HIV Knowledge Questionnaire

HIV-KQ-18

(18 item version)

Carey, M. P., & Schroder, K. E. E. (2002).

Development and psychometric evaluation of
the brief HIV knowledge questionnaire
(HIV-KQ-18). AIDS Education and
Prevention, 14, 174-184.
 HIV-KQ-18

For each statement, please circle True (T), False (F), or I dont know (DK). If you do not know, please
do not guess; instead, please circle DK.

True False I dont

know

1. Coughing and sneezing DO NOT spread HIV. T F DK

2. A person can get HIV by sharing a glass of water with

someone who has HIV. T F DK

3. Pulling out the penis before a man climaxes/cums keeps

a woman from getting HIV during sex. T F DK

4. A woman can get HIV if she has anal sex with a man. T F DK

5. Showering, or washing ones genitals/private parts,

after sex keeps a person from getting HIV. T F DK

6. All pregnant women infected with HIV will have babies

born with AIDS. T F DK

7. People who have been infected with HIV quickly show

serious signs of being infected. T F DK

8. There is a vaccine that can stop adults from getting HIV. T F DK

9. People are likely to get HIV by deep kissing, putting their

tongue in their partners mouth, if their partner has HIV. T F DK

10. A woman cannot get HIV if she has sex during her period. T F DK

11. There is a female condom that can help decrease a womans

chance of getting HIV. T F DK

12. A natural skin condom works better against HIV than does
a latex condom. T F DK

13. A person will NOT get HIV if she or he is taking antibiotics. T F DK

14. Having sex with more than one partner can increase a
persons chance of being infected with HIV. T F DK

15. Taking a test for HIV one week after having sex will tell a
person if she or he has HIV. T F DK

16. A person can get HIV by sitting in a hot tub or a swimming

pool with a person who has HIV. T F DK

17. A person can get HIV from oral sex. T F DK

18. Using Vaseline or baby oil with condoms lowers the

chance of getting HIV. T F DK
 Answer Key
HIV KQ 18

1 TRUE
2 FALSE
3 FALSE
4 TRUE
5 FALSE
6 FALSE
7 FALSE
8 FALSE
9 FALSE
10 FALSE
11 TRUE
12 FALSE
13 FALSE
14 TRUE
15 FALSE
16 FALSE
17 TRUE
18 FALSE
Original
1. Coughing and sneezing DO NOT spread
HIV
2. A person can get HIV by sharing a glass of
water with
someone who has HIV
3.Pulling out the penis before a man
climaxes/cums keeps
a woman from getting HIV during sex
4. A woman can get HIV if she has anal sex
with a man
5. Showering, or washing ones genitals/private
parts, after sex keeps a person from getting HIV
6. All pregnant women infected with HIV will
have babies
born with AIDS
7. People who have been infected with HIV
quickly show
serious signs of being infected
8. There is a vaccine that can stop adults from
getting HIV
9. People are likely to get HIV by deep kissing,
putting their
tongue in their partners mouth, if their partner
has HIV
10. A woman cannot get HIV if she has sex
during her period
11. There is a female condom that can help
decrease a womans
chance of getting HIV
12. A natural skin condom works better against
HIV than does
a latex condom.
13. A person will NOT get HIV if she or he is
taking antibiotics.
14. Having sex with more than one partner can
increase a
persons chance of being infected with HIV
15. Taking a test for HIV one week after having
sex will tell a
person if she or he has HIV
16. A person can get HIV by sitting in a hot tub
or a swimming
pool with a person who has HIV
17. A person can get HIV from oral sex
18. Using Vaseline or baby oil with condoms
lowers the chance of getting HIV
Spanish Translation
1. El toser o estornudar NO disemina el contagio de
VIH
2 .Una persona puede contagiarse con VIH si
comparte un vaso de agua que ha utilizado una
persona infectada con VIH
3. Sacar el pene antes de que el varn eyacule evita
que la mujer se contagie con el virus de VIH
durante las relaciones sexuales.
4. Una mujer puede infectarse con el VIH, si tiene
Relaciones sexuales anales (sexo anal) con un
hombre
5. Ducharse o lavarse los genitals/ partes privadas
luego de una relacin sexual evita que la persona se
infecte con VIH.
6. Toda mujer embarazada, infectada con VIH,
tendr nios infectados con VIH.
7. Personas que estn infectadas con el VIH,
Rpidamente muestran los signos o sntomas de
que est contagiado (a)
8. Existe una vacuna que evita que la persona adulta
se infecte con el VIH
9. Las personas son ms propensas a infectarse
cuando se dan besos profundos (French kiss), o
colocan sus lenguas dentro de la boca de sus
parejas que est infectada con VIH
10. Una mujer no puede contagiarse con VIH si tiene
relaciones sexuales durante su menstruacin
11. Existe un condn femenino que ayuda a reducer
el riesgo de infectarse con VIH
12. Un condn de piel es major para prevenir el
contagio de VIH que un condn de latex.
13. Una persona que est tomando antibiticos no se
contagiar con VIH si tiene relaciones sexuales
14. Tener relaciones sexuales con ms de una pareja
sexual aumenta la probalilidad de que se infecte
con VIH
15. Realizarse una prueba para determinar si se esta
infectado con VIH una semana despus de haber
tenido relaciones sexuales, puede determinar si
la persona est infectada o no
16. Una persona puede infectarse con VIH
compartiendo un bao termal (yacuzzi) o una
piscina con una persona infectada con el VIH.
17. Una persona puede contagiarse con VIH cuando
sostiene relaciones sexuales de forma oral
18. Utilizar vaselina o aceite para beb con los
condones reduce el riesgo de infectarse con VIH.
Back Translation
1. Coughing or sneezing do not spread HIV
2. A person can get HIV sharing a glass of water with
a person who has HIV
3. Pulling out the penis before a man ejaculates
climax/cums keeps the woman from getting HIV
during sexual intercourse
4. A woman can get HIV if she has anal sex with a man
5. Showering or washing genitals after having sex keep
the person from getting HIV
6. All pregnant women with HIV, will have children
with HIV
7. People who are infected with HIV, quickly show the
signs or symptoms of being infected
8. There is a vaccine that avoids adult from getting
HIV
9.People are more likely to get HIV by deep kissing
putting his/her tongue inside their partners infected
with HIV
10. A woman can not HIV if she has sex during her
menses (period)
11. There is a female condom which can help reduce a
woman change of getting HIV
12. A swing condom is better preventing tha HIV
contagion than the latex condom
13. A person who is taking antibiotics will not get
infected with HIV if he/she has sex.
14. Having sex with more than one partner increases
the changes of getting infected with HIV
15. Taking a test for HIV a week after having sex
(sexual relations) will tell a person if he/she has HIV
16. A person can get HIV by sharing a hot tub or a
swimming pool with a person infected with HIV
17. A person can get HIV from having oral sex
18. The use of vaseline or baby oil with condoms
reduces the risk of getting HIV
 HIV Knowledge Questionnaire (HIV-K-Q)

(45 item version)

Carey, M. P., Morrison-Beedy, D., & Johnson, B.

T. (1997). The HIV-Knowledge
Questionnaire: Development and evaluation
of a reliable, valid, and practical self-
administered questionnaire. AIDS and
Behavior, 1, 61-74.
 HIV-KQ-45

For each statement, please circle True (T), False (F), or I Dont Know (DK). If you do not know,
please do not guess; instead, please circle DK.

True False Dont

Know

1. HIV and AIDS are the same thing. T F DK

2. There is a cure for AIDS. T F DK

3. A person can get HIV from a toilet seat. T F DK

4. Coughing and sneezing DO NOT spread HIV. T F DK

5. HIV can be spread by mosquitoes. T F DK

6. AIDS is the cause of HIV. T F DK

7. A person can get HIV by sharing a glass of water with

someone who has HIV. T F DK

8. HIV is killed by bleach. T F DK

9. It is possible to get HIV when a person gets a tattoo. T F DK

10. A pregnant woman with HIV can give the virus to her
unborn baby. T F DK

11. Pulling out the penis before a man climaxes/cums keeps

a woman from getting HIV during sex. T F DK

12. A woman can get HIV if she has anal sex with a man. T F DK

13. Showering, or washing ones genitals/private parts,

after sex keeps a person from getting HIV. T F DK

14. Eating healthy foods can keep a person from getting HIV. T F DK

15. All pregnant women infected with HIV will have babies
born with AIDS. T F DK

16. Using a latex condom or rubber can lower a persons

chance of getting HIV. T F DK
 True False Dont
Know

17. A person with HIV can look and feel healthy. T F DK

18. People who have been infected with HIV quickly show
serious signs of being infected. T F DK

19. A person can be infected with HIV for 5 years or more

without getting AIDS. T F DK

20. There is a vaccine that can stop adults from getting HIV. T F DK

21. Some drugs have been made for the treatment of AIDS. T F DK

22. Women are always tested for HIV during their pap smears. T F DK

23. A person cannot get HIV by having

oral sex, mouth-to-penis, with a man who has HIV. T F DK

24. A person can get HIV even if she or he has sex with
another person only one time. T F DK

25. Using a lambskin condom or rubber is the best protection

against HIV. T F DK

26. People are likely to get HIV by deep kissing, putting their
tongue in their partners mouth, if their partner has HIV. T F DK

27. A person can get HIV by giving blood. T F DK

28. A woman cannot get HIV if she has sex during her period. T F DK

29. You can usually tell if someone has HIV by looking at them. T F DK

30. There is a female condom that can help decrease a womans

chance of getting HIV. T F DK

31. A natural skin condom works better against HIV than does
a latex condom. T F DK

32. A person will NOT get HIV if she or he is taking antibiotics. T F DK

33. Having sex with more than one partner can increase a
persons chance of being infected with HIV. T F DK
 True False Dont
Know

34. Taking a test for HIV one week after having sex will tell a
person if she or he has HIV. T F DK

35. A person can get HIV by sitting in a hot tub or a swimming

pool with a person who has HIV. T F DK

36. A person can get HIV through contact with saliva, tears,
sweat, or urine. T F DK

37. A person can get HIV from a

womans vaginal secretions/wetness from her vagina. T F DK

38. A person can get HIV if having

oral sex, mouth on vagina, with a woman. T F DK

39. If a person tests positive for HIV, then the test site will
have to tell all of his or her partners. T F DK

40. Using Vaseline or baby oil with condoms lowers the

chance of getting HIV. T F DK

41. Washing drug use equipment/works with

cold water kills HIV. T F DK

42. A woman can get HIV if she has vaginal sex with a
man who has HIV. T F DK

43. Athletes who share needles when using steroids can

get HIV from the needles. T F DK

44. Douching after sex will keep a woman from getting HIV. T F DK

45. Taking vitamins keeps a person from getting HIV. T F DK

 Answer Key HIV KQ 45
1 FALSE
2 FALSE
3 FALSE
4 TRUE
5 FALSE
6 FALSE
7 FALSE
8 TRUE
9 TRUE
10 TRUE
11 FALSE
12 TRUE
13 FALSE
14 FALSE
15 FALSE
16 TRUE
17 TRUE
18 FALSE
19 TRUE
20 FALSE
21 TRUE
22 FALSE
23 FALSE
24 TRUE
25 FALSE
26 FALSE
27 FALSE
28 FALSE
29 FALSE
30 TRUE
31 FALSE
32 FALSE
33 TRUE
34 FALSE
35 FALSE
36 FALSE
37 TRUE
38 TRUE
39 FALSE
40 FALSE
41 FALSE
42 TRUE
43 TRUE
44 FALSE
45 FALSE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
AIDS and Behavior, Vol. 1, No. 1, 1997

The HIV-Knowledge Questionnaire: Development and

Evaluation of a Reliable, Valid, and Practical
Self-Administered Questionnaire

Michael P. Carey,1*3 Dianne Morrison-Beedy,2 and Blair T. Johnson1

Received June 13, 1996; accepted July 5. 1996

We have developed and evaluated a self-administered questionnaire of knowledge about hu-

man immunodeficiency virus (HIV) infection for use in program evaluation. Formative work
led to the development of the 62-item HIV-Knowledge Questionnaire (HIV-K-Q),which was
administered to 409 women and 227 men. Item analyses resulted in the deletion of 17 items
that were either too easy or did not correlate well with the total score. Factor analysis on
the remaining 45 items resulted in a single factor labeled HIV Knowledge. The generalizability
of this one-factor solution was confirmed with data from 285 women and 76 men. Reliability
analyses revealed that the HIV-K-Q is internally consistent (alpha = .91) and stable over
1-week (r = .83), 2-week (r = .91), and 12-week (r = .90) intervals. Evidence for the validity
of the HIV-K-Q was assembled using known groups and treatment outcome analyses. Ad-
ditional evidence emerged from analyses that revealed associations between scores on the
HTV-K-Q and two related knowledge measures, and between HIV-K-Q scores and level of
educational attainment. Discriminant evidence was obtained through nonsignificant relation-
ships between the HTV-K-Q and potentially biasing constructs, including social desirability.
The HIV-K-Q requires a sixth-grade education, and 7 min to complete. The HIV-K-Q is a
reliable, valid, and practical measure of HIV-related knowledge that can be used with low-
literacy adults.
KEY WORDS: HIV; AIDS; knowledge; measurement; assessment.

INTRODUCTION tious agent that causes AIDS (Rosenberg, 1995). Al-

The Centers for Disease Control and Prevention
recently announced that acquired immunodeficiency
syndrome (AIDS) is the leading cause of death
among young adults in the United States. Estimates
indicate that 3% of African American men and 1%
of African American women in their 30s are living
with human immunodeficiency virus (HIV), the infec-
^Department of Psychology, Syracuse University, Syracuse, New
York.
2
College of Nursing, Niagara University, Niagara University, New
York.
3
Correspondence should be directed to Michael P. Carey, Depart-
ment of Psychology, 430 Huntington Hall, Syracuse University,
Syracuse, New York 13244-2340; e-mail: mpcarey@syr.edu
though rates of new infections among gay men have
declined, rates among heterosexual men and women
have either remained stable or increased. It has been
estimated that 71% of HIV-infected cases worldwide
involve heterosexual transmission (Ehrhardt, 1992),
and that poverty potentiates the risk of HIV infection
(O'Leary and Jemmott, 1995). AIDS incidence has in-
creased more rapidly in recent years among individu-
als born in 1960 or later compared to individuals born
earlier (Rosenberg, 1995). These data confirm that
AIDS affects people regardless of their gender, age,
race, or sexual orientation.
Because there is neither a cure nor a vaccine
against HIV and AIDS, behavioral change provides
the only protection against infection. Efforts to pre-
vent or reduce risk of HIV infection through behav-

61
1090-7165/97ABMX)61J12.5(M> O 1997 Plenum Publishing Corporation
62
ior change begin with an understanding of the de-
terminants of risk behavior. Several models have
been proposed to explain HIV-related risk behavior.
Prominent among these are the AIDS Risk Reduc-
tion Model (Catania et al., 1990) and the Informa-
tion-Motivation-Behavioral Skills model (Fisher and
Fisher, 1992). Both models identify knowledge (in-
formation) as an important determinant of risk be-
havior. Accordingly, most intervention programs
provide information about HIV-related transmission
and prevention to help participants to reduce their
risk of infection. In addition, these models invoke
motivational constructs, such as perceived threat or
risk, which require an accurate knowledge of the con-
sequences of HIV infection. To evaluate the validity
of these models and to determine the effectiveness
of interventions guided by them requires a reliable
and valid measure of HIV knowledge.
Several authors have developed measures to as-
sess HIV-related knowledge. Zimet (1992) described
a 22-item measure for adolescents that was based
upon a Centers for Disease Control (1988) brochure.
He reported that the test was internally consistent
(Kuder-Richardson formula 20 coefficient of .77),
but did not provide information on test development
or refinement. Item, factor, or validity analyses were
not conducted (or reported). Koopman et al. (1990)
also developed a measure for adolescents, the AIDS
Knowledge Test (AKT), a 52-item measure of HIV-
and AIDS-related knowledge. The AKT assesses six
domains (i.e., definitions, outcomes, risk behavior,
transmission, prevention, and HIV testing); however,
no factor analyses have been reported to confirm this
factor structure. Moreover, reliability analyses within
these domains indicated poor internal consistencies
(Cronbach's coefficient alphas ranged from .00 to
.57, M = .43); collapsing across these domains
yielded a more reliable coefficient (alpha = .82) that
was stable (i.e., test-retest reliability of .82 for 1 week
[Koopman et al., 1990]). The validity of the AKT has
not been examined.
Kelly et al. (1989) described the AIDS Risk Be-
havior Knowledge Test (ARBKT), a standardized 40-
item measure that they developed for use with gay
men. These authors provided detailed information re-
garding test construction and development; they also
conducted item, factor, and reliability analyses and at-
tempted to validate the ARBKT by examining pre- (M
= 87%) and post-scores (M = 93%) of 33 men who
underwent an AIDS education seminar. However, the
factor structure of the ARBKT has not been replicated
Carey, Morrison-Beedy, and Johnson
and evidence of the validity of this measure has been
modest. Moreover, the ARBKT uses a "true-or-false"
response format, increasing the likelihood that respon-
dents may guess, which adds error variance to test
scores. Finally, for our purposes (i.e., use with low-lit-
eracy adults), the ARBKT is limited because it was
developed with participants who were relatively well-
educated (M = 14.4 years; Kelly et al., 1989).
Dancy (1991) developed the AIDS Knowledge,
Feelings, and Behavior Questionnaire (AKFBQ) spe-
cifically for use with African American women. The
AKFBQ contains 107 items, of which 40 items are de-
voted to the assessment of HIV-related knowledge. Al-
though few details were provided regarding scale
development or item analyses, factor analysis on the
Knowledge subtest of the AKFBQ yielded a 15-factor
solution; this solution was not rotated or replicated,
and there was no discussion of its interpretation. Cron-
bach's alpha (.66) was reported for the entire subtest,
despite the factor solution. The validity of the
AKFBQ Knowledge subtest has not been examined.
Thus, although several measures of HIV-related
knowledge have been developed and described, few
have been adequately evaluated. The measure that has
been evaluated most thoroughly, namely the ARBKT
(Kelly et al., 1989), was evaluated primarily with data
provided by well-educated gay men. Given (a) the dif-
fusion of HIV disease into multiple communities (in-
cluding heterosexual women), (b) the need to evaluate
educational and risk reduction programs, and (c) the
absence of an appropriate measure of HIV-related
knowledge, the purpose of this program of research
was to develop and evaluate a self-administered ques-
tionnaire to measure knowledge about HIV infection.
We sought to develop a measure that was reliable and
valid, understandable to those with low-literacy skills,
and appropriate for use regardless of respondent age,
gender, and/or sexual orientation. Our aim was to de-
velop a measure that was brief but sensitive, so that
it could be used in the evaluation of focused HlV-edu-
cational, risk reduction, testing, and counseling pro-
grams. We anticipated that such a measure might also
be useful in theoretical model building and testing,
and in clinical (e.g., primary care) settings.
STUDY 1. SCALE CONSTRUCTION AND
FORMATIVE EVALUATION
The initial phase of the research involved an ex-
plicit articulation of the domain to be evaluated and
 HIV-Knowledge Questionnaire
the generation of items (Dawis, 1987; Haynes et al.,
1995). New information regarding HIV and AIDS
emerges daily, but most of this information is not di-
rectly relevant to the general public (i.e., non-health
professionals). Because our primary interest involves
direct service delivery to the general public, particu-
larly economically disadvantaged individuals who
tend to be less well-educated, we sought to develop
a measure that assessed basic knowledge rather than
the more sophisticated knowledge required of a
health-care provider. Therefore, the generation of
items was informed by three sources.
First, we studied carefully several existing meas-
ures (e.g., Dancy, 1991; Kelly et al., 1989; Koopman
et al., 1990; Zimet, 1992). We noted that prior meas-
ures (a) unintentionally encouraged guessing be-
cause they did not provide an "I don't know" option;
(b) contained attitudinal as well as knowledge items;
(c) included items for which the correct answer was
unclear due to newly emerging facts; and (d) con-
tained few items relevant to women's concerns. We
remained mindful of these limitations as we devel-
oped items.
Second, we consulted with eight local and na-
tional HIV and AIDS educators and researchers, in-
cluding faculty at the Center for AIDS Intervention
Research (Milwaukee, WI) and Syracuse University,
and HIV educators from a local AIDS service or-
ganization. These experts were interviewed to deter-
mine what they perceived to be the most relevant
and immutable facts. More than ten overlapping do-
mains were identified. These domains comprised
transmission vectors, myths regarding casual trans-
mission, risk reduction strategies, consequences of in-
fection, and treatment of HIV disease.
Third, we held a series of eight focus groups
with 45 low-income women to learn what informa-
tion and myths were widely held in the community
(Carey et al., in press a; see Zeller, 1993). The av-
erage participant was 26 years old, with a high
school education, who had two children; the major-
ity of participants were African American, single,
unemployed, with a family income of less than $8000
per year. Most of the women reported that they had
been tested for HIV, but none reported that she was
infected. We purposely included ethnic minority
women with limited educational backgrounds be-
cause prior instrument development had tended not
to include participants from this socioeconomic
background, and current epidemiological evidence
suggested that such persons are at increased risk of
63
infection (Rosenberg, 1995). These focus groups
were cofacilitated by a culturally diverse team of two
women, a doctoral-level nurse and a masters-level
social worker; both facilitators had extensive expe-
rience in women's sexual health. Material provided
by the participants confirmed that myths about cas-
ual transmission (e.g., by insect bite, hugging, or
kissing) and incorrect ideas regarding prevention
(e.g., using birth control pills or douching) remained
prevalent. In addition, we learned that accurate
knowledge regarding accepted transmission vectors
(e.g., anal sex) and effective prevention strategies
(e.g., male and female condom) remained incom-
plete. Participants also were poorly informed about
the role of antibody testing, and the treatment of
HIV disease.
Guided by these archival, professional, and gen-
eral public sources, we developed the HIV-Knowl-
edge Questionnaire (HIV-K-Q), a 68-item
self-administered questionnaire that tapped HIV
transmission (e.g., vaginal, anal, and oral sexual in-
tercourse; blood products; needle sharing; and per-
inatal), nontransmission (e.g., saliva, insect bites,
touching, sharing food), effective risk reduction
strategies (e.g., male and female condom, absti-
nence, monogamy following antibody testing), inef-
fective (e.g., douching, birth control pills, vaccine)
prevention methods, and consequences of infection
(e.g., asymptomatic period, treatment, disease
course).
These 68 items were then distributed to six
HIV experts who provided critiques of the items'
wording, content, and relevance to HIV risk reduc-
tion. Based upon this feedback, 6 items were
dropped because they were redundant with other
items, or were deemed not sufficiently important.
Minor revisions were made to the remaining 62
items to clarify the content and reduce the reading
level of the items; the final set of 62 items appears
in Table I.
STUDY 2. ITEM AND FACTOR ANALYSES
The purposes of Study 2 were (a) to obtain a
sample that was diverse with respect to age, gender,
ethnicity, income, educational attainment, and antici-
pated HIV-related knowledge; (b) to administer the
62-item HIV-K-Q; and (c) to conduct item and factor
analyses on these data in order to reduce and refine
the measure, and determine its factor structure.
64 Carey, Morrison-Beedy, and Johnson

Table I. Items in the HIV-Knowledge Questionnaire: Study 2

Item
number*
Scale Scale Percent s for Item-
A B correct item total^ Item
1 1 80.42 .40 .41 HIV and AIDS are the same thing
2 2 91.63 .28 .33 There is a cure for AIDS
3 3 89.09 .31 .34 A person can get HIV from a toilet seat
4 4 75.78 .43 .34 Coughing and sneezing DO NOT spread HIV1'
5 97.91 .14 A person can get HIV by sharing an injection needle with someone who has HIV*
6 80.87 .39 .17 A person can get HIV if she or he has sex with someone who shoots up drugs+
7 5 65.47 .48 .31 HIV can be spread by mosquitoes
8 6 64.28 .48 .35 AIDS is the cause of HIV
9 7 85.35 .35 .33 A person can get HIV by sharing a glass of water with someone who has HIV
10 96.41 .19 A person can get HIV by shaking hands with someone who has HIV
11 8 18.98 .39 .29 HIV is killed by bleach*
12 9 69.81 .46 .26 It is possible to get HIV when a person gets a tattoof
13 96.56 .18 A man can get HIV if he has sex with another man who has HIV1'
14 10 94.92 .22 .32 A pregnant woman with HIV can give the virus to her unborn baby+
15 11 85.20 .36 .40 Pulling out the penis before a man climaxes keeps a woman from getting HIV during sex
16 12 76.83 .42 .35 A woman can get HIV if she has anal sex with a man*
17 13 90.58 .29 .45 Showering, or washing one's genitals after sex keeps a person from getting HIV
18 95.07 .22 A man can get HIV if he has vaginal sex with a woman who has HIV*
19 14 93.27 .25 .29 Eating healthy foods can keep a person from getting HIV
20 15 66.22 .47 .45 All pregnant women infected with HIV will have babies born with AIDS
21 16 93.27 .25 .34 Using a latex condom or rubber can lower a person's chance of getting HIV*
22 95.22 .21 .34 Taking the Birth Control Pill keeps a woman from getting HIV
23 59.34 .49 .21 A diaphragm and the Birth Control Pill provide the same protection against HIV infection
24 91.18 .28 .12 Most people with AIDS will die from it*
25 17 91.48 .28 .40 A person with HIV can look and feel healthy*
26 29.30 .46 .22 There are more cases of AIDS in the USA than in the rest of the world
27 18 83.41 .37 .48 People who have been infected with HIV quickly show serious signs of being infected
28 19 90.28 .30 .48 A person can be infected with HIV for 5 years or more without getting AIDS*
29 20 81.32 .39 .42 There is a vaccine that can stop adults from getting HIV
30 21 78.62 .41 .33 Some drugs have been made for the treatment of AIDS*
31 96.41 .19 There is a blood test to tell if a person has been infected with HIV*
32 22 46.93 .50 .37 Women are always tested for HIV during their pap smears
33 23 71.30 .45 .34 A person cannot get HIV by having oral sex, mouth-to-penis, with a man who has HIV
34 32.59 .47 .15 In the. USA, most cases of AIDS resulted from sex between men and women
35 24 94.17 .23 .32 A person can get HIV even if she or he has sex with another person only one time*
36 39.61 .49 .13 A mother with HIV can pass it on to her baby by breast feeding*
37 25 54.56 .50 .40 Using a lambskin condom or rubber is the best protection against HIV
38 26 62.93 .48 .32 People are likely to get HIV by deep kissing, putting their tongue in their partner's mouth, if
their partner has HIV
39 87.89 .33 .25 Infection with HIV leads to AIDS*
40 27 56.05 .50 .38 A person can get HIV by giving blood
41 28 81.17 .39 .42 A woman cannot get HIV if she has sex during her period
42 29 94.02 .24 .39 You can usually tell if someone has HIV by looking at them
43 78.18 .41 .12 A person can get HIV by getting blood during surgery*
44 30 58.89 .49 .34 There is a female condom that can help decrease a woman's chance of getting HIV*
45 31 57.10 .50 .45 A natural skin condom works better against HIV than does a latex condom
46 32 86.55 .34 .49 A person will NOT get HIV if she or he is taking antibiotics
47 33 80.12 .40 .35 Having sex with more than one partner can increase a person's chance of being infected with HIV*
48 34 69.51 .46 .41 Taking a test for HIV 1 week after having sex will tell a person if she or he has HIV
49 35 83.70 .37 .42 A person can get HIV by sitting in a hot tub or a swimming pool with a person who has HIV
50 36 54.71 .50 .36 A person can get HIV through contact with saliva, tears, sweat, or urine
51 37 62.78 .48 .27 A person can get HIV from a woman's vaginal secretions (wetness from her vagina)1'
52 29.75 .46 .07 A person is more likely to get HIV if she or he has another STD (VD), such as herpes or
the clap*
HIV-Knowledge Questionnaire 65

Table I. Continued
Item
number*
Scale Scale Percent s for Item-
A B correct item total/' Item
53 21.82 .41 .11 Taking the AIDS drug AZT lowers the chance of a pregnant woman with HIV giving it to
her baby*
54 18.83 .39 .24 Outside of the USA, most cases of AIDS resulted because of IV (needle) drug use or men
having sex with men
55 38 63.53 .48 .36 A person can get HIV if having oral sex, mouth on vagina, with a woman*
56 39 51.27 .50 .34 If a person tests positive for HIV, then the test site will have to tell all of his or her partners
57 40 79.97 .40 .47 Using Vaseline or baby oil with condoms lowers the chance of getting HIV
58 41 80.57 .40 .40 Washing drug-use equipment with cold water kills HIV
59 42 93.42 .25 .33 A woman can get HIV if she has vaginal sex with a man who has HIV*
60 43 89.39 .31 .32 Athletes who share needles when using steroids can get HIV from the needles*
61 44 87.59 .33 .45 Douching after sex will keep a woman from getting HIV
62 45 90.58 .29 .46 Taking vitamins keeps a person from getting HIV
"Items marked with daggers (t) are true, those without are false.
*Items were dropped due either to low item-total correlations or to restriction of range.
Tor total of 56 items remaining after dropping the items with restriction of range (5, 10, 13, 18, 22, and 31) and subtracting the item
to be correlated.

Methods were recruited by a research assistant as they waited

Participants and Procedures
Participants were recruited from three subsam-
ples (primary care, university, and HIV/AIDS ex-
perts) so that we could assemble a diverse sample that
would permit generalization to the general popula-
tion. Combining the subsamples, participants (N =
669) included 409 women and 227 men (33 subjects
did not identity their gender). Their ages ranged from
15 to 76 years (M = 28.50 years, SD = 12.10). The
sample was diverse with respect to ethnic/racial back-
ground (64% European American, 25% African
American, 3% Native American, 3% Hispanic Ameri-
can, 4% Other, 2% did not identity), income (16%
less than $10,000; 16% $10,000-$19,999; 11%
$20,000-30,000; 5% greater than $30,000; 53% did
not report income), and educational level (range =
6-20 years, M = 12.27, SD = 2.36). Demographic
characteristics by subsample are provided in Table II.
to see a health care provider to receive medical care
at an urban primary care setting. Participants were
paid $5 to complete the HIV-K-Q and a demo-
graphic survey in a waiting room as they waited to
be seen. The undergraduate students were self-re-
cruited in response to posted announcements, and
were provided with course credit to complete a
longer survey that included the HIV-K-Q and a
demographic survey. They completed their survey in
groups of 20 or fewer in large classrooms on cam-
pus. The experts received a written invitation from
the first author asking them to complete the HIV-
K-Q and a brief demographic questionnaire. The ex-
perts completed these measures in private, and
returned them by mail. No compensation was pro-
vided to the experts.
Results
Item Analyses

Procedures
All participants completed a self-administered
survey that included the 62-item HIV-K-Q. Proce-
dures for recruitment of participants and data col-
lection varied by setting. The primary care patients
Means and standard deviations for the 62 HIV-
K-Q items from the primary care, university students,
and experts (N = 669) were calculated (see Table I).
Examination of the percent correct reveals that six
items were too easy (i.e., Mean >95% correct); these
66 Carey, Morrison-Beedy, and Johnson

Table II. Demographic Characteristics and HIV-Knowledge Questionnaire Scores Overall and Within Samples0
Study and sample
Study 2 Study 3
Community
Overall HIV/AIDS Primary care University Couples women
Demographic characteristic (N = 1033) experts (n = 40) (n = 350) (n = 279) (n = 152) (n = 212)
M age in years (s) 30.59 (11.56) 36.66 (9.38) 34.99 (12.97) 20.11 (1.95) 34.70 (6.79) 33.85 (11.18)
Gender
Females 694 (67%) 23 (58%) 241 (69%) 145 (52%) 76 (50%) 212 (100%)
Males 303 (29%) 16 (40%) 87 (25%) 124 (44%) 76 (50%) 0 (0%)
Not specified 36 (3%) 1 (3%) 22 (6%) 10 (4%) 0 (0%) 0 (0%)
Race
European American 591 (57%) 19 (48%) 190 (54%) 218 (78%) 141 (93%) 23 (11%)
African American 329 (32%) 16 (40%) 119 (34%) 29 (10%) 6 (4%) 159 (75%)
Native American 31 (3%) 1 (3%) 17 (5%) 0 (0%) 0 (0%) 13 (13%)
Hispanic American 30 (3%) 1 (3%) 11 (3%) 10 (4%) 0 (0%) 8 (4%)
Other or not specified 52 (5%) 3 (8%) 13 (4%) 22 (8%) 5 (3%) 9 (4%)
M number of children (s) 1.99 (1.68) 1.92 (1.73) 1.88 (1.52) 2.17 (1.71)
M education in years (s) 12.54 (2.40) 16.46 (2.72) 11.80 (1.78) 14.22 (2.37) 11.82 (1.89)
Household income level
Less than $10,000 228 (22%) 104 (30%) 0 (0%) 124 (58%)
$10,001-20,000 165 (16%) 104 (30%) 7 (5%) 54 (25%)
$20,001-30,000 134 (13%) 72 (21%) 45 (30%) 17 (8%)
$30,001-40,000 89 (9%) 30 (9%) 52 (34%) 7 (3%)
Greater than $40,000 60 (6%) 7 (2%) 48 (32%) 5 (2%)
Unknown 357 (35%) 40 (100%) 33 (9%) 279 (100%) 0 (0%) 3 (1%)
Internal consistency, .91 .85 .88 .83 .93 .91
45-item HIV-K-Q (a)
M HIV-K-Q proportion .72 (.18) .91 (.10) .69 (.17) .82 (.12) .52 (.14) .72 (.19)
correct (s)
"Due to rounding of individual percentage values, the sum of percentages sometimes differs from 100%.

items (numbered 5,10,13,18, 22, and 31 under Scale
A, Table I) were deleted.4
The standard deviations of the remaining 56
items revealed that there was sufficient variability to
retain these items. Next, point-biserial correlations of
each item with the 56-item total score were calcu-
lated. An item-total correlation > .25 was established
as acceptable; 11 items whose item-total correlations
were less than .30 were deleted (numbered 6, 23, 24,
26, 34, 36, 39, 43, 52, 53, 54 under Scale A, 'Eible I).
The remaining 45 items (Scale B, Table I) were used
in subsequent analyses.
Factor Analyses

4Because the primary use of the HIV-K-Q will be for the evalu-
ation of educational and risk reduction programs, it is appropri-
ate to delete items that the vast majority of respondents answer
correctly. This approach allows the generation of a relatively
brief test that results in a more normalized distribution of scores
for the general population. However, this item-analytic approach
may be less appropriate for some clinical applications where the
questionnaire will be used for ideographic assessment purposes
(e.g., to identify high-risk knowledge deficits). Thus, although the
items deleted at this stage of scale development do not add ap-
preciably to the scale for its primary purpose, they remain "criti-
cal" knowledge items for clinical use; these items can and should
be used in settings where an assessor needs to be certain that
respondents have the essential knowledge necessary to avoid in-
fection with HIV.
We performed a principal factor analysis on the
45-item HIV-K-Q and employed two criteria to de-
termine the number of factors to be retained: (a)
Kaiser-Guttman's criterion (i.e., factors with an
eigen value of greater than 1), and (b) examination
of the scree plot. Four factors met the Kai-
ser-Guttman criterion, explaining 59%, 13%, 10%,
and 8% of the variance, respectively. Because the
scree plot confirmed that eigen values leveled off af-
ter the fourth factor, only four factors were retained.
An oblique rotation revealed highly intercorrelated
factors |.42 < r < .581 that each correlated highly
with the 45-item total score (| r \s < .72), and the
HIV-Knowledge Questionnaire
individual factors were difficult to interpret as unique
factors; we tested whether the four factors might be
reduced further by conducting a secondary factor
analysis on the four factor scores (see Floyd and Wi-
daman, 1995). This analysis produced only one factor
that exceeded the Kaiser-Guttman criterion. More-
over, each of the original four factors loaded heavily
on this general factor, | .63 to .771, and the general
factor produced by the secondary factor analysis cor-
related almost perfectly with the raw knowledge
score, r = .99.
A confirmatory factor analysis fitting a latent
one-factor model to the 45 items revealed that (a)
goodness of fit was conventionally high (GFI = .810),
and adding factors failed to increase fit substantially
(GFI = .833 with two factors, .811 with three factors,
and .864 with four factors). Thus, one core factor
emerged from this analysis, which we labeled HIV
Knowledge.
STUDY 3. GENERALIZABILITY OF THE
FACTOR SOLUTION
The purpose of Study 3 was to cross-validate the
factor structure of the 45-item HIV-K-Q obtained in
Study 2 with data from two additional community-
based samples.
Methods
Participants
Participants were recruited from two subsamples
(urban women and married couples), so that we
could evaluate the stability of the factor structure
with adults living in the community, the target audi-
ence for the HIV-K-Q. Combining these two samples,
participants (N = 364) included 285 women and 76
men. Their ages ranged from 15 to 72 years (M =
34.2 years, SD = 9.6). The sample was diverse with
respect to ethnic/racial background (45% European
American, 45% African American, 4% Native
American, 2% Hispanic American, 3% Other, 1%
did not identify), income (34% less than $10,000;
17% $10,000-$19,999; 17% $20,000-30,000; 31%
greater than $30,000; 1% did not report income), and
educational level (range = 6-20 years, M = 12.8, SD
= 2.4). Demographic characteristics by subsample
appear in Table II.
67
Procedures
All participants completed a self-administered
survey that included the 45-item HIV-K-Q. Proce-
dures for recruitment of participants and composi-
tion of the survey varied by sample. The urban
women responded to posted announcements in a va-
riety of community-based settings (e.g., laundromats,
grocery stores, beauty shops, health and social service
clinics, bus stops). The announcements invited
women to participate in a Women's Health Project
to be held at an urban community-based organiza-
tion. Women completed the survey in small groups,
and were paid $15. The couples responded to news-
paper announcements requesting participants in a
study of marital communication. One male and one
female member of the research team visited couples
in their homes, and supervised the independent com-
pletion of the survey. Each couple was paid $20 for
their participation.
Results
We performed a confirmatory factor analyses to
determine the generalizability of the one-factor
model that emerged in Study 2. This analysis re-
vealed that the the goodness of fit for a one-factor
model was adequate (GFI = .660), and that adding
factors failed to increase fit (GFI = .658 with two
factors, .604 with three factors, and .666 with four
factors). Thus, one core factor emerged from this
analysis, which we again labeled HIV Knowledge.
STUDY 4. RELIABILITY
The purpose of Study 4 was to determine the
internal consistency and the test-retest stability of
the 45-item HIV-K-Q.
Methods
Participants and Procedures
Participants from all five subsamples described
in Studies 2 and 3 provided data for analyses of the
internal consistency of the 45-item HIV-K-Q. Com-
bining subsamples yielded N = 1,033 participants, in-
cluding 694 women and 303 men (36 subjects did not
identify their gender). Additional demographic char-
68
acteristics for the entire sample and for each sub-
sample can be found in Table II.
Analyses of the test-retest stability made use of
data from two sources. First, a subset of the urban
women's sample ( = 33) completed the HIV-K-Q
on three occasions; the first occasion as described in
Study 3, and then again at 2-week and 12-week retest
sessions. Second, a subset of the university students
also completed the HIV-K-Q on two occasions, sepa-
rated by approximately one week. Procedures for the
retest sessions were identical to those used for the
initial data collection.
Results
Internal Consistency
Internal consistency was determined with Cron-
bach's (1951) alpha. Alpha for the total sample (N
= 1,033) was .91, which reflects a high degree of in-
ternal consistency. Alpha was also calculated sepa-
rately for each subsample, and ranged from .83 to
.93 across the five subsamples.
Test-Retest Stability
A Pearson product-moment correlation coeffi-
cient was calculated for the urban women who took
the HIV-K-Q on three occasions, with 2-week and
3-month retest intervals ( = 33 and n = 25, re-
spectively). These calculations indicated high
test-retest reliability (r = .91 and r = .90 at 2 and
12 weeks, respectively; both ps < .0001). The
test-retest correlation for the university students
was r(130) = .83, p < . 0001.
STUDY 5. VALIDITY: COMPARISON OF
KNOWN GROUPS
The purpose of Study 5 was to assemble evi-
dence for the validity of the HIV-K-Q using "known
groups," that is, groups expected to differ with re-
spect to their HIV-related knowledge. Therefore, we
compared HIV-K-Q scores among "HIV experts,"
relatively well-educated college students, and three
community samples; we expected these groups to dif-
fer in their HIV-related knowledge as a function of
prior education and experience related to HIV and
AIDS. We predicted that the experts would demon-
Carey, Morrison-Beedy, and Johnson
strate more knowledge than the college students,
who, in turn, would demonstrate more knowledge
than the community samples.
Methods
Participants from the five subsamples described
in Studies 2 and 3 (N = 1,033) provided data for
these analyses.
Results
A one-way (group: primary care patients; urban
women; university students; HIV experts, community
couples) analysis of variance (ANOVA) revealed a
significant effect for group status, F(4, 1021) =
111.83, p < .0001. As expected, pairwise comparisons
revealed that the experts (91%) were significantly
better informed than the other four groups, and the
university students (82%) were more knowledgeable
than were the primary care patients (69%), urban
women (72%), and community couples (52%). Al-
though the primary care and urban women did not
differ from each other, both groups were more
knowledgeable than the community couples. All five
groups exceeded the score that would be achieved
by chance (i.e., 33%).
STUDY 6. VALIDITY: CHANGE RESULTING
FROM A PSYCHOEDUCATIONAL
INTERVENTION
To provide treatment-related evidence, data
from a sample of urban women who participated in
an HIV-risk-reduction program were used (Carey et
al., in press b). Because a component of the program
involved education about HIV and AIDS, we ex-
pected that treated participants would demonstrate
knowledge increases, whereas the control partici-
pants would not change on this dimension.
Methods
A subset of the urban women's sample (n = 78)
participated in a controlled clinical trial of an HIV-
risk-reduction program. This intervention program was
designed primarily to enhance HIV-related motivation
and skills, but also included one-half session (i.e., 45
min) devoted to basic facts regarding HIV and AIDS.
HIV-Knowledge Questionnaire
The women completed the HIV-K-Q independently 1
week before and 1 week after the program.
Results
Scores on the HIV-K-Q indicated that women
assigned to the intervention condition significantly
unproved their scores from pre- (M = 75%) to post-
intervention (M = 87%) assessments, f(42) = 6.08,
p = .0001, d = 0.90. In contrast, women in the con-
trol condition of this study did not improve their
scores, M = 71% and M = 72%, respectively, f(31)
= 0.79, p > .10, d = 0.13.
STUDY 7. VALIDITY: DISCRIMINANT
EVIDENCE
The purpose of Study 7 was to assemble dis-
criminant evidence to evaluate the validity of the
HIV-K-Q. Discriminant evidence obtains when meas-
ures not expected to correlate or not to correlate
very highly with the target measure show this ex-
pected pattern (Kazdin, 1995; Campbell and Fiske,
1959). Therefore, we selected reliable and valid in-
struments that measure important psychological con-
structs (i.e., social desirability, mood, and self-
esteem) that might influence responses to the HIV-
K-Q as well as other constructs (i.e., dyadic and sex-
ual adjustment, attitudes toward condoms) that,
although not directly related to HIV knowledge, are
indirectly related because of a common association
with sexual health. We did not expect to find signifi-
cant associations between any of these variables and
the HIV-K-Q.
Methods
Participants from the university student and cou-
ple subsamples described in Studies 2 and 3 provided
data for these analyses. These participants completed
additional measures at the same time that they had
completed the HIV-K-Q. The university students
completed the Social Desirability Scale, Positive and
Negative Affect Scale, Rosenberg Self-Esteem Scale,
and Multidimensional Condom Attitudes Scale,
whereas the couples completed the Social Desirabil-
ity Scale, Center for Epidemiological Studies Depres-
sion Scale, Index of Sexual Satisfaction, and Dyadic
Adjustment Scale. No sample was asked to complete
69
the entire battery in order to minimize the response
burden placed on participants.
The Social Desirability Scale (SDS; Crowne and
Marlowe, 1960) contains 33 true/false items and
measures the tendency to present oneself in a socially
desirable manner. The SDS is internally consistent
(alpha = .88) and stable (test-retest reliability r =
.89 at 1 month).
The Center for Epidemiological Studies Depressed
Mood Scale (CES-D; Radloff, 1977) contains 20
items designed to measure depressive symptoms in
the general population. When used with the general
population, the CES-D has an alpha of .85, with
test-retest correlations that range from .51 to .67
when tested over 2-8 weeks.
The Positive and Negative Affect Scale (PANAS;
Watson et al., 1988) contains two 10-item mood
scales that provide independent measures of positive
and negative affect. Positive affect reflects the extent
to which an individual feels enthusiastic, active, and
alert, whereas negative affect provides an index of
subjective distress, anger, and nervousness. Both
PANAS scales are internally consistent (alphas range
from .84 to .90), and stable (rs = .86 to .87).
The Rosenberg Self-Esteem Scale (RSES; Wylie,
1977) is a widely used measure that provides an in-
dex of an individual's sense of his or her general
worth or value.
The Multidimensional Condom Attitudes Scale
(MCAS; Helweg-Larson and Collins, 1994) contains
25 statements about condoms and yields five attitude
scores: reliability of condoms, pleasure of condom
use, identity stigma associated with condom use, em-
barrassment about negotiation and use of condoms,
and embarrassment about the purchase of condoms.
Internal consistency of factors range from .67 to 94
for men and .44 to .92 for women.
The Dyadic Adjustment Scale (DAS; Spanier,
1976) contains 32 items and provides a general meas-
ure of satisfaction in an intimate relationship. The
DAS has an alpha of .96, with a test-retest of .87.
The Index of Sexual Satisfaction (ISS; Hudson et
al., 1981) is a 25-item measure of the degree, sever-
ity, or magnitude of problems in the sexual compo-
nent of a couple's relationship. The ISS has a mean
alpha of .92, with a 2-hr test-retest of .94.
Results
Discriminant evidence was provided by the ab-
sence of significant correlations between the HIV-K-
70
Q and (a) the SDS, using either the undergraduate
sample, r(271) = -.09, or the couples, r(148) = -.09;
(b) positive or negative mood subscales from the
PANAS, r(277) = .00 and r(277) = -.05, respec-
tively; (c) the RSES, r(268) = -.05; (d) any of the
five subscales from the MCAS, rs ranging from -.05
to .11; (e) the CESD, r(148) = -.11; (f) the DAS,
r(148) = .16; or (g) the ISS, r(148) = -.05.
STUDY 8. VALIDITY: CONVERGENT EVIDENCE
The purpose of Study 8 was to assemble con-
vergent evidence for the validity of the HIV-K-Q.
Convergent evidence obtains when the measure be-
ing validated correlates with other measures that are
designed to assess the same or related constructs
(Kazdin, 1995; Campbell and Fiske, 1959). We an-
ticipated strong, positive associations between scores
on the HIV-K-Q and two extant measures of HIV-
related knowledge. However, because such correla-
tions between self-administered questionnaires may
also occur as a result of shared method variance
(Kazdin, 1995), we also examined the association be-
tween HIV-knowledge scores and level of educa-
tional attainment. Prior research (Peruga and
Celentano, 1993) suggested a positive association be-
tween these two variables, which is what we expected
to find.
Methods
Participants from the subsamples described in
Studies 2, 3, and 4 provided data for the analyses
that examined the relationship between HIV-K-Q
score and education level. For the analyses that ex-
amined the relations between the HIV-K-Q and
other measures, we collected data from a fresh sam-
ple of 50 primary care patients at an urban clinic.
The sample was 52% female, and diverse with re-
spect to ethnic/racial background (52% European
American, 41% African American, 6% Other) and
income (39% less than $10,000; 32%
$10,000-$19,999; 7% $20,000-30,000; 21% greater
than $30,000). Only one third had any college edu-
cation (range = 7-17 years, M = 12.57, SD = 2.37).
The patients were recruited by a research assistant
as they waited to see a health care provider to re-
ceive medical care at an urban primary care setting.
Participants were paid $5 to complete the HIV-K-Q,
the AIDS Risk Behavior Knowledge Test, the AIDS
Carey, Morrison-Beedy, and Johnson
Knowledge Test, and a demographic survey in a wait-
ing room as they waited to be seen.
The AIDS Risk Behavior Knowledge Test
(ARBKT; Kelly et al., 1989) is a 40-item measure of
AIDS risk behavior knowledge. The ARBKT has
been judged to be internally consistent (KR-20 =
.74) and stable (i.e., test-retest r = .84 over 2 weeks
[Kelly et al., 1989]).
The AIDS Knowledge Test (AKT; Koopman et al.,
1990) is a 52-item measure of HIV- and AIDS-related
knowledge. The AKT has been judged to be internally
consistent (i.e., Cronbach's alpha = .82) and stable
(i.e., test-retest r = .82 over 1 week [Koopman et al.,
1990]).
These two measures were selected as the best
available similar measures of HIV-related knowledge.
As we noted in the Introduction, these measures
have proven useful with the populations for whom
they were developed, but they may be less appropri-
ate for low-literacy adult men and women.
Results
HIV-K-Q scores correlated with both the AIDS
Risk Behavior Knowledge Test, r(48) = .42, p < .005,
and the AIDS Knowledge Test, r(47) = .56, p <
.0001. Using the larger sample, the HIV-K-Q scores
were correlated with higher levels of educational at-
tainment, r(386) = .48, p < .0001.
STUDY 9. READING LEVEL
Several studies have demonstrated that health
education and assessment materials are often writ-
ten at levels that exceed respondents' reading abili-
ties (e.g., Williams et al., 1995; Powers, 1988; Meade
and Byrd, 1989). Doak and Doak (1980) reported
that respondents' report of the number of years of
formal education tends to be four or five levels
higher than their actual reading ability based on the
Wide Range Achievement Test, a word pronuncia-
tion and recognition test. Given these findings, and
our desire to develop a measure that would be prac-
tical for urban, economically disadvantaged adults
(who tend to have the lowest functional literacy
skills [Williams et al., 1995]), the purpose of Study
9 was to determine the reading level of the 45-item
HIV-K-Q.
HIV-Knowledge Questionnaire
Methods
The instructions and all 45 items of the HFV-K-
Q were entered in a text file and split into passages
of approximately 100 words each. For purposes of
readability analysis, HIV AIDS, and Vaseline were
treated as proper names. Two formulas, the Flesch
and the Spache, were used to determine the read-
ability of the material. These readability formulas use
common features of words and passages (e.g.,
number of syllables, number of words, number of
sentences) to measure the difficulty of the material.
Results
Analysis of the 45-item HTV-K-Q with the Flesch
formula indicated that the material was at the pri-
mary-grade level, with most passages in the "fairly
easy" to "Very easy" range. Analysis using the Spache
formula revealed that the majority of the material
was below 4th-grade level. However, several "for-
eign" words increased the level to approximately the
7th-grade level; these "foreign" words include vac-
cine, antibiotics, genitals, and pap smears, that is,
words that are either medical or sexual in nature. Al-
though these words increase the reading level, they
are likely to be understood by low-literacy readers
due to their common use in the popular culture.
STUDY 10. COMPLETION TIME REQUIRED
To determine how long the HTV-K-Q (and indi-
vidual items) take to complete, we administered the
questionnaire to two samples. With the community
sample of adults used in Study 8, we recorded the
time to complete the paper-and-pencil questionnaire
in a natural environment. With a fresh sample of col-
lege students, we used a computer to administer and
record the time required to complete each item as
well as the entire questionnaire when it is computer-
administered.
Method
Community Sample
Participants were those subjects described in
Study 8, namely, 50 primary care patients at an urban
71
clinic. Participants were administered the HIV-K-Q
individually in a waiting room. They were instructed
to answer each question carefully and, when finished,
to return this survey and request the next one. A re-
search assistant privately recorded the time required
to complete the HIV-K-Q. Respondents were not
told that they were being timed.
College Sample
Participants were 28 female and 21 male under-
graduates (M age = 18.63 years, SD = 0.88) from
the same university population that was sampled for
Study 2. Participants expected to participate in a se-
ries of studies on "Social Attitudes" and were seated
in individual cubicles that precluded visual contact
with any other participant. Each cubicle contained a
computer through which all instructions and ques-
tions were administered. After a brief introduction
to the computer, the questionnaire was introduced
without participants' foreknowledge. Participants
were simply instructed that they should "answer the
following true-false items about HIV and AIDS."
Respondents were not told that their responses were
being timed. Each question from the 45-item HIV-
K-Q was then presented individually, with a unique
random order of items for each participant. Partici-
pants pressed the "a" key for a "True" response, "b"
for a "False" response, and "c" for a "Don't Know"
response. The computer recorded latencies to re-
spond to each question.
Results
Community respondents' mean proportion of
correct answers to the HIV-K-Q was .74 (SD = .14),
which does not differ significantly from the mean for
the earlier primary care sample (M = .69), F(l,
398) = 2.95, p = .0869. It took an average of 6.98
min (SD = 2.19) to complete the HIV-K-Q, with
times ranging from 3.50 to 12.00 min. The correlation
between time to complete the HIV-K-Q and the
score attained on it was moderate in magnitude,
r(45) = -0.29, p = .0536.
College student respondents' mean proportion of
correct answers to theHlV-K-Qwas .85 (SD = .07),
which does not differ from the mean for the earlier
university sample (M = .82), F (1,326) = 3.17, p =
.0762. It took an average of 4.12 min (SD = 0.77) to
complete the HIV-K-Q, ranging from 1.37 to 5.85
72
min. The time to complete the HIV-K-Q was unre-
lated to the score attained on it, r (48) = .02. The
mean time to respond to any given item ranged from
a low of 2.73 sec (for item 2 on Scale B, see Table I)
to a high of 10.28 sec (for item 41).
DISCUSSION
The purpose of this study was to develop and
evaluate the psychometric properties of the HIV-
Knowledge Questionnaire. Formative research
guided the generation of a 62-item scale that as-
sessed knowledge about the transmission, prevention,
and consequences of HIV infection. Item analyses in-
dicated that 17 of the 62 items should be deleted be-
cause of item ease or poor association with the total
scale. Many of the items deleted can be considered
"critical items," that is, items containing essential
public health knowledge regarding HIV transmission
and prevention. These items describe risk associated
with needle-sharing, men who have sex with men,
and heterosexual vaginal sex (Scale A, items 5, 13,
and 18, respectively). It is comforting to know that
more than 95% of our sample answered these items
correctly, and it is appropriate to delete such items
for our purposes (i.e., program evaluation). However,
these items may prove useful in certain clinical set-
tings where idiographic assessment of high-risk indi-
viduals is necessary; knowledge of such material
should never be assumed in clinical and counseling
settings with an individual client.
Principal factor analyses of the remaining 45
items suggested that a single-factor could account for
a significant amount of the variance; this single-factor
solution was replicated in a cross-validation sample.
Thus, these analyses demonstrate that the HIV-K-Q is
a unidimensional instrument that measures HlV-re-
lated knowledge. Reliability analyses indicated that the
HIV-K-Q is internally consistent and stable over inter-
vals as long as 3 months. Validity analyses revealed, as
expected, that the HIV-K-Q is not associated with
other constructs, including social desirability, negative
or positive mood, self-esteem, depression, dyadic ad-
justment, sexual satisfaction, or attitudes toward con-
doms. As expected, however, the HIV-K-Q is
associated with level of educational attainment, with
better educated respondents scoring higher on the
HIV-K-Q (Peruga and Celentano, 1993).
An additional source of validity evidence was pro-
vided by the association between the HIV-K-Q and
Carey, Morrison-Beedy, and Johnson
two related measures, the ARBKT (Kelly et at., 1989)
and the AKT (Koopman et al., 1990). The magnitude
of these associations (r = .42 and r = .56, respectively)
is in the "moderate" range, and provides adequate
support for the validity of all three measures (Kazdin,
1995).5 However, it is interesting that these correla-
tions did not differ from the correlation between the
HIV-K-Q and level of educational attainment (r =
.48). It cannot be determined whether these moderate
correlations reflect upon the HIV-K-Q, ARBKT; or
AKT However, we can say that the magnitude of a
validity coeffient is limited by the underlying reliabili-
ties of the measures being associated (Nunnally and
Bernstein, 1994). It is also likely that the reliability of
self-report of level of educational attainment is greater
than are the reliabilities of the three knowledge meas-
ures. Mindful of these considerations, the relatively
equivalent correlations among the knowledge meas-
ures and educational attainment are logical.
As expected, analyses of known groups indicated
that experts were better informed that college stu-
dents, who in turn were more knowledgeable than
were the community samples. An unexpected finding,
however, was that primary care and urban women
were more knowledgeable than were the community
couples. One explanation for this finding is that, rela-
tive to the two other samples, the community couples
might feel less compelled to stay informed regarding
HIV because of (a) their (presumably) exclusive sexual
relationship with a partner and (b) the relatively lower
rates of HIV infection in middle-income communities.
Overall, the results reported herein demonstrate
that the HIV-K-Q is a psychometrically strong instru-
ment that compares favorably to extant measures.
That is, although several authors have developed in-
struments to assess HIV-related knowledge, none of
these measures has been thoroughly evaluated; for
example, most extant measures have not been factor
analyzed nor evaluated for validity. The psychomet-
rically strongest measure is the ARBKQ (Kelly et al.,
1989), but this measure was developed only with col-
lege students and gay men and is written at a 9th-
or 10th-grade reading level. Although this measure
has proven useful with the population with which it
was developed, it is less appropriate for low-literacy
men and women, groups most at risk in the second
wave of the AIDS epidemic (Kelly et al., 1993).
Confidence in the validity of the HIV-K-Q can be much greater
due to the other sources of evidence described previously.
HIV-Knowledge Questionnaire
It is increasingly apparent that many diseases
pose a disproportionate threat to those who are eco-
nomically disadvantaged (Marmot et al., 1987; Wil-
liams, 1990); the epidemiology of HIV disease
(Rosenberg, 1995) provides but one illustration of
this pattern. The economically disadvantaged, in
turn, are more likely to be functionally illiterate, or
at least unable to read at levels required by many
self-administered questionnaires (Williams et al.,
1995). Therefore, to the extent that test developers
seek to provide instruments that are relevant to those
most in need, it is essential that self-report measures
be designed with low-literacy respondents in mind.
For those adults and adolescents who cannot read at
even a 6th grade level, new assessment modalities
will need to be developed. For example, audiotaped
(Boekeloo et al., 1994) or compact disc-administered
questionnaires may prove helpful in some contexts.
The development and psychometric evaluation of
a reliable and valid knowledge measure, based upon
formative research, that is culturally sensitive and ap-
propriate for women represents a significant advance
over previous measures; urban women are a tradition-
ally neglected but increasingly at-risk group for HIV
infection (O'Leary and Jemmott, 1995). Data pre-
sented here and elsewhere (Carey et al., in press b)
indicate that the HIV-K-Q can be used to evaluate
the effect on intervention programs for women. Al-
though the HIV-K-Q contains 45 items, it requires
only a 6th-grade reading level and can be completed
in approximately 7 min by low-literacy adults.
Knowledge regarding the mechanisms of HIV
transmission and prevention, and the consequences
of HIV infection, have been identified as determi-
nants of HIV-related risk-taking and protective be-
haviors (Catania et al., 1990; Fisher and Fisher,
1992). Reliable and valid assessment of knowledge
will permit testing of these and related models and
allow for more precise evaluation of programs de-
signed to reduce risk-conferring behaviors. Given the
current status of AIDS vaccines and treatments, re-
finement of theoretical models and behavioral inter-
vention programs offers the best hope to curb the
continuing HIV pandemic.
ACKNOWLEDGMENTS
This research was supported by grants from the
National Institute of Mental Health to M.EC. and
B.T.J., and a grant from the National Institute of
73
Nursing Research to D.M.B. The authors thank
Thomas Bazydlo, Jesse Dowdell, Ann Goodgion,
Gary Urquhart, and Monique Wright-Williams for
their assistance with recruitment; Laura Braaten,
Lauren Durant, Andrew Forsyth, Christopher Gor-
don, Beth Jaworksi, Daniel Purnine, Lance Wein-
hardt, and Ednita Wright for their help with data
collection; Jack Gleason for developing the data-
checking software; June Crawford for assistance with
the readability formulas; Kathy Sikkema, Seth Ka-
lichman, Tim Heckman, David Rompa, David Wag-
staff, Nina Wright, Sue Taylor-Brown, Deborah
McLean, Andrew Forsyth, Christopher Gordon, and
Jeffrey Kelly for their expert consultation; and
Stephen Haynes for his thoughtful suggestions re-
garding the improvement of this manuscript, espe-
cially the caveats contained in footnote 4.
REFERENCES
Boekeloo, B. O., Schiavo, L, Rabin, D. L., Conlon, R. T., Jordan,
C. S., and Mundt, D. J. (1994). Self-reports of HIV risk fac-
tors by patients at a sexually transmitted disease clinic: Audio
vs. written questionnaires. American Journal of Public Health,
84, 754-760.
Campbell, D. T., and Fiske, D. (1959). Convergent and discrimi-
nant validation by the multitrait-multimethod matrix. Psycho-
logical Bulletin, 56, 81-105.
Carey, M. P., Gordon, C M., Morrison-Beedy, D., and McLean,
D. A, (in press a). Low-income women and HIV risk reduc-
tion: Elaborations from qualitative research. AIDS and Be-
havior.
Carey, M. P., Maisto, S. A., Kalichman, S. C., Forsyth, A. D.,
Wright, E.r and Johnson, B. T. (in press b). Enhancing mo-
tivation to reduce the risk of HIV infection for economically
disadvantaged urban women. Journal of Consulting and Clini-
cal Psychology.
Catania, J. A., Kegeles, S. M., and Coates, T. J. (1990). Towards
an understanding of risk behavior: An AIDS risk reduction
model (ARRM). Health Education Quarterly, 17, 53-72.
Centers for Disease Control. (1988). "Understanding AIDS": An
information brochure being mailed to all U. S. households.
Morbidity and Mortality Weekly Report, 37, 261-269.
Cronbach, L. J. (1951). Coefficient alpha and the internal struc-
ture of tests. Psychometrika, 6, 297-334.
Crowne, D. P., and Marlowe, D. (1960). A new scale of social
desirability independent of psychopathology. Journal of Con-
sulting Psychology, 24, 349-354.
Dancy, B. L. (1991). The development of an ethnically sensitive
and gender-specific AIDS questionnaire for African-Ameri-
can women. Health Values, 15, 49-54.
Dawis, R. V. (1987). Scale construction. Journal of Counseling Psy-
chology, 34, 481-489.
Doak, L., and Doak, C (1980). Patient comprehension profiles:
Recent findings and strategies. Patient Counseling and Health
Education, 2, 101-106.
Ehrhardt, A A. (1992). Trends in sexual behavior and the HIV
pandemic. American Journal of Public Health, 82, 1459-1461.
Fisher, J. D., and Fisher, W. A. (1992). Changing AIDS-risk be-
havior. Psychological Bulletin, 111, 455-474.
74
Floyd, F. J., and Widaman, K. F. (1995). Factor analysis in the
development and refinement of clinical assessment instru-
ments. Psychological Assessment, 7, 286-299.
Haynes, S. N., Richard, D. C. S., and Kubany, E. S. (1995). Con-
tent validity in psychological assessment: A functional ap-
proach to concepts and methods. Psychological Assessment, 7,
238-247.
Helweg-Larson, M., and Collins, B. E. (1994). The UCLA multi-
dimensional condom attitudes scale: Documenting the com-
plex determinants of condom use in college students. Health
Psychology, 13, 224-237.
Hudson, W. W., Harrison, D. F., and Crosscup, P. C. (1981). A
short-form scale to measure sexual discord in dyadic relation-
ships. Journal of Sex Research, 17, 157-174.
Kazdin, A. E. (1995). Preparing and evaluating research reports.
Psychological Assessment, 7, 228-237.
Kelly, J. A., St. Lawrence, J. S., Hood, H. V., and Brasfield, T.
L. (1989). An objective test of AIDS risk behavior knowledge:
Scale development, validation, and norms. Journal of Behavior
Therapy and Experimental Psychiatry, 20, 227-234.
Kelly, J. A., Murphy, D. A., Sikkema, K. J., and Kalichman, S. C.
(1993). Psychological interventions to prevent HIV infection
are urgently needed: New priorities for behavioral research
in the second decade of AIDS. American Psychologist, 48,
1023-1034.
Koopman, C, Rotheram-Borus, M. J., Henderson, R., Bradley, J.
S., and Hunter, J. (1990). Assessment of knowledge of AIDS
and beliefs about AIDS prevention among adolescents. AIDS
Education and Prevention, 2, 58-70.
Marmot, M. G., Kogevinas, M., and Elston, M. A. (1987). So-
cial/economic status and disease. Annual Review of Public
Health, 8, 111-135.
Meade, C, and Byrd, J. (1989). Patient literacy and the readability
of smoking education literature. American Journal of Public
Health, 79, 204-206.
Nunnally, J. C, and Bernstein, I. H. (1994). Psychometric theory
(3rd ed.). New York: McGraw-Hill.
Carey, Morrison-Beedy, and Johnson
O'Leary, A., and Jemmott, L. S. (1995). General issues in the pre-
vention of AIDS in women. In A. O'Leary and L. S, Jemmott
(Eds.), Women at risk- Issues in the primary prevention of AIDS
(pp. 1-12). New York: Plenum Press.
Peruga, A., and Celentano, D. D. (1993). Correlates of AIDS
knowledge in samples of the general population. Social Sci-
ence and Medicine, 36, 509-524.
Powers, E. (1988). Emergency department patient literacy and the
readability of patient-directed materials. Annals of Emergency
Medicine, 17, 124-126.
Radloff, L. S. (1977). The CES-D scale: A self-report depression
scale for research in the general population. Applied Psycho-
logical Measurement, 1, 385-401.
Rosenberg, P. S. (1995). Scope of the AIDS epidemic in the
United States. Science, 270, 1372-1375.
Spanier, G. B. (1976). Measuring dyadic adjustment: New scales
for assessing the quality of marriage and similar dyads. Jour-
nal of Marriage and the Family, 38, 15-28.
Watson, D., Clark, L.A., and Tellegen, A. (1988). Development
and validation of brief measures of positive and negative af-
fect: The PANAS scales. Journal of Personality and Social Psy-
chology, 54, 1063-1070.
Williams, D. R. (1990). Socioeconomic differences in health: A
review and redirection. Social Psychology Quarterly, 53, 81-99.
Williams, M. V., Parker, R. M., Baker, D. W., Parikh, N. S., Pitkin,
K., Coates, W. C, and Nurss, J. R. (1995). Inadequate func-
tional health literacy among patients at two public hospitals.
Journal of the American Medical Association, 274, 1677-1682.
Wylie, R. C. (1977). The self-concept: A review of methodological
considerations and measuring instruments. Lincoln, NB: Uni-
versity of Nebraska.
Zeller, R. A. (1993). Combining qualitative and quantitative tech-
niques to develop culturally sensitive measures. In D. G. Os-
trow and R. C. Kessler (Eds.), Methodological issues in AIDS
behavioral research (pp. 96-116). New York: Plenum Press.
Zimet, G. D. (1992). Reliability of AIDS knowledge scales: Con-
ceptual issues. AIDS Education and Prevention, 4, 338-344.