PATTERNS OF MATE SELECTION AMONG THE ELDERLY: DO THE HOMOGAMOUS NORMS APPLY?
Norma A. Winston and G. Jeffrey Klepfer, University of Tampa
Using data gathered from 9325 marriage licenses filed in a Florida county in 1995, we examined the impact of age at marriage, marital parity and marital status on the age differences between spouses. We compared the results for those under age 55 with those 55 and older. Conclusions were drawn about the applicability of age-related homogamous norms in mate selection to the elderly. Results indicated that younger brides and grooms married spouses more similar to themselves in age than did older brides and grooms. Also, we found that a greater number of previous marriages was associated with greater spousal age differences for younger but not older brides and grooms. Finally, our results demonstrated a relationship between previous marital status (i.e., single, divorced or widowed) and age homogamy, again, for younger, but not for older, brides and grooms. We point to an imbalanced sex ratio, a cultural double standard and differing mate selection needs as possible explanations for these age homogamy associations.
A substantial number of marriages take place in later life (while the number of marriages per thousand population aged 55 and older is smaller than that of any other marriageable age category, it is large enough to warrant study. In 1990, the number of marriages among those aged 55+ was 14.1 per 1000 population aged 55+). Despite this, comparatively little attention has been given to the characteristics of the mate selection process among the older cohorts (see, for instance, Burch, 1990; Dressel, 1980). Social scientists generally agree that mate selection is influenced by cultural and social structural factors. But do the same factors operate for all persons or are there differences by age cohort and associated stage in the life cycle?
Theoretical Perspectives
The most commonly quoted theoretical perspective on mate selection is that of homogamy or assortive mating (e.g., Burgess & Wallin, 1963). This perspective presumes that the normative structure operates through cultural conditioning to direct people to select a mate with characteristics similar to their own. It follows then that individuals who marry are expected to be close in age. An underlying assumption of homogamy is that greater differences in age between spouses will lead to higher probabilities of marital instability.
A second perspective, less common than homogamy, involves the concept of the marriage squeeze (Elder & Rockwell, 1976). This perspective presumes that as the supply of men declines and the demand for them increases, women have little choice but to broaden the age range of acceptable partners (Spanier & Glick, 1980). This perspective is particularly applicable to the older cohorts where the sex imbalance becomes increasingly marked with age (the sex ratio is computed as the number of males per 100 females. The overall sex ratio in the older population is 65 men per 100 women in the year, 2000. Among those 85+ the sex ratio is 38 men per 100 women in the same year (Atchley, 2000, p. 30)). Adams (1979) and Becker (1981) offer variants on the "marriage squeeze" as it affects the older population. They suggest that as people move beyond the typical age for marriage, the pool of eligibles is significantly reduced and homogamous norms become less relevant. As a result, a person may marry someone whose age differs greatly from their own because they do not expect to do better by further searching or waiting (Becker, 1981, p. 232).
The third, and least commonly used perspective, emphasizes the double standard of aging. The assumption here is that women are much more likely to be penalized for aging than are men and that these differences affect preferences in mate selection (Bem, 1993; Wolfe, 1990). Some indicators that the double standard does operate include much more variability in men’s age at marriage compared to that of women, a greater frequency of older man/younger woman marriages and less social support for older woman/younger man unions (Cowan, 1984).
Previous Research Findings
The effect of the following three independent variables on age differences between spouses was focused on in this research: age at marriage, marital parity (number of previous marriages) and previous marital status (single, divorced or widowed). Since little research has focused on mate selection among the elderly specifically (e.g., Dressel, 1980), relevant findings must be extrapolated from studies based on samples spanning a much broader age range.
With respect to the age variable, Glick and Landau (1950), Hollingshead (1951 ) and Presser (1975) reported that age differences between singles who marry tend to increase with an increase in the age of each spouse. This not the case among those who have been previously married. Marital parity is, of course, closely associated with age, since an individual married for the second, third or subsequent time is usually older than one marrying for the first time. Thus by virtue of increasing age, with second or subsequent marriages, homogamous norms weaken and age differences between spouses increase (Glick & Landau, 1950; Presser, 1975).
There is disagreement among researchers as to whether age at marriage or marital parity is more important in determining age homogamy. Despite this, "scholars agree that as the marriage market among people of older ages decreases in size, age differences increase regardless of the sequence of the marriage" (Breckenridge, 1989, p. 17). That is, marriages become more heterogamous in terms of age.
Breckenridge (1989) found gender differences in the effects of age at marriage and marital parity on age heterogamy. The husband’s age at marriage explained a much larger amount of the variation in age differences between spouses than did the wife’s age at marriage. Likewise, the husband’s marital parity had a much greater impact on age heterogamy than did the wife’s marital parity. These findings provide some support for the theoretical perspective of the double standard. In addition, they suggest the importance of looking closely at the impact of sex differences on the mate selection process.
In the case of previous marital status, Bowerman (1953) theorized that the homogamous norms would operate such that "like would choose to marry like" (p. 176-177). However, Bowerman found this to be the case in marriages contracted between singles only. Among the previously married, the tendency toward homogamy decreased as age increased. Thus, older divorcees and widow(er)s were less likely than those in younger cohorts to marry others of similar status.
Hypotheses
Given the above, we predicted that (1) for grooms there would be an inverse relationship between age and age homogamy, (2) for grooms there would be an inverse relationship between marital parity and age homogamy and (3) there would be no relationship between previous marital status and age homogamy.
Method
The data were gathered by examining all marriage certificates recorded (i.e., marriages that actually took place) in Hillsborough County, FL, for the year 1995 (N= 9325). Ninety-five percent of the sample was under 55 years of age while the remainder (5%) was 55 and over. Of those under 55, 8770 were males and 9009 were females, while of those aged 55-plus, 551 were males and 308 were females. We tested each of the hypotheses on the whole sample initially, and then made comparisons by gender between those under 55 and those 55 and older, when appropriate. We tested homogeneity of variance using Levine’s method for all two-sample comparisons.
Results
Age at Marriage and Age Homogamy
We hypothesized that older grooms would evidence less age homogamy than younger grooms and, indeed, found this to be the case. Groom age correlated positively with spousal age difference (r [9314] = +.468, p < .001), accounting for 21.90% (r²) of the variance in age difference. An independent sample t-test supported this finding when spousal age difference was compared for grooms younger that 55 years of ages (N = 8794) and those 55 or older (N = 522). Younger grooms differed in age from their brides by a mean of 4.53 years (S.D. = 4.39), while those older differed by 11.17 years (S.D. = 8.73): t (9314) = 17.25, p < .001, r² = .031.
Although we did not predict a relationship between age and age homogamy for brides, we found an association similar to that for grooms. Bride age correlated in a significantly positive direction with spousal age difference (r [9314] = +.221, p < .001), accounting, however, for considerably less of the variance in age difference (r² = .049). An independent sample t-test of the relationship between bride age and spousal age difference revealed that brides younger than 55 (N = 9020) were more homogamous in age (M = 4.81, S.D. = 4.84) than were brides 55 years and older (N = 296, M = 7.73, S.D. = 7.61): t (9314) = 6.56, p < .001, r² = .005. Figure 1 shows mean spousal age differences for both younger and older brides and grooms.
Marital Parity and Age Homogamy
Secondly, we hypothesized that an increase in marital parity would be associated with a greater spousal age difference (heterogamy) among grooms. This relationship also was evidenced in our data (r [9314] = + .293, p < .001), with marital parity accounting for 8.59% (r²) of the variance in spousal age difference.
A one-way analysis of variance (ANOVA) testing the effect of groom's marital parity on spousal age difference demonstrated a significant relationship between these variables: F (2, 9313) = 511.50, p < .001. The mean age differences for first (N = 5450), second (N = 2655) and third or subsequent (N = 1211) marriages were 3.64 (S.D. = 3.78), 6.21 (S.D. = 5.48) and 7.72 (S.D. = 6.44), respectively. The percent of variance in spousal age difference accounted for by groom marital parity was 9.90% (h ²). Post hoc comparisons of mean differences were analyzed using independent sample t-tests and all differences were found to be significant. For the difference between first and second marriages, this analysis yielded t (8103) = 21.76, p < .001; between first and third or subsequent marriages, t (6659) = 21.24, p < .001; and between second and third or subsequent marriages, t (3864) = 7.08, p < .001.
We further investigated the relationship between groom marital parity and spousal age difference by introducing groom age as a variable. Marital parity and spousal age difference positively correlated (r [8764] = +.249, p < .001, r² = .062) for grooms under 55 years of age, but were not related for grooms 55 or older (r [548] = +.062, p = .150, r² = .003). A one-way ANOVA for marital parity and spousal age difference in younger grooms revealed F (2,8763) = 352.019, p < .001, h ² = .074. Post hoc independent sample t-tests administered for each pair of means resulted in significant values for all comparisons: between first (N = 5429, M = 3.61, S.D. = 3.65)) and second marriages (N = 2367, M = 5.72, S.D. = 4.83), t (7794) = 19.04, p < .001; between first and third or subsequent marriages (N = 970, M = 6.69, S.D. = 5.36), t (6397) = 17.20, p < .001; and between second and third or subsequent marriages, t (3335) = 4.87, p < .001.
Table 1 and Figure 2 present mean spousal age differences for levels of marital parity and groom age. They also reveal the extent of the difference between younger and older grooms’ mean spousal age differences for first marriages (t [5448] = 2.97, p = .008), second marriages (t [2653] = 9.17,p < .001) and third or subsequent marriages (t [1209] = 9.09, p < .001), all of which are significant.
Although not predicted, we also found that bride marital parity related inversely with age homogamy (r [9313] = +.167, p < .001, r² = .028. A one-way ANOVA for the effect of bride's marital parity on spousal age difference revealed a significant relationship, yielding F (2, 9312) = 162.02, p < .001, with mean age differences of 4.13 (S.D. = 4.47)for first marriages (N = 5376), 5.86 (S.D. = 5.36) for second marriages (N = 2750) and 6.19 (S.D. = 5.55) for third or subsequent marriages (N = 1189). Bride marital parity accounted for 3.36% of variance in spousal age difference, and we found two of the three mean difference comparisons to be significant: between first and second marriages (t [8124] = 14.61, p < .001) and between first and third or subsequent marriages (t [6563] = 12.00, p < .001.
As with grooms, brides' marital parity and spousal age difference positively correlated for the below-55 age group (r [9005] = +.159, p < .001, r² = .025), but not for the 55-and-older age group (r [306] = +.007, p = .905, r² = .00005). In addition, ANOVA results indicated a significant relationship between marital parity and spousal age difference for the younger brides (F [2,9004] = 148.81, p < .001, h ² = .032). Post hoc independent sample t-test comparisons revealed significant differences between mean age differences for first (N = 5355, M = 4.10, S.D. = 4.41) and second marriages (N = 2599, M = 5.78, S.D. = 5.22), t [7952] = 14.19, p < .001, and for first and third or subsequent marriages (N = 1053, M = 6.03, S.D. = 5.28), t [6406] = 11.15, p < .001; but not for second and third or subsequent marriages.
Table 1 and Figure 3 display these mean spousal age differences for levels of marital parity and bride age. Also, they show that the difference between younger and older brides’ mean spousal age differences was significant for first marriages (t [5374] = 3.13, p = .005), second marriages (t[2748] = 2.45, p = .015) as well as third or subsequent marriages (t [1187] = 2.13, p =.034).
Previous Marital Status and Age Homogamy
Finally, we predicted that previous marital status for spouses in their first or second marriage (i.e., single, divorced from first marriage and widowed from first marriage) would not relate with age homogamy. We found that, especially for younger spouses (under 55 years of age), previous marital status did affect spousal age difference.
An ANOVA between previous marital status and spousal age difference for our full sample of grooms in first or second marriages yielded F (2,8095) = 349.23, p < .001. The proportion of variance in spousal age difference accounted for by groom’s previous marital status was .0794 (h ²). Post hoc independent sample t-tests demonstrated that all three pairs of marital status mean comparisons were significant. Previously single grooms (N = 5452) evidenced a mean spousal age difference of 3.64 years (S.D. = 3.78), which was significantly less than the divorced grooms’ (N = 2475) mean spousal age difference of 6.00 years (S.D. = 5.25): t (7925) = 20.10, p < .001. Also, mean spousal age difference for previously single grooms was significantly different from that for widowed grooms (N = 171, M = 9.26, S.D. = 7.58), t (5621) = 9.66, p < .001, as was the case between divorced and widowed grooms, t (2644) = 5.54, p < .001.
For grooms younger than 55, the same pattern of significant effects was revealed as for the full sample of grooms (F [2,7786] = 236.38, p < .001, h ² = .06), but there was no significant marital status effect on spousal age difference discovered for grooms 55 years and older. In post hoc analyses, we found that the younger previously single grooms (N = 5431, M = 3.61, S.D. = 3.66) were less different in age from their brides than were the younger divorced grooms (N =2314, M = 5.66, S.D. = 4.78), t (7743) = 18.51, p < .001. These previously single grooms also were less different in age from their brides than were the younger widowed grooms (N = 44, M = 8.71, S.D. = 6.76), t (5473) = 5.00, p < .001. Finally, we found that the divorced and widowed younger grooms were significantly different in spousal age difference, t (2356) = 2.97, p = .005.
Table 2 and Figure 4 show the mean spousal age differences for levels of previous marital status and groom age. They also demonstrate that the difference between younger and older grooms’ mean spousal age differences was significant in the case of previously single grooms (t [5454] = 2.97,p = .008) and divorced grooms (t [2473] = 7.74, p < .001), but not for widowed grooms.
For our full sample of brides in first or second marriages, an ANOVA between previous marital status and spousal age difference also revealed a significant relationship, F (2,8113) = 126.26, p < .001, h ² = .03. Independent sample t-test post hoc analyses again evidenced significant differences between all paired mean comparisons. Previously single brides (N = 5373, M = 4.12, S.D. = 4.47) were less different in age from their grooms than were both divorced brides (N = 2551, M = 5.79, S.D. = 5.28), t (7922) = 13.83, p < .001, and widowed brides (N = 192, M = 6.90, S.D. = 6.30), t(5563) = 6.06, p < .001. Divorced brides were significantly less different in age from their spouses than were widowed brides, t (2741) = 2.37, p < .001.
Our data revealed a significant relationship between marital status and spousal age difference for brides younger than 55 years of age (F [2,7941] = 115.70, p < .001, h ² = .03), but not for those 55 and older. Post hoc independent sample t-tests showed significant mean differences between younger brides who were previously single (N = 5352, M = 4.09, S.D. = 4.41) and those who were divorced (N = 2486, M = 5.76, S.D. = 5.19), t (7836) = 13.88, p < .001, as well as between the younger previously single brides and those who were widowed (N = 106, M = 6.49, S.D. = 6.01), t(5456) = 4.08, p < .001.
Table 2 and Figure 5 exhibit these mean spousal age differences for levels of previous marital status and bride age. They also illustrate that the difference between younger and older brides’ mean spousal age differences was significant only for previously single brides (t [5371] = 3.13, p = .005), not for divorced or widowed brides.
Discussion
The results presented indicate that the norms for age homogamy do not operate similarly across the life span. Males who were aged 55 and older were less likely to marry age homogamously than those who were under 55. This was true regardless of whether the marriage was the first or a subsequent one. In fact, age differences between spouses were even greater for older grooms in first marriages than older grooms in second or subsequent marriages and regardless of previous marital status. Similar findings were reported for females. However the strength of the associations were weaker for females.
These findings provide further support for those reported in previous research (Breckenridge, 1989; Bowerman, 1953; Glick & Landau, 1950; Hollingshead, 1951; Presser, 1975). A causal explanation for the decline in age homogamy among older brides and grooms is beyond the scope of this paper. However, the trend we have documented could be accounted for in several ways. First, with aging, the sex ratio becomes increasingly imbalanced making it more difficult to find an age homogamous mate. It stands to reason then that the norms pertaining to appropriate age differences between mates would be broadened resulting in an increasing number of heterogeneous matches. Second, the greater degree of age heterogeneity found in the marriages of older males compared to older females, may reflect the double standard in our culture of allowing males a greater age range from which to choose a spouse. Third, it may well be that older persons have different needs in mate selection, fulfillment of which leads to a greater proportion of age heterogamous matches than is the case among younger couples. This may be why previous marital status and marital parity were only weakly associated with age difference between bride and groom. These and other possible explanations warrant further investigation.
References
Adams, B. (1979). Mate selection in the United States: A theoretical summarization. In Contemporary theories about the family (pp. 259-267). New York: The Free Press.
Atchley, R. (2000). Social forces and aging: An introduction to gerontology. (9th ed.). Belmont: Wadsworth Thomson Learning.
Becker, G. (1981). A treatise on the family. Cambridge, MA: Harvard University Press.
Bem, S. (1993). The lenses of gender: Transforming the debate on sexual inequality. New Haven: Yale University Press.
Bowerman, C. (1953). Assortive mating by previous marital status: Seattle, 1939-1946. American Sociological Review, 18, 170-177.
Breckenridge, D. (1989). Trends in age heterogamy among native-born Americans, 1800-1950. Unpublished doctoral dissertation, Brigham Young University.
Burch, T. (1990). Remarriage of older Canadians: Description and interpretation. Research on Aging 12, 546-549.
Burgess ,E., & Wallin, P. (1943). Homogamy in social characteristics. American Journal of Sociology 49, 117-124.
Cowan, G. (1984). The double standard in age-discrepant relationships. Sex Roles 11, 17-23.
Dressel, P. ( 1980) Assortive mating in later life: Some initial considerations. Journal of Family Issues, 1, 379-396.
Elder, G., & Rockwell, R. (1976). Marital timing in women’s life patterns. Journal of Family History 1, 34-53.
Glick, P., & Landau, E. (1950). Age as a factor in marriage. American Sociological Review 15, 517-529.
Hollingshead, A. (1951). Age relationships and marriage. American Sociological Review, 16, 492-499.
Presser, H. (1975). Age differences between spouses. American Behavioral Scientist, 19, 190-205.
Spanier, G., & Glick, P. (1980). Mate selection differentials between whites and blacks in the United States. Social Forces, 58, 707-725.
Vera, H., Berardo, F., & Vandiver, J. (1991). Age irrelevancy in society: The test of mate selection. Journal of Aging Studies 14, 81-96.
Wolfe, N. (1990). The beauty myth: How images of beauty are used against women. London: Chatto and Windus.
Table 1
Mean Spousal Age Differences for Levels of Marital Parity
+ = p<.001, ^ = p<.01, * = p<.05, ** = p>.05,NS
Table 2
Mean Spousal Age Differences for Levels of Previous Marital Status
+ = p<.001, ^ = p<.01, * = p<.05, ** = p>.05,NS
Figure 2
Figure 3
Figure 4
Figure 5