Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 56
CHAPTER 3
CONTRACEPTIVE DECI SION-MAKING AMONG ADOLESCENTS
Sandra L. Hofferth
Unlike many other activities which require a conscious effort, con-
ception is highly likely (given sexual activity) unless some action is
taken to prevent it. One writer suggested that if it were the reverse,
civilization would have long since died out. There is a certain cost
to making a decision to act--it requires motivation, time, resources,
and knowledge. It requires a change in behavior--overcoming inertia,
as it were. This takes time and energy. Thus it is really no surprise
that the risk of pregnancy is highest during the first months of sexual
activity {Zabin et al., 1979~. However, it appears that that risk is
also higher the younger the girl. Although contraception may be
problemmatic for older women as well (Tanfer and Horn, 1984), it may
be especially problemmatic for teenagers. This is because effective
use of contraception is linked to the process of defining oneself as
sexually active, becoming aware of pregnancy risk and its consequences,
developing motivation to prevent pregnancy and taking active steps to
prevent an unwanted pregnancy
The major question that will be addressed in this chapter is, given
sexual activity, what determines whether a teenager takes active steps
to prevent pregnancy? The chapter is organized as follows. Background
is first provided on contraceptive use among teens compared to adults.
Second, a framework is provided to organize the review. Third is a
discussion of some methodological issues, in particular, different
ways to measure contraceptive use and their implications. The research
that bears on the issue of contraceptive use versus non-use is pre-
sented, following the model presented earlier. Finally, the f indings
are summarized, conclusions drawn and implications for further research
presented.
BACKGROUND
Table 2.7 displays the number of women age 15-44 exposed to the
risk of unintended pregnancy who are currently practicing contracep-
tion. Women exposed to the risk of unintended pregnancy includes those
practicing contraception, and those not practicing contraception who
had sexual intercourse in the last three months and were not pregnant,
56
OCR for page 57
57
postpartum, seeking pregnancy, or noncontraceptively sterile. Never-
married women of all ages are less likely than married women to prac-
tice contraception; of these, teens are the least likely to practice
contraception (Bachrach, 1984~. The proportion contracepting is very
high: 9 of ten women at risk of unintended pregnancy. Never-married
women are less likely than married women to practice contraception,
and, of these, teens are the least likely to contracept: 69 percent
of never-married teens 15-19 were practicing contraception. Thus 30
percent of never-married teens 15-19 at risk of an unwanted pregnancy
were not contracepting.
There are substantial black-white differences, with black teen-
agers less likely than white teenagers to be contracepting {Table 2.8~.
However, since blacks who are exposed to the risk of pregnancy are
likely to be younger and less economically well-off, and since the
young and the poor are less likely to use contraception, this differ-
ence may merely reflect socioeconomic differences between the two
groups. This will be explored further in a later section.
Table 2.9 shows contraceptive use by Hispanic teenagers compared
with non-Hispanic black and white teenagers. Hispanic teenagers are
less likely than white but more likely than black teenagers to be cur-
rently contracepting. This table includes teenagers of all marital
statuses. More Hispanic teenagers are married, which may explain the
rather small differences among the three groups. A comparison of
sexually active never-married teenagers of Hispanic and non-Hispanic
descent would be more appropriate but the data are not currently
available.
One problem with statistics is deciding what the appropriate com-
parison group is for teenagers. Is it women 20-44 of all marital
statuses? A much larger proportion of older women than teenagers will
be married, and marriage improves contraceptive practice. A better
choice might be never-married women 20-44. The only problem is that
there are few women above 30 who have never been married. Those who
haven't may be unusual and therefore, not constitute an appropriate
comparison group. A good choice, therefore, for a comparison group
appears to be never-married women 20-24 (and perhaps 25-291. They are
similar in marital status and, given the increasing proportion in their
twenties who have never married, probably not exceedingly different
from their married peers in other respects. Fortunately, several
sources contain information on contraceptive use among never-married
women in their twenties, making possible comparisons with teenagers.
It is clear that contraceptive use does improve with age. Table 2.7
showed that 81 percent of never-married women 20-24 practiced contra-
ception compared with 69 percent of those 15-19 in 1982. A second
study found that 85 percent of never-married sexually active women
20-24 were currently using contraception (Tanfer and Horn, 1984~.
Table 2.8 shows the methods currently used by never-married users
15-19 and 20-24 in 1982. Teens are more likely than 20-24 year aids
OCR for page 58
58
to use the pill and the condom; they are less likely to use the IUD,
diaphragm or other methods. Teens are unlikely to be sterilized for
contraceptive purposes. The probability of sterilization is higher
for 20 to 24 year olds, and surprisingly high given their age. (Of
course, many of these young women may have children, though unmarried
and their partners may also have had children and had vasectomies).
Although there are differences between the two age groups, the dif-
ferences are not that striking. The largest difference appears to be
in condom use. Almost 4 times as many teens say that it is their
current method, compared with 20-24 year olds.
Table 2.9 shows the current method used by Hispanic and non-
Hispanic teenage contraceptive users in 1982. Two thirds of Hispanic
teenagers use the pill and 9 percent use the IUD or are sterilized,
which results in a very high level of use of effective methods (pill,
IUD, sterilization), a level as high as that of blacks, among whom
three quarters use the most effective methods. Again, this table
includes teenage women of all marital statuses. The high level of
effective contraceptive use among Hispanic teenage women is probably
due to the higher proportion of married teenagers among Hispanics.
One important issue is that of trends over time. According to
data from the National Survey of Family Growth, among 15 to 24 year
old currently married users, pill and IUD use declined substantially
between 1973 and 1982, while use of sterilization and barrier methods
such as the condom and diaphragm increased (Bachrach, 1984: Tables 7
and 8~. Trends among teenagers are somewhat different. According to
data on never-married teenagers 15 to 19, between 1971 and 1976 pill
use increased. Then, between 1976 and 1979, pill and JUD use dropped
while diaphragm and condom use increased. Data from the 1982 NSFG
survey suggest that the use of both pill and diaphragm among teenagers
increased since 1979 (Table 2.81. Since the measures are not identical
the size of the increase cannot be taken as definitive. However, the
data certainly show that the decline in pill use since the late 1970s
has been reversed. Sterilization showed major increases among young
women as well as older women, though levels are still very low among
young unmarried women.
The model used here for contraception was developed by Delameter
(1983~. It assumes that individuals determine and evaluate the
potential consequences of their actions and, after weighing these
consequences, act so as to maximize their benefits or minimize their
costs. This calculation need not be made consciously or even care-
fully. These basic assumptions are common to several theoretical
models in use today: The Health Belief model (Nathanson and Becker,
1983), the Fishbein model (Fishbein, 1972; Fishbein and Jaccard, 1973),
the economic utility model (Bauman and Udry, 1981; Becker, 1960;
Willis, 1973; Nambocdiri, 1972),j, and the Luker model (Luker, 1975;
Philliber et al., 1983~.
hi.
OCR for page 59
59
In this model, the proximal factors affecting contraceptive be-
havior are the perceived probability of pregnancy, given intercourse,
and its frequency, and the cognitive assessment of pregnancy, as well
as positive and negative experience with contraceptives. The psycho-
social factors in the model of sexual intercourse presented in Chapter
1 determine whether and how often intercourse takes place. In addi-
tion, these same factors {society, family, peers) also affect the
individual's perceived probability of pregnancy and cognitive assess-
ment of pregnancy and well as attitudes toward contraception and some
portion of experience with contraceptives.
METHODOLOGICAL ISSUES
A variety of measures have been used to study contraceptive use.
One of the first issues to address is at what point should contracep-
tive use be measured: first intercourse, last intercourse, in last
month, currently? There appears to be no apparent rationale for using
one or the other measure. One distinction does appear to be important.
That is the distinction between use of contraception at first inter-
course and continuing use of contraception. Contraceptive use at
first intercourse {Table 2.2) is poorer than current use (Table 2.8~;
this is the case for those whose first intercourse occurred when they
were under 18 as well as those whose first intercourse occurred when
they were 18 and older (Table 3.3; Figure 3.1~. Nonuse at initial
intercourse and during the first months of sexual activity appears to
be problemmatic: the risk of pregnancy is highest in those months
(Zabin et al., 19801. The decision to first use contraception appears
to be an important one. And the factors associated with initiating
contraception may differ from those associated with continuing contra-
ceptive use, once initiated. Thus the second major concern will be a
discussion of the factors associated with continued contraceptive use
including the difficulties associated with using it, change from one
method to another, and the effectiveness and consistency of use.
A second important issue is which of the variety of measures of
current use that have been utilized by researchers are most appro-
priate. Measures include whether ever used contraception, frequency
or regularity of use, use at last intercourse, use during last month.
Since these measures are all relevant to one concept, continuing use,
the results of the research will be discussed together. One study
(Herold, 1980} comparing these various measures will be discussed.
Another issue is how to measure the type of contraception used and
its effectiveness. Questions involve not only the time point at which
use is measured (first, last intercourse, current use) but also whether
and how methods are grouped: e.g., medical (pill, IUD) vs. non-medical
methods, prescription (pill, IUD, diaphragm) vs. non-prescription
methods; more effective vs. less effective methods; rank ordering of
effectiveness; scale score based on use effectiveness.
OCR for page 60
60
One of the major problems with the research in contraception
appears to be its static nature. Contraceptive use at one point in
time may not be a good proxy for use over the period of a year or
more. Panel data that bears on this issue will be described.
Another important issue is that of comparability with a sample of
similar older women. Where possible this chapter compares the results
for teenagers with those for older women to see if and where the
findings differ.
Finally, most of the analytic work so far has been based on data
collected by Kantner and Zelnik (cf. Zelnik et al., 1981) in 1971,
1976 and 1979. Although the National Survey of Family Growth con-
ducted in 1982 was released in 1984, so far no analytic studies of the
type that would show factors associated with contraceptive decision-
making have been completed. There is no reason to believe that the
factors associated with contraceptive initiation and continuation have
changed over time. Therefore, it makes sense to discuss the results
obtained from that earlier survey as a baseline for understanding.
Unfortunately, neither the National Survey of Young Women nor the
National Survey of Family Growth provides the kind of information
about potential rewards and costs of pregnancy and the perceived
probability of pregnancy that could shed light on details of the
theoretical model. Thus most of the studies discussed here are
reduced form models with no intervening variables. This is not
problemmatic as long as it is recognized that background factors are
not expected to have substantial direct effects on contraceptive
initiation and continuation. To fill in the detail on intervening
factors this review focuses on a few small scale studies that have
collected such information. Although these studies are generally
conducted on highly selective samples, the information they provide is
useful in evaluating the model.
RESEARCH
Frequency of Sexual Intercourse
There is only one study so far to explore factors associated with
the frequency of sexual intercourse among never-married teenagers.
Zeloik et al. (1981) found that white teen women who reported that
they had marriage plans, who used a medical contraceptive method at
last intercourse, and who had had 4 or more partners had a higher
frequency of intercourse than their peers. For black teen women,
having had a greater number of partners and having used a medical
method at last intercourse were associated with a greater frequency of
sexual intercourse.
OCR for page 61
61
Probability/Risk of Pregnancy
What is the n best" way to measure contraceptive use depends on the
purpose of the study. If researchers wish to measure how adequately
individuals protect themselves from an unwanted or unplanned pregnancy,
they would first need to adequately measure motivation to avoid
pregnancy. Some individuals, for example, may be trying to get
pregnant. Others may be infertile for a variety of non-contraceptive
reasons. Finally, others may not care if they become pregnant or
not. Some individuals may be knowledgeable about their menstrual
cycle and may contracept actively only during the high risk parts of
the cycle. In this case it would be helpful to know how accurate the
individual's knowledge is about the menstrual cycle and at what time
of the cycle the questions are asked.
Table 2.10 shows perception of risk of pregnancy to be poor among
teenage women, even among those sexually experienced. In 1976 only
two out of five could name the time of month of greatest pregnancy
risk. On the other hand, a number of contraceptive methods do not
require any knowledge of time of greatest risk, and some studies sug-
gest that users of such methods are least knowledgeable about timing
of pregnancy risk (Presser, 19777. Thus it may not be critical to
know exactly when pregnancy is most likely to occur to use contracep-
tion effectively, only that the risk is relatively high. The associa-
tion between knowledge about pregnancy risk and use of contraception
is not very strong, though it has been found in a few studies
(Philliber et al., 1983; Cvetkovich and Grote, 1980~. The relationship
between knowledge of methods and contraceptive use is also not very
strong (Flaherty and Marecek, 1982; Poppen, 1979~. This appears to be
because most teens can name several methods. What kind of knowlege
leads to earlier and more consistent use of contraception is an issue
that needs additional exploration
Cognitive Assessment of Pregnancy
How do teenagers feel about pregnancy and childbearing? Do they
want to become pregnant? Do they not care? Do they not want to avoid
it strongly enough to contracept adequately? What information do we
have about this issue?
There is, of course, an important distinction between childbearing
intention and wantedness of a child. In particular, most young child-
less women eventually do want children. Thus many teenage pregnancies
are timing failures; this does not imply that children are not wanted
ever. A child may be wanted, but not at that time. Three categories
of child-bearing intentions at the time of conception are commonly
distinguished:
1) Timing success - a child is wanted at that time;
2) Timing failure - a child is wanted, but not at that time;
3) Unwanted - a child is not wanted at that time or ever.
OCR for page 62
62
Of those young women who become pregnant as teenagers, about 45
percent have abortions (excluding miscarriages from the total) (see
Pregnancy Resolution). Presumably abortions represented unwanted
pregnancies. Early studies showed that of those teens who experienced
a first birth as a teenager, only about half (45 percent) could be
considered intended or timing successes, that is, both wanted ever and
wanted at that time (Trussell and Menken, 1978~. The majority (55
percent) of first births to teenagers, represented either timing
failures or unintended or unwanted births. Thus, of the total
pregnancies to teenagers, only about 25 percent were intended, that
is, wanted at that time. This study was conducted at a time when a
large proportion of teenagers married. The proportion of pregnancies
that could be considered intended has probably draped, but no recent
research was available to address this question.
Initiating Contraceptive Use
What proportion of youth used a contraceptive method at first
intercourse? Among those who did not use contraception at first
intercourse, when did they begin using contraception? What methods
were used at first intercourse among users? What methods did later
initiators first use?
Table 2.1 shows that in 1979 almost half of all teen women reported
that they used a contraceptive method at first intercourse. The per-
centage of teen women in 1982 who used a method at first intercourse
is identical to that in 1979. There appears to have been no major
shift between 1979 and 1982 in use at first intercourse. This repre-
sents a substantial improvement over 1976, when 40 percent of teen
women said that they contracepted at first intercourse (Zelnik and
Kantner, 1978~.
The pattern of contraceptive use at first intercourse varies sub-
stantially by age at that time {(unpublished tabulations from the NSFG;
Zelnik and Shah, 19831. Under a third of women and men who first had
intercourse at under 15, half of those who first had intercourse at 15
to 17 and three-fifths of those who first had intercourse at age 18 or
older used a contraceptive method at first intercourse. There is very
little sex difference. The race difference is sharper; blacks are
less likely than whites to have used a method at first intercourse.
However, in an analysis of the 1976 National Survey of Young
Women, Zeloik and colleagues (1981) found that after adjusting for
other differences between young women, in particular the age at first
sex, current age, SES and family stability the race difference was not
statistically significant. The difference in use of a contraceptive
method at first sex by age at the time remained highly significant;
women who were older at first intercourse were much more likely than
women who were younger to have used a method at that time.
OCR for page 63
63
Mosher and Bachrach (1986) conducted a multivariate analysis on
data from the 1982 National Survey of Family Growth. They found that,
for women 15-44 in 1982, there was a substantial race difference in
use of contraception at first intercourse--whites were much more likely
than blacks to use a contraceptive method at that time. When they
controlled for differences between blacks and whites in education of
mother, year of first intercourse, ethnicity/religion, age at first
intercourse and whether ever discussed menstrual cycle with a parent,
that race difference declined slightly but did not disappear. This
analysis differs from that of Zelnik et al. in that 1) it was not re-
stricted to teenagers, and 2) it did not control for family stability
or religiosity, which were included in the previous analysis. Thus we
cannot draw any conclusions as to whether a change has occurred such
that there is now a true race difference in contraceptive use at first
intercourse net of other differences, where there was not one in 1976.
An analysis comparable to that of Zelnik et al. (1981) should be
conducted on the 1982 NSFG or other recent data.
Zelnik et al. (1981) focused primarily on the influence of back-
ground factors on use of contraception at first intercourse. A few
relationships were significant, but overall the relationships were not
very strong. Young women with better educated parents and in intact
families were more likely to have contracepted at first intercourse.
Surprisingly, women younger at the time of the survey were also more
likely to have contracepted at first intercourse. Given that age at
first intercourse is also controlled, this suggests that more recent
cohorts of young women, particularly blacks, are more likely than
early cohorts to use contraception at first intercourse.
What proportion of those who used a method at first intercourse
used a prescription method--the pill, IUD or diaphragm? Use of such
methods requires planning and a doctor or clinic visit. According to
data from the 1979 National Survey of Young Women black teenagers are
much more likely than white teenagers to report having used a prescrip-
tion method at first intercourse: 41 percent compared to 15 percent
of whites (Zelnik and Shah, 1983~. Even when other factors are con-
trolled, using the data from a similar 1976 survey, Zelnik et al.
(1981) found the black-white differences to be large and statistically
significant. Age at first intercourse continues to be important, with
those older at first intercourse more likely to have used a medical
method at that time. Again, only a few variables were associated with
use of a prescription method, and the total proportion of variation in
contraceptive use explained by these variables was very small.
This analysis of medical methods points out some of the pitfalls
of relying on one analysis of contraceptive use. For example, the
condom, which is not a medical method and was therefore not included
in the analysis discussed above, is effective if used properly. Is an
analysis, therefore, adequately measuring contraceptive use at first
intercourse if condom use is ignored? Table 2.3 shows the distribution
of contraception used at first intercourse among all sexually ex
OCR for page 64
64
perienced respondents and among users only. Comparing black and white
female users in 1979 (Table 2.3) we see that white women users relied
heavily {three quarters to four-fifths) on a male method (condom and
withdrawal); in contrast, half of black female users reported using
female and half reported using male methods, with most of the female
methods being prescription methods. The reports of white male users
are very similar to those of white female users, with 7 out of 10
reporting use of a male method at first intercourse. However, the
reports of black male users are very different from those of black
female users. Of the former, 6 or 7 out of 10 say that they used a
male method at first intercourse. Of course, there is no reason these
figures have to be the same. In fact, the data suggest that the first
partner is older and, therefore, probably already experienced. Data
are similar for 1982 (Table 2.4), except that among users the use of
female prescription methods at first intercourse appears to have de-
clined among black female teens and the use of the condom has in-
creased. Sample differences make this conclusion tenuous, however.
The planning status of first intercourse does appear to be related
to use of contraception (Table 2.31. Young women who planned their
first intercourse were more likely to have used contraceptive methods
than those who didn't plan it, but the differences are small. Young
men who planned their first intercourse were more likely to have used
a male method.
Table 2.5 shows reasons respondents gave for not using a method at
first intercourse. Under 5 percent said they wanted a pregnancy or
didn't care; another small proportion thought pregnancy impossible.
Almost 20 percent said they didn't know about contraception. The
majority said it was unavailable, didn't think about it, or didn't
want to use it.
Contraceptive Use in the First Six Months After Sexual Debut
Half of all initial premarital pregnancies occur in the first six
months of sexual activity (Zabin et al., 1979), and more than one-fifth
in the first month. Yet data show that teenagers delay coming to a
clinic--the average delay is about 1 year after initiation of sexual
activity. What is the pattern of contraception during the period?
Research on the sequence of contraceptive use from first inter-
course is greatly needed. Table 2.1 shows the percentage distribution
of premaritally sexually active women age 15-19 by contraceptive use
status and race in 1982, 1979 and 1976. About one-third of the young
women in 1979 reported using contraception at first intercourse and
using it consistently thereafter. Fifteen percent contracepted at
first intercourse, but not always, 25 percent did not use a method at
first intercourse but did at some time afterwards, and 27 percent
claim to have never used contraception. Since 1979 there has been an
apparent decline in the proportion of teen women who never used con
OCR for page 65
65
traception and an increase in the proportion who used it at sometime,
although not at first intercourse.
Of course, Table 2.1 does not show us how long after first inter-
course it takes to adopt contraception. Table 2.6 shows, for those
who did not use a method at first intercourse, but who had ever used a
method by the interview date in 1982, the length of time between first
intercourse and first contraceptive use. It is clear that the older
the age at first intercourse, the quicker a young woman adopts contra-
ception. Twenty-two percent of those under 15 compared to 53 percent
of those 18-19 at first intercourse adopted contraception within one
month; 42 percent of the former and only IS percent of the latter de-
layed more than one year. Among those who do not use contraception at
first intercourse, blacks were consistently slower to adopt contracep-
tion than whites. This difference is smallest among the earliest
initiators, and is surprisingly large among later initiators {although
sample sizes are small). Since blacks were also less likely than
whites to have used contraception at first intercourse, blacks appear
to be at much higher risk than whites of a pregnancy at or soon after
first intercourse. In 1976, on average those 15-17 year olds who did
not use contraception at first intercourse did so within about 6 months
(excluding those who did not use contraception by the time of the sur-
vey or prior to pregnancy or marriage). Among those teenagers visiting
clinics, the average delay in 1980 was more than a year (Kicker, 1984'
This figure understates the average delay since many (41.7 percent of
all 15-19 year olds) used no contraception between first intercourse
and either a pregnancy, marriage or the survey date. The most imp
portent reason young women gave for delaying a family planning visit
to a clinic was that they simply didn't get around to it. Two other
major reasons cited were n fear of family discovery" and n relationship
with partner not close enough.
What factors would be expected to be related to contraceptive use
at first intercourse? 1) correct perception of pregnancy risk and time
in the month of greatest risk, 2) motivation to avoid pregnancy. So
far there has been little attempt to measure these concepts and test
their association with contraceptive use at first intercourse or rela-
tionship to contraceptive adoption soon after. Table 2.10 shows that
sexually inexperienced women are less likely than experienced women to
correctly perceive the time of greatest pregnancy risk in the menstrual
cycle.
There is very little work on adoption of contraception after first
intercourse--either the process or the types of methods used. Very
little research has even categorized young women by length of time
since f irst intercourse. Research to sort out this process, preferably
using a life table methodology, is needed.
OCR for page 66
66
Contraceptive Continuation
For this review three aspects of contraceptive continuation are
important: 1) whether currently using, 2) regularity of use and 3)
effectiveness of method.
Current Contraceptive Use
Probably the most common measures are 1) whether used a contracep-
tive method at last intercourse, and 2) whether currently using a con-
traceptive method. The reference point for the second measure is
usually specified as during the last month or last 4 weeks preceding
the survey and is specified as among those sexually active. Some
analysts, in addition, eliminate those who cannot become pregnant and
those who are trying to become pregnant. Some studies utilize the
first and others utilize the second measure. Comparisons of the
results obtained (for example, Tanfer and Horn, 1985) show that the
distributions are very similar. So it makes sense to consider these
two together.
There was substantial improvement in current contraceptive use
between 1971 and 1982. Slightly under half of all women 15-19 used a
method at last intercourse in 1971; close to 2/3 of all teens used a
method at last intercourse in 1976 {Zelnik and Kantner, 1977~. 1982
data show that 71 percent of teens exposed to the risk of an unwanted
pregnancy are currently practicing contraception (Bachrach, 1984~. In
1971 black teenagers appeared to be similar to whites in contraceptive
practice. When other factors were controlled, such as differences in
socioeconomic status and in family structure, no significant difference
between blacks and whites in use at last intercourse remained {Zelnik
et al., 19811.
One of the most important factors associated with currently using
contraception are current age and length of time sexually active.
Zelnik et al. (1981) controlled simultaneously for current age and for
age at first intercourse. Thus age at first intercourse really rep-
resents length of time since first intercourse. Of two young women
with similar ages, the one who became sexually active earlier has been
active longer. Zeloik et al. {1981) found that women who were older
at first intercourse (and therefore had been sexually active less time)
were more likely to contracept at last intercourse. Using the same
data set, Devaney and Hubley (1981) found current age but not age at
first intercourse to be associated with contraceptive use. That is,
older teens were more likely to be contracepting at last intercourse
than younger teens. The Devaney and Hubley study included a larger
set of control variables, which may explain the difference in findings.
Studies using the National Survey of Young Women found frequency
of intercourse (Devaney and Hubley, 1981) and number of partners
(Zelnik et al., 1981) to be associated with use of contraception at
OCR for page 67
67
last intercourse. A women who was engaged to be married (Devaney and
Hubley, 1981) and one who had been pregnant {Zelnik et al., 1981) were
likely to be contracepting. A history of past use was associated with
current use. Teens who contracepted at first intercourse (Zelnik et
al., 1981) and those who had few unprotected months after first method
use (Philliber et al., 1983) were more likely to be contracepting at
last intercourse.
Family background and interpersonal factors contribute to use of
birth control. Young women with better educated parents were more
likely to be contracepting (Zelnik et al., 1981~. A young woman in a
family receiving welfare was less likely to have contracepted at last
intercourse than a young women not in a family receiving welfare
(Philliber et al., 1983~. Living in group quarters was associated
with greater use of contraception at last intercourse (Devaney and
Hubley, 1981~. Finally, Philliber et al. (1983) found that the more
persons who knew about the teenager's birth control use the more
likely she was to be contracepting at last intercourse.
One factor strongly associated with use of contraception at last
intercourse is educational expectations. The higher the educational
expectations, the more likely a young woman is to have used contra-
ception at last intercourse (Devaney and Hubley, 1981), a relationship
that holds for whites and blacks alike. This variable may be a proxy
for motivation to prevent pregnancy.
Philliber et al. (1983) tested the Luker formulation of a decision-
making model on youth attending a New York City youth center. In-
cluded in the model were a series of socioeconomic background charac-
teristics as well as a series of variables measuring perceived advan-
tages and disadvantages of pregnancy and perceived pregnancy risk.
She found that using effective contraception at last intercourse was
associated with a high score on the subjective probability of preg-
nancy, a low score on perceived advantages, a high score on perceived
disadvantages of pregnancy, a low estimate of the probability of using
abortion if pregnancy occurs, and a low rating of disadvantages of
birth control. Those with a high level of ego development were also
more likely to use effective contraception. Few background factors
added significantly to the model. This research provides strong
support for the value of a decision-making model. Further research
should address the issue of the factors affecting individual percep-
tions of consequences and the ways these affect behavior that is, the
mechanisms whereby background factors lead to differential contracep-
tive use.
Regularity of Contraceptive Use
Regularity of contraceptive use has been measured in a variety of
ways. Flaherty and Marecek (1982) divided their sample into contracep-
tors versus non-contraceptors. The former reported use of contracep
OCR for page 68
68
Lion either always or usually, the latter reported rarely using con-
traception, and were using nothing at the time of interview. Poppen
(1979) and Cvetkovich and Grote (1980) used two measures: 1) ever
versus never had unprotected intercourse and 2) frequency of con-
traceptive use in the last three months {always, almost always,
usually, sometimes, never), including all methods, even rhythm.
Nathanson and Becker's definition of contraceptive use was the pro-
portion of time subsequent to the baseline interview that a respondent
at risk of pregnancy (sexually active and not pregnant) was using a
medical method of contraception (mostly oral contraceptives) (Nathanson
and Becker, 1985~. Ager (1982) defined contraceptive non-use as 1)
program discontinuance or 2) method discontinuance. The former refers
to whether still in program or not; the latter refers to those not
continuously practicing contraception, compared to those who were
either continuously practicing effective contraception or were not at
risk over the interval. Zelnik et al. (1981) divided women into two
groups: ever and never users of contraception.
Unfortunately, except for Zelnik et al. (1981), none of these
studies controlled for length of time since first intercourse, and
only a few controlled for current age. Since these respondents are at
all stages of sexual experience the results should be taken with cau-
tion. Zelnik et al. (1980) showed that both age at first intercourse
and current age are related to frequency of contraceptive use. Net of
current age, age at first intercourse measures length of time or dura-
tion since first intercourse. Results of their study show that the
older the current age and the longer the time since first intercourse,
the less likely a teenager is to have always contracepted. The former
represents a cohort effect: older teens were born before younger
teens; more recent cohorts appear to use contraception sooner than
earlier cohorts. The latter represents pure length of exposure effect.
The longer the period in which to have contracepted, the less likely
the respondent is to have always contracepted.
In another study which controlled for background factors as well
as characteristics of the respondent and current relationship and
which used a different data set, Furstenberg (1983) did not find either
duration of exposure to pregnancy risk or current age to be associated
with contraceptive continuation (continuous contraceptive use over 15
months as measured by two different measures). This suggests that
factors included by Furstenberg and not by Zelnik et al. (1981), such
as parental employment, n steadiness" of the relationship, academic
performance, school/employment status, and convenience of method, may
explain the impact of current age and length of time since first
intercourse on contraceptive continuation. However, this is purely
speculation, since the research to test such an hypothesis has not
been conducted.
None of the studies cited showed significant black-white differ-
ences in frequency of contraceptive use. Although one study suggested
that sexually active blacks may be slightly more likely to have ever
OCR for page 69
69
contracepted (Devaney and Hubley, 1981), the evidence certainly is not
sufficient to reject the conclusion that contraceptive regularity
differs little by race.
Cvetkovich and Grote (1980) found that black males whose mothers
have a high education use contraception more regularly; Zelnik et al.
(1981) found a similar effect of parental education on the contracep-
tive use of black females. Flaherty and Maracek (1982) found that
girls who talked with their mothers and cited the mother as a source
of birth control information used contraception more regularly; how-
ever, she did not control for other differences between young women.
Fox (1980) found a similar effect of maternal communication on
daughters' contraceptive use, but it disappeared with controls for
other factors such as socioeconomic background and family structure.
Flaherty and Marecek (1982) found type of maternal discipline associa-
ted with frequency of contraceptive use. In particular, daughters who
had experienced parental rules and punishments for violations and for
whom restrictions of privileges were favored over corporal punishment
for misbehavior were more likely to use contraception. Again, however,
these researchers did not control for potentially confounding factors.
Thus these results should be taken as suggestive, not definitive.
One important hypothesis is that young women who don't perceive
opportunities other than motherhood open to them will be less motivated
to prevent pregnancy than women who perceive better opportunities for
jobs and careers. Nathanson and Becker (1983) found that black girls
who perceived better opportunities for non-reproductive roles were more
likely to be regular contraceptors than those who didn't. However, the
association was weak for black teens and there was no association for
white teens. They did find that the older the preferred marriage age,
the greater the continuity of contraception among white and black teen-
agers (Nathanson and Becker, 19831.
Religion doesn't appear to be related to frequency of contraceptive
use. Flaherty and Marecek (1982) showed a positive effect of religios-
ity in general. Devaney and Hubley (1981) found that blacks who said
that religion was important to them were more likely to have ever-used
contraception than others. In contrast, whites who said they attend
church regularly were less likely to have ever-used contraception.
Their particular religious affiliation was not associated with ever
having used contraception. However, using the same data, Zelnik et
al. (1981) did not find either religiosity or religion to be associa-
ted with ever having used contraception, net of other factors.
One study based on samples from two urban areas and one small town
found that knowledge about sex and contraception were associated with
greater frequency of contraceptive use among white males {Cvetkovich
and Grote, 19801. Fear of side effects of contraception and perception
of harmful effects of contraception were associated with less frequent
use (Popper, 1979; Cvetkovich and Grote, 1980~. Self-esteem was asso-
ciated with more frequent contraceptive use for both whites and blacks
(Cvetkovich and Grote, 1980~.
OCR for page 70
70
For a discussion of the effects of the clinics themselves on con-
traceptive use, see Programs and Policies, Chapter 9.
Contraceptive Effectiveness
The final measure of contraceptive continuation to be discussed is
contraceptive effectiveness, that is, the level of effectiveness of the
contraceptive method the women/couple used at last intercourse or is
currently using. Effectiveness is measured by the failure rate. It
is the proportion of women exposed to the risk of unintended pregnancy
who would become pregnant if they used a given method and no other for
one year. Measuring effectiveness therefore, requires a knowledge of
method used and an estimation of its associated failure rate. There
are several potential ways of measuring failure rate. The theoretical
or biological failure rate is impossible to measure but is often ap-
proximated by the lowest failure rate measured among different popula-
tion groups. The actual use effectiveness rate, in contrast, is the
average failure rate for the group to which the individual belongs,
which will be higher than that of the theoretical rate.
Table 2.11 shows use failure rates for unmarried women over the
period 1979 to 1982, from the National Survey of Family Growth (Grady
et al., 19861. Failure rates vary substantially by age, contraception
method, duration of exposure, poverty ratio income, race, parity and
contraceptive intention of the woman. In general failure rates are
higher the younger the woman. Among those who intend to delay a birth,
failure rates are highest for those 18-19 years old, not for those
under 18. This is probably due to the fact that 18-19 year old single
women have the highest frequency of intercourse (see Chapter 1~. There
was not enough information in the NSFG to control for frequency of
intercourse. Among those who seek to prevent a birth entirely, failure
rates are highest among those under age 18, and decline gradually to
their lowest level among women 30-44. Surprisingly, failure rates are
higher up to age 29 among women who seek to prevent an additional birth
rather than those who seek to delay a birth. Grady et al. ( 1986)
speculate that either 1) young women redef ine their intentions at the
time of conception after the fact, or 2) young single women who are
preventing a pregnancy are doing so because they have already had one
or more unwanted births. Those who intend to prevent at a young age
may be ineffective users and high risk nonusers, while the older age
groups would include many post married women who had successfully
regulated their fertility. Black women have no higher risk of un-
intended pregnancy than women of other racial groups; the similarities
are probably exaggerated since abortions are underreported more by
black than white women. Women who have had children have higher
failure rates than those who don't. Such women may be more fecund.
Finally, failure rates are higher among women with lower incomes;
those below the poverty level have the highest failure rate.
OCR for page 71
71
Grady et al. (1985) found that the failure rates for single women
were somewhat lower than those for married women. They hypothesized
that this difference was due to the fact that abortions were under-
reported more by single than by married women. When they adjusted the
use-failure rates to take into consideration this reporting difference,
the use-failure rates for single and married women were more similar.
Comparing single with married women, failure rates for the former are
higher for the pill, lower for the condom and use of no methods. It
is likely that the differential frequency of intercourse among married
women explains their greater failure rates for the condom and no
method, while greater inconsistency of pill use explains the greater
failure rates of pill use among unmarried compared with married women.
The lowest possible failure rates measured among different population
groups are presented in Grady et al. (1986~.
Although other factors, such as interpersonal relationships and
contraceptive attitudes may also affect use effectiveness, no research
has been conducted using such factors to distinguish effectiveness.
In practice, almost no one has simply taken these theoretical or
use effectiveness scores and used them as a dependent variable.
Contraceptives are generally rank ordered by effectiveness and then
grouped into categories such as medical and non-medical or effective
and non-effective. An example is that of Zelnik et al. (1981) who
distinguished medical (prescription) from non-medical (non-prescrip-
tion) methods. An example of research using effectiveness ratings as
a dependent variable is that of Polit et al. (1981~. The dependent
variable n at risk to pregnancy" combines four types of information:
a) the percentage of time that the couple used a method of birth
control, b) the theoretical effectiveness of the methods used, c) the
use-effectiveness of the methods used and d) assessments of how effec-
tively this couple used their contraception. The final score ranged
from 3 to 90, with 3 representing low risk (e.g., consistent users of
the pill) and 90 representing highest risk of pregnancy (e.g., no
birth control at all).
Current age and age at first intercourse predict use of effective
contraception. Older teens are more likely to have used a medical
method at last intercourse (Zelnik et al., 1981~. The longer the
duration of exposure to sexual activity (the younger the age at first
intercourse) the more likely to have used a medical method. Young
black women with better educated parents are more likely to have used
a medical method at last intercourse (Zelnik et al., 1981~. Net of
other factors, blacks are more likely than whites to have used a medi-
cal method at last intercourse (Zelnik et al., 1981~. Not listing a
religion and not being religious are associated with a higher probabil-
ity of using a medical method at last intercourse (Zelaik et al.,
1981).
These authors also found that young women who have been pregnant
are more likely to use a medical method at last intercourse (Zelnik et
OCR for page 72
72
al., 1981~. In contrast, Polit et al. (1981) found that couples in
which the female had been pregnant were less effective contraceptors.
Characteristics of the relationship are important predictors of
effectiveness of contraception. The greater the number of partners,
the more committed the relationship with the partner, and the more
frequently intercourse occurs, the more effective the contraception
(Thompson and Spanier, 1978; Herold, 1980; Zelaik et al., 1981~. The
greater the likelihood of couple continuity (boy's report) and the
higher the communication rating, the more effective the contraception
(Polit et al., 19811.
Attitudes toward sex and contraception have been found to affect
the effectiveness of the contraception couples use. Herold (1980)
found that teen females who have favorable attitudes toward contracep-
tion, a low level of embarrassment about contraception, and a low
level of premarital sex guilt are more likely to use effective contra-
ception.
Beliefs about who should have responsibility for birth control
have also been found to be associated with more effective contracep-
tion. Polit et al., (1981) found that teen women who believe that the
female should take responsibility for birth control tend to be more
effective contraceptors. Although the same authors (Polit et al.,
1981) hypothesized that males and females who were rated high on
decision-making skills would be better contraceptors, in fact, she
found no difference among females and that couples in which the male
had good decision-making skills were less effective contraceptors. She
concluded that, contrary to her hypotheses, contraception is more suc-
cessful when one person, particularly the female, takes responsibility
for it.
Some Methodological Issues
Presumably the reason that we are interested in measuring contra-
ceptive use is that we want to know the risk of unwanted or unintended
pregnancy that women incur. Thus it would seem important to validate
these measures. To what extent do they measure what they purport to
measure that is, pregnancy risk. Herold (1980) attempted to validate
the different measures of contraceptive use by looking at their inter-
correlation and by looking at the association of each variable with
other potential determinants. He concluded that the weakest measure,
that is, the one with fewest associations with other measures of
contraceptive use and apparently with the most random variance, is
whether a young women ever used any method at all. The strongest
measures are those of effectiveness. Unfortunately, Herold did not
explore which of these measures best predicts avoidance of pregnancy.
Two studies have explored the relationship between these measures
and eventual pregnancy {Koenig and Zelnik, 1982; and Furstenberg et
OCR for page 73
73
al., 1983~. Both studies show that women who claim to be using con-
traception consistently, that is, who always use a contraceptive
method at intercourse, have a much lower likelihood of pregnancy than
women who are not consistently using a contraceptive method, than
women who are currently using a contraceptive method, and than women
who have never used a contraceptive method. Distinguishing between
medical and non-medical methods helped discriminate levels of pregnancy
risk only slightly. Unfortunately, however, Furstenberg (1982) also
discovered considerable inconsistency in respondents' reports of con-
traceptive use. Using a measure of consistency of use based on retro-
spective accounts 15 months after the initial survey, 73 percent of
clinic patients surveyed were consistent users. However, there were
substantial inconsistencies in reporting of use between the 6 month
and the 15 month interviews. If those whose reports at the two time
points are inconsistent are not included with consistent users, then
the percent who are consistent users drops to 43 percent. The authors
concluded that "Retrospective measures contain a large element of
error because of the respondents' inability or unwillingness to recall
past use accurately. Without more frequent interviews and a good deal
of probing regarding inconsistent reponses, investigators are likely
to exaggerate rates of contraceptive continuation" (Furstenberg et
al., 1983:217~.
Prediction of pregnancy was strongly affected by the measure of
contraceptive continuation used (Furstenberg et al., 1983~. The most
refined measure, the measure that included prospective as well as
retrospective reports of contraceptive use over a 15 month period, was
the best predictor of pregnancy and was also best explained by a number
of characteristics of adolescents; random error appeared to be mini-
mized. The authors concluded that better measures of contraceptive
use are needed before factors affecting that use can be identified and
addressed.
Males
Very little contraceptive research has included males. Few
national surveys have collected information on male contraceptive use
among adults, let alone among teenagers. The only survey that included
young men, the 1979 Zelnik-Kantner National Survey of Young Men, 17-21,
has not been fully analyzed. The research there is suggests that know-
ledge about and attitudes toward birth control, as well as self-esteem,
are very important to contraceptive use as reported by males (Cvet-
kovich and Grate, 1980~. Unfortunately, knowledge is poor. In one
study of teen males (Finkel and Finkel, 1975), fewer than one-half
could correctly identify the time during the menstrual cycle when
conception is likeliest to occur. Attitudes and knowledge of males
toward contraception are also less than favorable. In her study of
contraceptive decision-making in adolescent couples, Polit et al.
(1981) found that, in general, men were less knowledgeable about
specific contraceptives and had less favorable attitudes toward con
OCR for page 74
~4
7
traceptive use than women. In particular, more males than females
thought oral contraceptives were dangerous. In spite of these
findings, males do appear to use effective (male) methods at first
intercourse (Table 2.5~. They also appear to use male methods cur-
rently or at last/most recent intercourse, as reported by their part-
ners (Table 2.8~. No data are yet available on current use or use at
last/most recent intercourse as reported by males.
Partners
..
Among adults, both partners have been found to have an independent
contribution to the contraceptive decision; however, the important
decision-maker is still the woman. In studies primarily of college
students, Thompson and Spanier (1978) and Herold (1980) found partner
influence very important in contraceptive use. Very little research,
however, has been conducted using teenage couples. Polit (1983) found
that the couples best protected against unwanted pregnancy were those
in which one person, (usually the female) took charge. Luker (1975)
found male partners to be perceived by women as frustrated and feeling
relatively powerless in preventing pregnancy (p. 133J:
Men are socially cast into the role of passive spectators to
contraceptive decision making by the same social and technological
changes that structure the way in which women take risks. If both
reponsibility and accountability are defied as exclusively female,
men have neither the social means nor the personal motivation to
take more active interest.
However, her sample was comprised of women seeking abortion; the
feelings and perception of male frustration and powerlessness are
likely to be heightened in this sample.
Polit et al. (1981) found that the agreement between partners'
responses to questions about contraceptive practice ranged from 60 to
90 percent, the correlations between responses ranged from .18 to
.90. These results suggest substantial variation between teen part-
ners' reports of contraceptive use.
Experience with Contraceptives
It could be expected that individuals' experience with contracep-
tion would affect subsequent behavior. It has been hypothesized, for
example, that perceived/experienced side effects of the pill (as well
as adverse publicity) resulted in the apparent decline in pill use
between 1976 and 1979 among teenagers (see Table 2.8), and its re-
surgence between 1979 and 1982 may be due to more recent reports that
show very low risks for teenagers and some positive effects, such as
protection from some cancers. As important as side effects are, only
one project has studied reported side effects and contraceptive use
OCR for page 75
75
(Mindick and Oskamp, 19803. This study could not adequately test the
association because the authors did not statistically correlate the
reported side effects with reports of actual contraceptive use, but,
rather, the likelihood that a relationship between side effects and
unwanted pregnancy or clinic discontinuation existed was rated by
project personnel based on clinic records, and these ratings were then
used as a dependent variable for analysis.
SUMMARY AND CONCLUSIONS
This review has been organized around a rational decision-making
model of contraception, in which the proximate determinants are fre-
quency of intercourse, perceived probability of pregnancy, willingness
to use abortion as backup if pregnancy occurs, positive advantages and
disadvantages of pregnancy and contraception, and positive and nega-
tive experience with contraception. Although research is still rela-
tively scarce in this area, what there is supports the importance of
such factors in distinguishing degree of contraceptive use. The
family appears to have little influence on contraceptive use; few
studies have examined peer influence. Societal influence may operate
through accessibility and availability. A recent study suggests that
broader societal attitudes and values re sex and contraception may
also affect contraceptive use (Jones et al., 1985~; no other research
was identified on this issue.
In this review contraceptive use was divided into two major areas
of substantive interest: initiating use and continuing use. There is
very little research on contraceptive use at first intercourse; only
two studies deal with the process of initiating contraceptive use
after first intercourse (Koenig and Zeloik, 1982; Zabin et al., 1979~.
Initiating contracepive use is heavily influenced by the age of the
young woman at the time. The younger the woman the less likely to
have used contraception at first intercourse and the longer the delay
before initiating use. In addition, recent birth cohorts of women
appear to be more likely to use contraception at first intercourse
than early cohorts. There are no race differences in contraception at
first intercourse, once other factors such as socioeconomic status are
controlled.
Background factors had only weak influences on the initiation of
contraception. Young women with better educated parents and in intact
families were more likely to have contracepted at first intercourse.
Black women are more likely to have reported using a medical method at
first intercourse. Women who planned their first intercourse were
more likely to have used a contraceptive method than those who didn't
plan it.
There is substantially more work looking at contraceptive continua-
tion, current use of contraception, regularity of use, and effective-
ness of use. There is some evidence that those who become sexually
OCR for page 76
76
active at a young age are less likely to be currently contracepting,
less likely to have always contracepted, but more likely to be using a
medical method currently. Frequency of intercourse and being engaged
to be married are associated with a higher likelihood of contracepting
at last intercourse and with more effective contraception.
Women who perceived a greater probability of pregnancy, more dis-
advantages and fewer advantages of pregnancy, more advantages of birth
control and who were less likely to say that they would resort to abor-
tion if they became pregnant were more likely to report having used
contraception at last intercourse and to report having used effective
methods.
One small scale study found knowledge about sex and contraception
to be associated with greater frequency of contraceptive use among
males and females. The relationship between previous pregnancy and
contraceptive use is not clear as one study found such young women to
be more likely to use a medical method and the other found such young
women less effective contraceptors.
One of the most important problems in studying contraception is
the appropriate measurement of contraceptive use. First, very little
research has attempted to validate the various measures of contracep-
tive use standardly used. Second, the validation studies that have
been conducted have found substantial inconsistencies in reports of
contraceptive use. This is especially crucial for looking at consis-
tency of use over time. The results of such studies suggest that one
of the reasons reseachers have had so much trouble identifying factors
associated with good or the contraceptive practice may be the poor
reliability of the measures of contraceptive use. This is especially
important for determining use effectiveness, which, at the present
time, is based solely on individual reports of contraceptive use.
In conclusion, birth control use is largely a function of current
age and age at first intercourse, relationship with partner, perceived
risk of pregnancy, acceptance of abortion, attitudes toward contracep-
tion, desirability of pregnancy, and experience with contraception.
It is only weakly related to knowledge about contraception. The
family plays a relatively small part. Society may affect the availa-
bility and accessibility of contraception, which may affect contracep-
tive practice. The part peers play is not well known. Why some youth
are effective contraceptors and others are not is still a little
researched issue. Research suggests that women of all ages have
trouble with contraception in actual use; teenagers differ only
slightly from their 20-24 year old unmarried peers~in practice. How-
ever, since their pregnancy rates also appear to be higher, how good
our understanding is of actual contraceptive practice is in question.
Nor do we know the full extent of differences in pregnancy or contra-
ceptive failure rates by age.
OCR for page 77
77
This paper concludes with a list of important issues:
1. What is the process of becoming a user and a regular user?
This issue has not been explored although it is extremely important.
2. How should contraceptive use be measured? More studies are
needed to better measure actual contraceptive use.
3. What is the contribution of male partners to effective contra-
ception? This is an important question but one that only a few in-
vestigators (Finkel and Finkel, 1975; Polit et al., 1981; Shea and
Freeman, 1983; Shea et al., 1983; Cvetkovich and Grote, 1980) have
explored.
4. Are there race/ethnic group differences in contraceptive
practice? Although Zelnik et al. (1980) show distinct black-white
differences in contraception at first intercourse, ever used con-
traception, and contraception at last intercourse, these differences
disappear when background factors are controlled. There is no differ-
ence between blacks and whites in use at first or last intercourse or
ever use, net of other factors. However, one difference does hold
up. Black users are more likely to use a medical method at first and
at last intercourse, probably due to greater clinic access for blacks.
Blacks and whites do differ in types of methods used, with blacks more
likely to use medical methods. Recent data suggest differential use
effectiveness. Differences by ethnicity (e.g., Hispanic background)
are small probably due to the fact that Hispanic teenagers are more
likely than other teenagers to be married.
5. Is having been pregnant associated with better or poorer con-
traception? Zelnik et al. (1980J found that, net of other factors, a
previously pregnant woman was more likely to use a method at last
intercourse, and to use a medical method. Polit et al. (1981) in
contrast, found that, net of other factors, a previously pregnant
woman was a less effective contraceptor, measured in terms of
effectiveness of current contraception.
6. Do actual or perceived opportunities and alternatives affect
contraceptive use? It is commonly hypothesized that they do; however,
very little research has been conducted on this issue. The research
that has been conducted shows little support for the hypothesis.
Representative terms from entire chapter:
failure rates
