by the corrected
correlations. The two more affective dimensions (ide-
‘logy and sensitivity), which shared a common item
format, were most strongly related for the total
sample (.62), and were strongly correlated for women
(61) and men (61) separately as well, Allcorrelations
are sufficiently low to support the argument that each
of the gender awareness components are distinct from
fone another, elthough they share a common theme,
and hence contribute uniquely tothe global construct.
PART 3, CONVERGENT AND
DISCRIMINANT VALIDITY
In Part 3 we explored evidence for conver.
gent and discriminant validity by examining hypoth-
esized patterns of expected relationships between
scores on each of the three gender awareness mea-
sures and scores on two prominent instruments that
assess gender-role ideology in a broad sense (the
SowRole Egalitarianism Seale and Attitudes To-
ward Women Scale), a measure of sociat desirabil-
lay (the Marlowe-Crowne Social Desirability Scale),
and a measure of negative affect (from the Positive
AlfectiNegative Affect Schedule). It was predicted
that these individuals who scored higher on the Gen-
Ger Awaieness Inventory-VA would be more cgali-
tarlan in their atitudes toward women and men (as
shown by the Sex-Role Egalitarianism Seale) and
have more positive attitudes toward women (as shown
by the Attitudes Toward Women Scale) than wouldin-
dividuals who scored lower on the Gender Awareness
Inventory-VA; this outcome would provide evidence
for convergent validity. It was also predicted that gen-
derawareness would not be significantly influenced by
social desirability or negative affectivity, which would
provide support for iseriminant validity.‘Gender Awareness Inventary-VA.
Method
Pantciponts and Procedure
Assubset (1 = 114;59% women, 41% men)ofthe
participants in Part 2 was administered the additional
‘measures of gender-role ideology, social desirability,
‘and negative affectivity. Information in the third col
‘umn of Table { profiles this group. To ease respondent
burden for many of our participants and to enhance
response rates, different combinations ofthese vali
ity measures were randomly assigned for inclusion in
the main questionnaire packet: thus, these individu-
als received a questionnaire packet that contained the
Gender Awareness Inventory-VA and 2 counterbal-
anced combination of the other measures.
Measures
‘The Sex-Role Egalitarianism Scale—Form KK
(King & King, 1990, 1993) is a 25-item gender-role
ideology measure (a = 93) that assesses attitudes to
ward the equality of women and men across content
domains that represent marital, parental, employ:
‘ment, Social-interpersonal-heterosexual, and educa-
tional roles. Respondents are asked to judge the ex-
tent of their agreement or disagreement with each
statement using a S-point, 1 = srongly disagree 10
= strongly agree, Likest response format. A total
score is computed as the sum across all items, with
higher values indicating a more egalitarian attitude.
Sample items include “Keeping track of a child’ ac
tivities should be mostly the mother’s task [reverse-
scored],”* Women can handle job pressures as well as
men can,” and “A worsan should have as much right
to ask a man for a date as 2 man has to ask a woman
fora date”
‘We used the 25-item short form (a = 86) of the
Attitudes Toward Women Scale (Spence, Helmreich,
& Stapp, 1973) to assess participants’ attitudes regard
ing women's roles in society, Each item statement on
this instrument was judged by means of a 4-point
Likert response format, with 1 = disagree strongly
and 4 = agree strongly. A total score across items was
derived; higher scores indicated more positive atti=
tudes toward women, Sample items are “A woman
should not expect to go to exactly the same places or
to have quite the same freedom of action as a man
[reverse-scored}," “In general, the father should have
‘greater authority than the mother in the bringing up
of children freverse-scored],” and “There should be a
255
strict merit system in job appointment and promotion
without regard to sex”
‘The Marlowe-Crowne Social Desirability Scale
(Crowne & Marlowe, 1964) is a 33item instrument
{a = 84) that measures an individual's tendency to
Sescribe himsel herself favorably or i a socially de-
sirable manner. Scores were computed asthe number
of items responses in the Keyed direction based on a
‘nvelfese format, Higher seores indicate more social
desirability. Sample items are “Thave never intensely
disliked anyone T sometimes feel resentful when I
don't get my way” and “There have been occasions
when I took advantege of someone [reverse-sored]”
‘The index of negative affectvity was taken
fiom the Positive AflecuNegative Affect Schedule
(Watson, Clack, & Tellegen, 1988), a measure of
‘ood stes The negative afectivty subscale consists
Of 10 items (« = 5) such as “distressed” “guilty.”
and “irritable.” which are judged on a S-point Likert
scale that ranges from 1 = very slightly or nota all
105 = extremely
Analyses
‘Fo establish evidence for convergent and dis-
criminant validity, Pearson product-moment corre-
lations were computed between scores on each of
the three Gender Awareness Inventory-VA measures
and scores on the four measures described immedi-
ately above. In addition, corrections for attenuation
clue to unreliability in measurement were applied to
the correlations. Asin Part2, tests for respondent gen
der differences were computed for each component
of the Gender Awareness Inventory-VA.
Results and Discussion
Table IV presents the results of the correlational
sunalyses for the cotal sample and stratified by gender.
For the total semple, each of the gender awareness
components was significantly correlated with scores
on both the Sex-Role Egalitarianism Seale (range
32-58 for uncortected correlations, range = 5-5
for corrected correlations) and the Attitudes Toward
Women Scale (range = .33-.65 for uncorrected cor
relations, range = .52~.76 for correcied correlations).
‘Moreover, asexpected, the largest correlations within
‘each set were found for the associations between mea-
sures oflike constructs, thats, gender role ideology as
assessed by the Gender Awareness Inventory-VA and
the two more general gender-role ideology measures,
the Sex-Role Egalitarianism Scale and the Attitudes256
Salgado, Vogt, King, and King
“Table IV. Correlations Berween Scureson the Gender Awareney ventory. VA and Other Measures
RES AWS MCSD PANAS Negative atleetviy
‘Toatample
Tdeslogy 119. SBS" 85 6ST. T6 102 ~-08/—-10 1H 1/08
Sensitivity 13 “aor/as* 95 Serysst 102 —0/—04 He a7
Knowledge 114 32/49" 96 3B/S0 05 “aT/-27 tS iny=a8
Women
Ideology 6 aBISY 7 SLO ot a
Sensi 66 yA ST anys ot 81 16/39
Knowledge 67 alta" 58 Soya 2 a aynas
Mea
Ideology er Bas 4 a -24)—28
Senstivty se Joys 4 a 13/43
Knowledge ae ary7s & 3
Now SRES= rm Seale AWS = Atitudce Toward Women Sal; MCSD
ity Scale: PANAS = Postve AfectNegntie Affect Schedule. Uncorrected
‘correlations at ted before the sh and corrected correlations se listed ater the ash
“p< 05
‘Toward Women Scale. With the exception ofthe neg:
lgible correlations between the sensitviy measure
and the Sex-Role Egalitariaism Seale for women,
the coztelaions computed separately for women and
men were consistent with those computed forthe to
tal sample. The low cortelations between sensitivity
and Sex-Role Egalitaranism Seale scores for women
ray recapitulate previously documented differences
Jn what is measured by the Sex Role Bgalitarianism
Scale vis--vis what is measured by the Attitudes To-
ward Women Scale (see King & King, 1993, 1997)
With regard to discriminant validity, the associations
between the genderawarenessmeasures and the mea-
sures of social desirability and negative aifectivity
were negligible and nonsignificant for the total sample
aswellas for womenand men separately. This suggests
that scores on the Gender Awareness Inventory-VA,
are not influences by the tendency to present one
self in a postive light nor by one's negative mood
state. Overall, the pattern of relaiorshigs shown
in Table JV support both convergent and disrio:
nant validity, a8 hypothesized. There were no signifi
cant differences between male and female health-care
workers on the ideology, sensitivity, and knowledge
components, (111) = —1.29, ns; (111) = ~0.24, ns
and (12) = 1.72, ne: respectively
PART 4. CRITERION-RELATED VALIDITY
In order to examine evidence for criterion.
related validity in Part 4, a series of vignettes
Were constructed to assess health-care judgments
related to female patients. It was predicted that
those individuals who scored higher on the Gender
Awareness Inventory-VA would provide more posi-
tive health-care related judgments of female patients
than would individuals who scored lower on the Gen:
der Awareness Inventory-VA.
Method
Participants and Procedure
A sample of 281 health-care workers were
drawn from two additional Veterans Health Admin-
istration facilities using proportional stratified ran-
dom sampling by site, employee gender, and em-
ployee’s amount of patient contact. Of those sampled,
147 completed the Gender Awareness Inventory-VA,
252% response rate. Respondents later received a
vignette measure; 92 (63%) of these were completed.
(Characteristics of those who returned both the G
der Awareness Inventory-VA and the vignette mea-
sures (V = 90, 62% women, 38% men) are shown in
the fourth coluron of Tabie 1
Because wo waves of data collection were neces-
sary for this part of the study, an abridged Margione
method (1995) was used to reduce participant bur-
den. The frst wave of data collection included (a) an
initial Jetter to introduce the project and provide
information about the two tasks that would follow,
(b) the Gender Awareness Inventory-VA packet, and.
(e) a teminder postcard. The second wave of dataGender Awareness Inventory-VA.
collection occurred approximately 3 months after the
administration of the Gender Awareness Inventory
VA, and included (a) an inital letter to reintroduce
the’ project and provide information about the vi-
gnetie component, (b) the vignette component itsel,
and (c) a reminder postcard. Although participant
numbers were assigned to link responses across the
‘wo measures, care was taken to ensure both the con-
fidentialty and anonymity of responses.
Measures
Inaddltion to the Gender Awareness Inventory.
‘VA. each participant received a vignette measure that
included five short scenarios. Four of these required
respondents to make judgments relevant tothe cate
provided to female patients, and the fifth was a fol
Vignettes were developed on the basis of their impi-
cations forthe care provided to female patients,
“The first of these vignettes required respon:
dents to judge the appropriateness of a Veterans
Health Administration committees decision not 19
take disciplinary action against a health-care worker
‘who revealed confidential information from a female
patients medical records. The second vignette re-
{quired respondents to judge the appropriateness of
4 doctor’ decision not to address privacy concerns
raised by a female patient. Both of these vignettes
pertain tothe important issue of the extent ro which
female patients are provided with an appropriate evel
of privacy within the Veterans Health Administra
tion setting. Although there have been impressive im-
provements in the facilitation of privacy to women
veterans, probleme still remain in this area (General
Accounting Office, 1999). Each ofthese vigneitesre-
quired respondents to provide ratings on a sale rang-
ing from 1 = strongly disagree to 7 = strongly agree.
‘The third vignette required respondents to make
Judgments of the importance of programs and set
vices that specifically benefit female patients com-
pared with other, mote general, programs and ser
Vices proposed for a hypothetical veterans hospital
Clearly, optima} pealth-care occurs when needed pro-
‘grams and services are available to female patients
“Tis vignette requited respondents to choose one of
two options fom & total of six pais of programs
and services. Four of the pairs contained an option
that specifically benefited female patients (eg. hie
ing health-care providers who are expes's in women's
health cafe) contrasted with a more general pro-
tram or service (€. educating patients about teat-
257
‘ments and programs offered by the Veterans Health
‘Administcation); the remaining to pairs served as
foils After examining trequenties of endorsement
and intercortelations among scores onthe service and.
‘program pairs, one pair of program and services was
Aetermined 0 be inappropriate due toits inability to
discriminate among participants (Le, slmost all p
ticipants selected one of the options) and was elimi-
naied from further analyses.
‘A fourth vignette cequired respondents to rate
the appropriateness of an employee's decision to di
rect 0 distraught patient to the psychiatry service,
based on the assumption thatthe patient's problems
are psychiatric in nature, Halt of the respondents e-
ceived a vignette in which the patient was male and
the other half ofthe respondents received a vignette
in which the pauient was female, Past research (¢
Turpin, Darcy, Weaver. & Kruse, 1992) has suggested
that healthcare workers tend to interpret women's
health problems 28 mote likely than men’s health
problems to originate from psychitsc causes, and
that this tendency has negative implications for the
care provided to female patients.
Analyses
‘We first examined the responses to each of the
‘ocal vignettes with attention to descriptive statistics,
‘most notably frequency distributions ofthe responses
to each scenario. We then calculated bivasiate corre-
lations between scores on the three measures of the
Gender Awareness Inventory-VA and scores on the
vignette messures to examine the eriterion-related
validity of the Gender Awareness Inventory-VA. Fi-
nally, tests for respondent gender differences were
‘computed for eech component of the Gender Aware-
‘ness laventory-VA.
Results and Discussion
Although participants generally tended toward a
‘more “gender-aware” perspective in their responses
to the vignettes, there were individual differences in
judgments related to the health care of female pa-
tients. Responses to the vignettes that assessed judg-
ents of confidentiality and privacy revealed the least
variance; many healin-care workers indicated that vi
lations of these standards of care are not acceptable.
About two thirds of the respondents (67%) strongly
disagreed with the hypothetical committee’s decision
not to take disciplinary action against a seceptionist’s
violation of patient confidentiality, and an even larger258
number ofthe respondents (88%) strongly disagreed
with the hypothetiea! doctor's decision not to address
4 woman veteran’s privacy coneemns. On the other
hand, support for programs and services that specif.
ically benefit women was weaker. Although slightly
more than one third ofthe respondents (35%) en-
dorsed at least one gender-specific program or se
vice, nearly two thirds of she respondents (64%) dia
not endorse even one program or service that speci
icaly benefits female patients. With regard to the
psychiatic judgments, although nearly half of the
respondents (49%) strongly disagreed that it was ap
propriate to assign psychiatric causes to the health
problem of the female patient in the vignette, 25%
‘endorsed a possible psychiatric origin tothe woman's
health problem. Iti interesting to note tht a simi-
Jar pattern was obtained in the responses to the by-
pothetical male veteran; there was litle difference
In the average agreement scores between those who
judged the woman and those who judged the man,
1(89) = ~0.32, ns. It is possible that veterans, both
women and men, are seen by health-care stati as a
patient population that is likely 1o have health prob-
lems that originate in psychiatric eauses.
Inthe next set of analyses, we examined the 1e-
lationship between scores on the Gender Awareness
Inventory. VA and the vignetic judgments for the to-
tal sample and for men and women separately. We
first corzelated scores on the confidentiality and pri-
vacy vignettes with the three gender awareness com-
ponents While gender-sole ideology was unrelated to
confidentiality judgments for he sample as a whole,
r= ~07, and for women, r = .05, gender-role ide-
logy was related to confidentiality judgements for
ale tespondents, r = —42, p< 1, More specie
cally, men who had less positive autudes and be-
liefs about fermale patients were more likely to agree
with a hypothetical committe’s decision not 10 Lake
action against a health-care employee who revealed
‘confidential information from a female ptients med-
ical records On the other hand, gender-role ideol-
‘ey was significantly elated to privacy judgmeats for
the sample as a whole,r = ~.23, p < 65.for women,
1 = —25, p= 05, and was in the expected direc.
tion for men, although it did rot reach a conven-
tionat level of significance, r = ~.18. Thus, men and
women (Laken together) who fad less positive ati-
tudes and beliefs about female patients were more
likely to agree with a doctor's decision not to adress
2 female patients privacy concerns. Theze was also
a trend for knowledge to be related to privacy judg-
tmenis,r = -.17, p < 10, for the sample as @ Whole,
Salgado, Vogt, King, and King
which suggests that individuals who were less know!-
edgeable about women veterans were slightly more
likely to agree with @ doctor's decision not to ad-
dress a fomale patient's privacy concerns. The cor-
relations between knowledge and privacy judements
were similar for men and women when computed
separately, r = ~.18 and —20 respectively, although
these correlations did not meet conventional levels of
significance,
‘We then computed a summed score of the num-
beer of endorsed programs and services that benefit
female patients and correlated this Score with scores
ton each of the gender awareness measures. Findings
revealed a significant relationship between sensitiv.
ity and endorsement of progrants and services that
benefit female patients for the sample as a winole,
= 27, p <.05,anda trend in this direstion for both
women and men, r= 25, p< 10;r= 29, p= 10,
respectively. These findings suggest that individuals,
‘who were more sensitive to the needs of female pa-
tients were more likely to support initiatives that bet
efit female patients, Finally, we correlated psychiatric
judgments of the female patient with scores on each
‘ofthe three gender awareness components for the to-
tal sample. There was no relationship between any
of the gender awareness components and judgments
of the appropriateness of referring a distressed fe-
‘male patient to the psychiatry service for the sample
as 2 whole or for women and men separately. There
was, however, a marginally significant negative corre
Iation between psychiatric judgments and knowledge
for women, r = ~38, p = .07. This finding, although
based on a fairly small sample size (n = 24), suggests,
that more knowledgable female health-care staff were
less likely to agree with the employee's judament that
the female patients’ problem was psychiatricin nature
than female health-care staff who were less knowl
edgeable about women veterans and their health-care
needs.
‘Taken altogether, these findings provide prelim-
inary support for the criterion-related validity of the
Gender Awareness Inventory-VA. and suggest that
scores on this measure may have significance for a
ange ofjudgments related to the health care provided
to women veterans. It might be noted that there was a
3.month interval between the administration of the
Gender Awareness Inventory-VA and the vignette
judgments, which provided a rather conservative test
‘of validity, as the size of the computed correlations
‘was likely to attenuate with the passage of time,
‘The independent samples ‘tests once again
yielded no significam differences between male and‘Gender Awareness Inventory-VA.
female health-care workers on the ideology, sensi-
tivity, and knowledge components, (89) = 1.6, ns;
1(89) = 0.01, ns; and #(89) = 1-72, ns, respectively
PARTS. FACTOR STRUCTURE
Part § was an inital examination of the factor
structure of the Gender Awareness Inventory-VA.
Method
Pantcipants and Procedure
For this part of the study, we used the responses
from 481 individuals (63% women, 37% men) who
completed the Gender Awareness Taventory-VA, ei-
‘ther as participantsin Parts?,3,or or ascontributors
to anormative evaluation of gender awareness among
health-care workers (Vogt et al, 2001). The fifth co:
‘umn of Table [ presents demographic characteristics
{or these respondents,
Anatyses
We used Browne, Cudeck, Tateneni, and Mels
(1998) Comprettensive Exploratory Facior Anly-
sis (CEFA) procedures and software. Unlike more
conventional exploratory factor analysis programs,
‘CEFA provides extensive statistical information con-
cerning the viability of the solution, including a dis-
‘crepancy index or chi-square statistic, the root mean
square error of approximation (RMSEA; Steiger,
1990), and standard errors for the factor loadings in
the rotated solution. The maximum likelihood est
mator was employed; three factors were initially ex-
tracted; and the Solution was rotated t0 an oblique
criterion in which the 20 ideology items were targeted
tooad on one factor, the 29 sensitivity tems were tar-
geted toload on another factor, and the 20 knowledge
inems were targeted to load on a final factor. As pre
scribed by Browne et al, the matrix of correlations
among items was analyzed, Please note that results
fare not reported separately for men and women, a5,
the available number of cases did not warrant sepa:
rate factor analyses for these subgroups.
Results and Discussion
The three-factor solution that coincided with
the throe components of gender awareness yielded
259
reasonably good fit to the data, x2(2142, N= 481) =
4086.50, p <.01, with RMSEA = 0.043, and its
‘9% confidence interval = 0.041-0.046, Browne and
‘Cudeck (1993) judged RMSEA values in the range of
(0.05 or lower to indicate “close fi." With regard tothe
loadings for the rotated solution, 15 ofthe 20 items
proposed to load on the ideology factor had critical
ratios that exceeded 2.00 and 90% confidence inter
vals that didnot include 0;24 ofthe 29 items proposed
toload on the sensitivity factor had critica ratios that
exceeded 2.00 and 26 of 29 had 90% confidence in-
tervals that did not include O; and 15 of the 20 items
proposed to load on the knowledge factor had criti
‘al ratios that exceeded 2.00 and 90% confidence in
tervals that did not include 0. Thus, in general, there
eassupportforastructure that differentistes between
{he three components, roughly indexed Dy the sets of
items designated to load on each factor.
Of the three factors, the more interpretable ap-
peared to be gender-role ideology. For this factor, 15,
‘of the 20 items constructed to be indicators of ideo!
‘ogy had loadings greater than 0,40 (average of 0.59),
and forall items the cross-ioadings on gender sensi-
tivity and knowledge were quite low, with all but two
hhaving values less than 0.15 for sensitivity and all but
five having values less than 0.15 for knowledge, As
expected, the loadings were considerably lower for
knowledge, a possible function of the restricted renee
introduced by the dichotomous scoring ofthese items.
For this factor, 10 of the 20ivems constructed to be in-
dicators of knowledge had Joadings greater than 0.25,
(average of 0.34), and forall items, the cross-loadings
fon gender sensitivity and gender-role ideology were
quite low, with all but three having values less than
0.10 for gender sensitivity and all but one having val
ues less than 0.10 for gender roe ideology.
‘The strength and pattern of loadings for gender
sensitivity, however, were much morecomplex. Asub-
set of nine items with strong loadings on sensitivity
refers to empathy and concern for women as care-
aivers (eg. “Hospitals should have flexible hours to
rect the needs of mothers,” “Women who bring chi
dren with them to appointments disrupt hospital rou-
tine.” [reverse-scored]); these items had loadings that
ranged from 0.20 00.83 (mean = 0.57). The loadings
were weaker on the remaining 20 items; they ranged
from 0.04 to 033 (average of only 0.17), and a few
had rather strong associations with the ideology fac:
tor(eg,,"Women complain too much abouthow hard
itis to get to the hospital” [reverse-seored], 0.53 with
ideology, “Women who tell their doctors about abuse
at home often make it sound worse than it realy is”260
{peverse-scored], 0.56 with ideology, and “Women ask
for to0 many extra police and safety services in this
hospital” [reverse-scored], 0.61 with ideology). The
‘correlation between the ideology and sensitivity fac-
tors was 030, sll indicative of the discriminant vali
ity ofthese two affective components