Relationship Characteristics and Feelings About Pregnancy Among Black and Puerto Rican Young Adults

Abstract

CONTEXT

Unintended pregnancy is common among black and Hispanic young adults in the United States. How pregnancy intentions form and change is poorly understood, although research indicates that intentions and attitudes are dependent on partners’ views and other relationship factors, and are different by gender.

METHODS

A sample of black and Puerto Rican men and women aged 18–25 from low-income neighborhoods in two cities were surveyed in 2007–2008. Using data on 520 serious and casual sexual relationships reported by 460 respondents, generalized ordered logistic regression analysis was conducted to identify individual- and relationship-level correlates of how respondents would feel if they became involved in a pregnancy with a particular partner.

RESULTS

About one-quarter of respondents reported each of four possibilities of how they would feel about a pregnancy with a particular partner—very upset, a little upset, a little pleased and very pleased. In 45% of relationships, respondents believed that their partners would be very pleased about a pregnancy, whereas they themselves would be very pleased in only one-quarter of cases. Overall, women were less likely to feel positive about a pregnancy than were men (odds ratio, 0.3). Respondents’ positive feelings about their relationships were associated with a strong tendency toward more positive feelings about a pregnancy (2.1), as was a measure of how positive respondents thought their partners would feel (1.5–2.6). The latter association was particularly strong among women (1.7).

CONCLUSIONS

Relationship characteristics were associated with feelings about pregnancy for both genders. Future research should utilize a more comprehensive framework for conceptualizing and examining sexual relationships.

In the United States, unintended pregnancies are common among young people. In 2006–2010, among women aged 20–24 who had given birth in the previous three years, half reported that their most recent pregnancy had been unintended (17% that it had been unwanted, 22% mistimed by two or more years, and 12% mistimed by less than two years).¹ Unintended pregnancy rates vary among subgroups of the population; they are significantly higher among women whose income is less than 150% of the federal poverty level than among those living at more than four times the poverty level, and higher among blacks and Hispanics than among whites.¹ Unintended pregnancy can have various disadvantages related to both public health outcomes (e.g., later prenatal care, less breast-feeding and higher abortion rates) and public finance (e.g., high delivery costs).¹ It may also present setbacks for some men and women, by affecting their financial, educational and personal circumstances.^2–4 For these reasons, national Healthy People 2020 goals include lowering the unintended pregnancy rate, increasing the use of contraceptives and reproductive health services, and expanding financial access to such services.⁵ However, in the last decade, the prevalence of unintended pregnancy has been relatively unchanged.¹

Pregnancy intentions and desires are complex and involve multiple dimensions.^6,7 The 2010 National Survey of Family Growth included 11 questions that could be used to assess pregnancy intention. In addition to the traditional questions related to the wantedness of pregnancy at a particular time, the survey asked respondents how they felt about a pregnancy, whether they thought their partner wanted the pregnancy, and how much they planned or acted to avoid or seek pregnancy.¹ The concept of ambivalence about pregnancy—mental states on the spectrum between clearly planned and clearly unwanted—has also emerged as conceptually and analytically important, to better reflect women’s and men’s feelings and perspectives.^8–12 Notably, the National Survey of Family Growth also asks men directly about their intentions for pregnancy involvement with their partner, reflecting years of research showing that men’s views not only are often different from their female partners’, but also have been associated with some maternal and child health outcomes (e.g., timing of prenatal care, birth weight).^13–17

Given the conceptual complexity of pregnancy intention and desire, it is no surprise that they are also thought to be influenced by a host of variables, ranging from individual characteristics to broader contextual ones.¹⁸ Of particular interest to this study is the focus in recent years on how relationship characteristics may be related to pregnancy intention and desire. Marital status has often been used as a proxy for various aspects of relationships.¹⁹ Other research has demonstrated more direct ways to measure the attributes of relationships and sex partners, and to assess how they are related to pregnancy intention and desire. In qualitative studies, the status and quality of a relationship, as well as hopes for the relationship, have been prominent in respondents’ explanations of their pregnancy intentions.^2,3,20 For example, Higgins et al. found that in a sample of socially advantaged and disadvantaged men and women, ambivalence about pregnancy, and the inconsistent contraceptive use that followed, sometimes reflected emotional intensity or a desire to cement a relationship with a particular partner, even when one or both partners did not clearly want a pregnancy because of other issues.²¹

In quantitative research, various aspects of pregnancy intention or desire (the wantedness of a recent pregnancy or birth, feelings about a recent pregnancy or birth, or desire for a pregnancy in the near future) have frequently been associated with relationship characteristics. Based largely on data from women reporting on their relationships with men, these studies have found associations with measures of relationship satisfaction,¹⁰ quality^22–24 and duration;^24,25 marital status, including cohabitation;^12,24,26,27 age, racial or ethnic differences between partners;²⁷ partner’s desire for a pregnancy;^26,28 and partner’s provision of emotional or financial support.^24,26 Most results have pointed in the expected directions: Lower relationship quality, stability and duration were associated with a lower desire for pregnancy. However, there have been exceptions. For example, in Wilson and Koo’s multivariable analysis, neither relationship duration nor a global measure of relationship quality was associated with a current desire for a pregnancy with a particular partner.²⁴ Rijken and Thomson found mixed results in the Netherlands: Women in medium-quality relationships were more likely than those in low- or high-quality relationships to have a first child with their male partner.²³

Men’s views have been increasingly represented and explored in research on pregnancy intention and desire, but few studies have examined how relationship characteristics are related to men’s thinking about pregnancy. A couples study among youth aged 14–21 identified various similarities and differences in partners’ desire for a current pregnancy.²⁸ For example, both men and women believed their partner wanted the pregnancy more than they did, but women were more accurate than men in assessing their partner’s intentions. This and other studies raise questions about whether and how men take relationship characteristics into consideration when thinking about pregnancy. Some qualitative studies have found that men’s feelings about their partners and relationships featured prominently in their pregnancy intention and desire, alongside economic and other factors.⁸ In contrast, some men have been characterized as living within norms of masculinity that may devalue feelings or frame pregnancy as primarily about taking financial responsibility for one’s children.^29,30 Given that men and women experience pregnancy and parenting differently (both biologically and socially), characteristics associated with pregnancy desire may differ for men and women.

Our study follows this line of inquiry, and examines how relationship characteristics are associated with pregnancy intention, as measured by how the respondent would feel if a pregnancy occurred with a particular partner. The sample consists of urban, young adults who are black or Puerto Rican—demographic groups at high risk of unintended pregnancy.¹ Moreover, unlike many studies, ours employed multiple measures of relationship characteristics, as well as data from both men and women.

METHODS

Data Source

We analyzed data from the Philadelphia and Hartford Research and Education on Sexual Health and Communication (PHRESH) project, which was funded by the Centers for Disease Control and Prevention and carried out from 2004 through 2008 by the University of Connecticut and the Family Planning Council of Southeast Pennsylvania. The study was approved by the institutional review boards of the Centers for Disease Control and Prevention and partner institutions.

For this study, we relied on a quantitative survey conducted in 2007–2008 as part of this larger project. Study teams recruited participants from Philadelphia and Hartford neighborhoods with relatively high STD and teenage pregnancy rates and relatively large populations of blacks or Puerto Ricans. Criteria for participation were self-identification as Puerto Rican, African American or black; being aged 18–25; not being pregnant; having been born on the U.S. mainland or in Puerto Rico; being fluent in English; and having had sex with a heterosexual partner in the preceding six months. By design, the survey recruited roughly equal numbers of males and females, and of Puerto Rican and black participants.

The study used a time-place sampling methodology.^31,32 In each city, study teams mapped neighborhood sites at which members of the target population visited or gathered on a regular basis (e.g., laundromats, bus stops, corner stores and parks). Teams used that information to create time-space sampling frames from which they randomly selected recruitment sites. Most venues were eligible for selection for all sampling periods, which were scheduled during the day, on weekdays and weekends. During these periods, which were generally about four hours long, research staff approached individuals who appeared to be from the target population and assessed their eligibility for participation.

In Hartford, the sampling frame comprised 41 venues distributed across nine neighborhoods; 478 people were screened. Of these, 108 were deemed ineligible, 76 eligible people declined to participate and 53 did not show up for interview appointments (scheduled because of inclement weather on the day of recruitment); thus, 241 individuals (65% of those eligible) completed interviews. The Philadelphia study team created three sampling frames, two for blacks and one for Puerto Ricans; the frames crossed seven zip codes and included 207 venues. Overall, 325 people were screened. Seventy-four were deemed ineligible, and nine eligible individuals declined to participate; 242 men and women (96% of those eligible) completed interviews. In both cities, interviews lasted 20–30 minutes each, and participants were compensated $35.

Survey Content

Using closed-ended items, the survey asked participants about all partners in the last six months, and questioned them extensively about their most recent serious and casual heterosexual partners (as defined by respondents, but explicitly excluding one-night stands). For these two types of partners, we had information about relationship duration (in months); whether the relationship was ongoing at the time of the survey; the couple’s frequency of sex over the past six months; whether they had had a child together; whether they were living together; whether the respondent believed that his or her partner had had other partners since the couple had been together; the couple’s commitment to monogamy; the respondent’s feelings about the relationship; how the respondent would feel about a pregnancy with the partner; and how the respondent thought the partner would feel about a pregnancy.

The measure of commitment to monogamy distinguished respondents who reported that both they and their partner wanted to be exclusive within their relationship (regardless of whether they actually were) from those who reported that at least one of them did not want to be exclusive; the latter respondents were further identified by whether they would have been upset or would not have cared if their partner was not monogamous.

Relationship feelings were assessed on a scale created from multiple survey items. We began with nine measures that we felt were most relevant. Using results from a varimax, rotated factor analysis, we focused on a single factor, containing five items, which had an eigenvalue of 3.48 and for which each item loaded at 0.70 or greater. The items related to how much the respondent cared about the partner, how much the respondent thought the partner cared about him or her, how important the relationship was to the respondent, how likely it was that the couple would be together in one year, and how much the partner “was there” for the respondent when she or he needed it. Response options ranged from 5 (most negative) to 23 (most positive). Cronbach’s alpha for the five-item scale was 0.88; we used a standardized version of the scale in the analyses. Responses were grouped into quartiles; the most positive were assigned to the fourth quartile.

The measure of feelings about pregnancy reflected how participants would feel if they or their partner got pregnant then; the four response categories were “very upset,” “a little upset,” “a little pleased,” “very pleased” and “don’t know.”

Other respondent characteristics used in this analysis were gender, race or ethnicity, age, educational attainment, parity, having had more than one partner in the last six months and the importance of avoiding a pregnancy then. Marital status was not assessed, given the low prevalence of formal marriage in these populations. For relationships that had ended by the time of the survey, participants were instructed to respond to relevant questions in terms of how the relationship was while they and their partners were together.

Analysis

In selecting covariates and models, we examined both continuous and categorical forms of variables that could be represented both ways (i.e., relationship duration, frequency of sex, relationship feelings, age, partner’s feelings about a pregnancy, number of children). For each variable, the results were similar, so we selected the most parsimonious form to include in the models, which in most cases was the continuous form. For the same reason, we dichotomized the respondent’s parity into any children versus none. Given the potential for high correlation among some of our covariates, we also systematically assessed various combinations of related variables for inclusion, examining how effect sizes and statistical significance changed in the presence of the other variables. In one case, because two variables effectively canceled each other out, we retained the more pertinent one (i.e., we included “ever had a child” and dropped “had a child together”). Every variable was examined in a multivariable model, regardless of the significance of bivariate results; those variables that remained nonsignificant in both bivariate and multivariable models were eventually dropped from the multivariable analysis (e.g., education and age).

We combined serious and casual relationships into a single data set so that relationships were the units of analysis. We excluded 10 relationships because one partner was sterilized. Sixty respondents reported on both a casual and a serious relationship, and thus contributed two observations to our analytic data set; we adjusted standard errors for the regression results for this clustering, using Stata’s cluster option.

We present descriptive statistics and results from multivariable regression analyses that used the respondent’s feelings about a pregnancy with the index partner as the dependent variable. In first attempting ordered logit regression models, we found that the assumption of proportional odds was violated. After assessing alternative approaches, we adopted a generalized ordered logit model. This model uses partial proportional odds to allow the estimates for variables that meet the proportional odds assumption to remain constant (as in a typical ordered logit), while the estimates for variables that do not meet the proportional odds assumption can vary. We applied this model using Stata’s user-derived command gologit2, via the autofit option, which uses an algorithm based on iterative Wald tests to ascertain which variables included in a regression meet the proportional odds assumptions at p<.05.^*33

We show two multivariable models, one with no interaction terms and one assessing the interaction of gender with the relationship feelings scale and with the partner’s feelings toward a pregnancy. We hypothesized that the associations of both measures would be stronger for women than for men, following the conventional notion that women more strongly value emotional aspects of relationships and are more influenced by their partner’s interest in a pregnancy.

RESULTS

Sample Characteristics

After excluding survey participants who did not report a recent serious or casual partner and relationships in which a partner was sterilized, the sample included 520 relationships reported by 460 young adults (Table 1). Nearly six in 10 participants were aged 18–21, and the remainder were 22–25. Forty-nine percent had at most a high school education or GED, and 13% had attended at least some college. Thirty-two percent had had more than one sexual partner in the last six months. Half of respondents did not have any children, and 78% said it was very or extremely important to avoid pregnancy right then.

TABLE 1.

Percentage of young adults, and of their sexual relationships, by selected characteristics, Philadelphia and Hartford, 2007–2008

Characteristic	%
RESPONDENTS	(N=460)
Female	50
Race/ethnicity
Black	50
Puerto Rican	50
Age
18–21	58
22–25	42
Educational attainment
<high school	38
High school/GED	49
≥some college	13
Had >1 partner in past 6 mos.	32
No. of children
0	50
1	29
≥2	21
How important to avoid pregnancy now
Not at all	8
A little	5
Somewhat	9
Very	37
Extremely	41
RELATIONSHIPS	(N=520)†
Duration (mos.)
0–3	18
4–6	12
7–12	12
13–24	17
25–36	14
≥37	26
Ongoing at time of survey	82
Couple’s monogamy commitment‡
Both wanted to be exclusive	69
≥one did not want; respondent would be upset if partner not monogamous	13
≥one did not want; respondent would not care if partner not monogamous	18
Frequency of sex in past 6 mos.
Daily	15
Several times/week	41
Once/week	16
Several times/month	14
Once/month	6
<once/month	8
Had a child together	28
Cohabiting	25
Believed partner had not been monogamous since relationship began	41
Respondent’s feeling if pregnancy occurred
Very upset	24
Little upset	23
Little pleased	26
Very pleased	25
Don’t know	2
Partner’s feeling if pregnancy occurred‡
Very upset	14
Little upset	16
Little pleased	23
Very pleased	45
Don’t know	2

^†Because of missing data, Ns range from 500 to 520.

^‡According to respondent.

Note: Percentages may not add to 100 because of rounding.

The median duration of relationships was 17 months (not shown); 18% of them had a duration of three months or less, and 26% had a duration of more than three years. Eight in 10 relationships were ongoing at the time of the survey. In 69%, the respondent reported that both partners wanted to be exclusive. On average, the frequency of sex in these relationships was high: In 56%, respondents reported having had sex either daily or several times a week over the past six months. In 28% of relationships, the couples had had a child, and in 25%, they were living together at the time of the survey. In 41% of relationships, the respondent believed that his or her partner had had sex with someone else during their time together. Overall, respondents reported largely positive feelings toward their partners and these relationships. On the five-item scale, for which response options ranged from 5 (most negative) to 23 (most positive), the median score was 20 (not shown).

Respondents varied widely regarding how they would feel about a pregnancy with their partners. Roughly a quarter of responses fell into each of the four categories. In contrast, in 45% of relationships, respondents believed that their partners would be very pleased. Data on 504 relationships included responses for both the respondent and the partner; in 37% of these, respondents thought that their partners would feel more pleased about a pregnancy than they would, and in 49%, respondents thought that their partner would feel the same as they would (not shown).

Bivariate Findings

After the exclusion of relationships in which respondents said they did not know how they or their partner would feel about a pregnancy, as well as relationships for which data were missing for variables used in the multivariable analysis, the analytic sample consisted of 498 relationships reported by 439 respondents. Black and female respondents generally felt less positive about a pregnancy than their Puerto Rican and male counterparts (Table 2). Feelings about pregnancy did not vary significantly by age-group or educational attainment (not shown). Having had multiple partners in the past six months was associated with generally more negative feelings toward a pregnancy. Thirty-one percent of respondents who already had a child reported that they would feel very upset about a pregnancy, compared with 18% of those without a child; however, the same proportion of each group said they would feel very pleased (26%).

TABLE 2.

Percentage distribution of relationships, by respondents’ feelings about pregnancy with a particular partner, according to selected characteristics

Characteristic	Very upset	Little upset	Little pleased	Very pleased	Total
Gender*
Female	30	24	24	23	100
Male	20	22	29	29	100
Race/ethnicity*
Black	28	25	26	21	100
Puerto Rican	21	21	28	31	100
No. of partners in past 6 mos.***
1	18	23	28	31	100
≥2	34	23	24	19	100
No. of children**
0	18	26	30	26	100
≥1	31	20	23	26	100
Relationship duration (mos.)***
0–3	41	19	14	26	100
4–6	34	30	20	16	100
7–12	21	18	37	24	100
13–24	22	32	25	21	100
25–36	18	21	27	34	100
≥37	15	20	35	30	100
Couple’s monogamy commitment**
Both wanted to be exclusive	16	23	30	31	100
≥one did not want; respondent would be upset if partner not monogamous	35	20	24	21	100
≥one did not want; respondent would not care if partner not monogamous	52	23	16	9	100
Frequency of sex in past 6 mos.***
Daily	18	12	23	47	100
Several times/week	17	25	30	28	100
Once/week	27	24	32	17	100
Several times/month	27	27	29	17	100
Once/month	48	32	6	13	100
<once/month	48	14	19	19	100
Had a child together
Yes	21	21	27	31	100
No	26	23	27	24	100
Cohabiting***
Yes	10	20	27	42	100
No	29	24	27	20	100
Partner’s feeling if pregnancy occurred***,†
Very upset	67	15	13	4	100
Little upset	18	52	14	16	100
Little pleased	13	20	53	15	100
Very pleased	19	16	22	42	100
Relationship feelings scale***,‡
1st quartile	54	22	13	12	100
2nd quartile	24	31	30	14	100
3rd quartile	16	19	34	31	100
4th quartile	9	19	27	45	100

^*p<.05.

^**p<.01.

^***p<.001.

^†According to respondent.

^‡The most positive responses were assigned to the fourth quartile.

Notes: Chi-square tests were used to assess significant differences in percentage distributions across categories. Percentages may not add to 100 because of rounding.

Associations between feelings about a pregnancy and relationship duration were significant, but did not follow consistent patterns. In 31% of relationships in which both partners wanted to be monogamous, respondents said they would feel very pleased about a pregnancy. In contrast, in only 9% of relationships in which at least one partner did not want to be monogamous and the respondent would not care if their partner had sex with someone else did individuals report feeling very pleased.

Relationships characterized by relatively high frequencies of sex and by cohabitation were both associated with more positive feelings about pregnancy. For example, in 42% of relationships in which partners cohabited, the respondents said they would feel very pleased about a pregnancy, whereas this was reported in only 20% of non-cohabiting relationships.

Respondents’ perceptions of how their partners would feel about a pregnancy were positively associated with their own feelings. In relationships in which respondents thought that their partners would be very upset by a pregnancy, respondents were most likely to say that they themselves would be very upset, whereas in relationships in which partners would be expected to be very pleased, respondents were most likely to feel that way themselves.

Finally, scores on the relationship feelings scale were positively associated with feelings about pregnancy. For example, in relationships that received the lowest scores, 54% of respondents said they would be very upset about a pregnancy and 12% said they would be very pleased. In contrast, in relationships with the highest scores, 9% said they would be very upset and 45% would be very pleased.

Multivariable Findings

For characteristics for which results did not vary across outcomes (e.g., gender, race and ethnicity, multiple partners), odds ratios from the generalized ordered logit regression are interpreted in the same way as ordered logit results: A value above 1.0 is associated with a tendency toward more positive feelings about pregnancy with a particular partner, and a value below 1.0 is associated with a tendency toward more negative feelings. For characteristics for which results did vary across outcomes (e.g., ever had a child and relationship duration), odds ratios are interpreted in a similar way but indicate how the tendency differs across feelings categories.³³ For example, a particular characteristic may be associated with large differences in the likelihood of respondents’ reporting feelings on one end of the spectrum, yet may show no significant differences on the other end. Thus, contrary to multinomial logit results that look similar to these, odds ratios less than 1.0 do not indicate a reduced likelihood of being in a particular feelings category.

In the multivariable model excluding interactions, many of the associations identified at the bivariate level remained, and they accounted for approximately 18% of the variation in the outcome (Table 3, model 1). Women were more negative about a pregnancy than men (odds ratio, 0.3). Compared with respondents who did not have a child, those who did were more likely to say they would feel very or a little upset, as opposed to pleased, if a pregnancy occurred in their current relationship (0.3 and 0.6, respectively).

TABLE 3.

Odds ratios from multivariable generalized ordered logit regression analysis assessing how respondents would feel about a pregnancy with a particular partner, by selected characteristics

Characteristic	Model 1				Model 2
	Nonproportional odds			Proportional odds	Nonproportional odds			Proportional odds
	Very upset	Little upset	Little pleased		Very upset	Little upset	Little pleased
Respondents‡
Female				0.3***				0.1***
Black				0.7†				0.7†
Ever had a child	0.3***	0.6*	1.2		0.3***	0.6*	1.2
Multiple partners in past 6 mos.				0.8				0.8
Relationships
Duration	1.2*	1.1	0.9		1.2*	1.1	0.9
Couple’s monogamy commitment
Both wanted to be exclusive (ref )				1.0				1.0
≥one did not want; respondent would be upset if partner not monogamous				0.9				0.9
≥one did not want; respondent would not care if partner not monogamous				0.6†				0.6
Frequency of sex in past 6 mos.				1.2*				1.2*
Cohabiting‡				1.9**				1.9**
Partner felt positive about a pregnancy with respondent§	1.5**	1.8***	2.6***		1.1	1.4*	2.1***
Relationship feelings scale				2.1***				2.3***
Interactions
Female × relationship feelings scale								0.8
Female × partner’s positive feelings about a pregnancy§								1.7**
Pseudo-R2				0.18				0.19

^*p<.05.

^**p<.01.

^***p<.001.

^†p<.10.

^‡Dichotomous measures; all other characteristics for which no reference group is shown are continuous.

^§According to respondent.

Notes: For proportional odds, estimates reflect a tendency toward or against more positive feelings about a pregnancy. For nonproportional odds, estimates greater than 1.0 indicate an increased likelihood of being in a more positive feelings category, and estimates less than 1.0 indicate an increased likelihood of being in the current or more negative feelings categories. See text, Multivariable Findings, for further explanation. ref=reference group.

Relationship duration was associated only with expecting to feel more positive than very upset about a pregnancy (odds ratio, 1.2). Reporting a higher frequency of sex and cohabitation remained associated with generally more positive feelings toward pregnancy (1.2 and 1.9, respectively). The more positive respondents thought their partners would feel about a pregnancy, the more positive they felt themselves, and this association strengthened across the feelings categories (1.5–2.6). Finally, the higher the score on the relationship feelings scale, the more positive a respondent’s expected feeling about pregnancy (2.1).

When we tested the interaction between gender and the relationship feelings scale (model 2), the result was non-significant, indicating that the size and nature of the associations between the scale and pregnancy feelings were similar for men and women. In contrast, the interaction between gender and the perceptions of partner’s feelings about pregnancy was large and significant (odds ratio, 1.7), indicating that female respondents who thought that their partner would have positive feelings about a pregnancy tended to have positive feelings themselves. In other words, the women in the sample are driving the positive association between respondents’ feelings and their view of their partner’s feelings about a pregnancy. Associations for other measures remained largely unchanged in the interaction model.

DISCUSSION

This analysis found that numerous individual and relationship measures were independently associated with how this sample of high-risk young adults would feel about a pregnancy, supporting the broad notion that many proximate determinants of pregnancy intention and desire may operate at the relationship level.

A direct, composite measure of respondents’ feelings toward a partner and their perceptions of their partner’s feelings about a possible pregnancy were strongly and positively associated with the respondents’ own expected feelings toward a pregnancy. It is not surprising that these measures were correlated, but these associations have not been measured or identified in much previous quantitative research. Past research has often used measures of pregnancy intention or desire that are more global rather than partner-specific, or has had fewer measures of relationship characteristics to draw from. Studies that have assessed relationship characteristics and feelings more directly have used various measures and sometimes identified different associations. For example, in Wilson and Koo’s analysis of desire for pregnancy with a specific partner, the primary measure of relationship quality was a composite of questions related to exclusivity, satisfaction and communication, and it was not significant in multivariable analysis.²⁴ Rijken and Thomson measured relationship quality with a 13-item scale that included components related to support, conflict and overall feelings;²³ it was associated with having a first birth (a reasonable proxy for an intended pregnancy in a Dutch sample) among women, and a second birth among men, but in different patterns. Similarly to what we found, Blake et al. found that among pregnant women, if the father had desired the pregnancy, women were more likely to have felt happy about it themselves.²⁶ Sipsma et al. reported similar associations for men and women in a sample of adolescents.²⁸

Our assessment of gender interactions provides additional evidence to support the inclusion of men in research on pregnancy intentions and desire. Contrary to our expectations, the results suggest that men’s feelings about the relationship, like women’s, were positively associated with their feelings about pregnancy. Perhaps for many of these men, pregnancy desire was part of a larger relationship narrative related to love and intimacy. The high prevalence of longer term relationships in our sample may have biased our results toward this particular finding. Moreover, research among adolescents found that pregnancy desire had a stronger association with relationship satisfaction among women than among men.²⁸ Future research can extend these findings by examining young men’s perspectives on relationship quality and intimacy.

Contrary to a study by Sipsma et al., we found that the association of the partner’s feelings toward a pregnancy was stronger for female than for male respondents.²⁸ This, along with the finding that women were generally more negative about a pregnancy than men, points to some of the gender dynamics of pregnancy and parenting. Perhaps women interpreted their perceptions of men’s happiness about a pregnancy as evidence that the relationship was ready for a serious commitment or a child (or another child), when in fact men did not feel this way. Women may also have interpreted their perceptions as strong signs that those men would provide needed, ongoing emotional and material support, which might have made pregnancy more welcome to them. The degree to which individuals take on their partners’ views (consciously or subconsciously) about whether and when to have a child is an important area for future research. It not only reflects potentially important gender norms, but also could prove to be an area that merits increased attention in programs aiming to reduce unintended pregnancy.

We also found that in multivariable models, the measure of partner exclusivity was not associated with feelings about a pregnancy, while living together was positively correlated. Other research has found similar results for cohabitation, though in some studies, it was one of the primary measures of relationship status or quality.^12,27 In our sample, cohabitation was independently and positively associated with feelings about a pregnancy, even when feelings about the relationship and other relationship dimensions were accounted for. In this sample, cohabitation could not only represent the opportunity and presence of resources to raise a child jointly, but also indicate that a relationship was more serious or stable, and thus more prepared to support a pregnancy and parenting.

Limitations

Participants were black or Puerto Rican young adults recruited in low-income neighborhoods in two cities; therefore, they likely shared, to some degree, socioeconomic and life stage characteristics. This may help explain why age and education were not significant. Clearly, the findings are not generalizable to all young adults; results of analyses of more diverse samples may differ. Nevertheless, it is valuable to have information about these groups, as they are among the ones at highest risk of unintended pregnancy.¹

Another limitation is that the survey did not provide much information about respondents’ partners, such as their age, parity, economic status or relevant attitudes. In addition, the information was based on respondents’ points of view. Potentially important dimensions of relationships were not expressly represented in our data, such as those related to conflict, power, control and communication. Moreover, we lacked measures for many variables that may relate to respondents’ attitudes and feelings about pregnancy, such as their outcome expectations, plans and opportunities.

Our main outcome measure differed from that used in some other research on pregnancy intention and desire: It assessed respondents’ reactions to a hypothetical pregnancy with a particular partner and may not reflect what they would have felt in response to a real pregnancy. This measure may have been particularly vulnerable to respondents’ wishful thinking about their relationship and where it might lead; as such, it may have been essentially another measure of participants’ feelings about the relationship. (In fact, Sipsma and colleagues used pregnancy desire to predict relationship satisfaction.²⁸) On the other hand, this measure is not subject to modifications that might result when individuals are asked about pregnancies retrospectively, and it aligns with previous research that highlights the relevance of inquiring about pregnancy intention and desire in a partner-specific way.^1,19

Conclusions

Overall, this study helps demonstrate that relationships matter to pregnancy intention and desire, for both men and women, and that researchers need to expand their examination of relationships to include multiple dimensions. We need quantitative research that, instead of representing relationships in an ad hoc way or through distant proxies, utilizes a comprehensive framework for conceptualizing relationships and more standardized measurements. By drawing from qualitative research about relationships and pregnancy intention, from psychometric research related to relationships and from analyses like this one, it is possible to develop such a framework, as well as valid measurements for use in future research. Such efforts could lead to a better understanding of pregnancy intentions and desires, and enable pregnancy prevention programs to become more responsive to how young adults form their feelings and attitudes toward pregnancy.