Kenrick Keefe:Age preferences in mates resources as social status that females value in a mate (Buss 1989;Kenrick et al.1990;Sadalla et al.1987).He Oldes1 prelerred loses physical resources in the form of health and virility, ●Youngest preferred however,and his economic position may also begin to 20- drop as he ages (e.g.,Jencks 1979).Older males decrease their rate of copulation,and a lowered rate ofcopulation is in turn related to lessened probability of insemination (James 1974).As we noted above,however,testosterone and sperm counts in older males remain adequate for fertilization. Leonard(1989)discusses a factor that would incline females against choosing males who are much older than they are: Females...should show a preference for males who have achieved full maturity and high rank,which will 3040 0 60 typically mean older males,but perhaps males still MALE'SAGE FEMALE'S AGE young enough to retain effectiveness as providers until Figure 1. Age differences preferred in mate advertisements, the females'last reproductive efforts have reached plotted as minimum and maximum differences from advertiser's adulthood.If one assumes (conservatively)that a age female marries at age 20 and can expect to bear her last child no later than age 40,and that that child will require some degree of parental care until age 20,then maximum desired age for a partner.When these data were broken down by the advertiser's gender and decade if pair bonds are long-lasting,she should prefer a male of age,they yielded the pattern shown in Figure 1. who is currently a good provider and likely to remain so for 40 years.This would mean a male older than she is, As can be seen in the figure,female specifications but not much more than 10 years older (p.26). remained fairly constant throughout the age range. Leonard's reasoning could be extended to older Females were interested in partners who,on average, females choosing a mate.Compared with a 20-year-old,a ranged from 0 to 5 years younger,and from 8 to 11 years 30-year-old female would have 10 fewer years of expected older than themselves.Males,on the other hand, remaining parental care,and could "afford"to marry a changed their preferences in a systematic fashion as their slightly older man,with fewer remaining productive own age increased.Males in their 20s were equally years,but more present resources. attracted to women above and below their own age, In summary,males are hypothesized to begin with a specifying partners ranging,on average,from 5 years preference for females around their own age,but,as they younger to 5 years older.As males got older,however, grow older,to develop a preference for women who, their preferences increasingly diverged from those of although not absolutely younger,are progressively females in the same age group.Among males in their 50s younger than themselves.We hypothesize that females and 60s,the maximum acceptable age was several years will begin with a preference for older males,and com- below their own age,and the minimum specified was pared to males,show less variation in that preference almost a generation discrepant from their own age.Analy- over their lifespan.This gender and age-differentiated ses of variance indicated a significant sex x decade interac- prediction was tested in six studies. tion for both minimum (F,4,190 =3.81,p<.01),and maximum age differences (F,4,190 =2.97,p <.02) Analyzed separately by sex,male advertisers showed 3.Study 1:Age preferences in mate significant decade changes for both minimum (F,4,126= 18.56,p<.001)and maximum preferences (F,4,126= advertisements 13.96,p <.001).There was a significant,but smaller, Several social psychologists have looked to classified decade difference for female advertisers'minimum pref- newspaper advertisements as an unobtrusive measure of erences (F,4,63 =4.34,p <.01).The variations in economic exchange in the mating process(e.g.,Harrison females'maximum preferences were not significant(F,4, Saeed 1977)."Lonely hearts"advertisements allow for 63=1.05,p>.39). the nonreactive naturalistic examination of mating prefer- When age preferences from mating advertisements are ences,and social psychologists have regarded them as examined more carefully,they yield results that were not particularly well suited for examining what are usually predicted by social economic models.If there is a"norm" tacit economic exchange considerations.Because earlier for men to prefer younger women it should show up most researchers considered only the general "norm"that markedly in younger men,who tend to be more con- females mate with older males and found support for that cerned with behaving in a stereotypically sex-typed man- expectation,they did not further partition the preference ner.A number of studies have found that men become data.A test of the present hypothesis calls not only for less concerned with gender-specific norms as they age separating male and female advertisers,but also for (e.g.,Deutsch et al.1986).The interesting feature of dividing them by the advertiser's age. younger men's preferences,however,is that they ex- We analyzed a sample of 218 personal advertisements tended equally above and below their own ages.Thus, from three issues of a "singles"newspaper(the Arizona these data provide no evidence that a preference for Solo).We used only advertisements that specified the age younger women is a consistent feature of the normatively of the advertiser,and that indicated a minimum and/or defined role for males. The results do fit well with an evolutionary model.The 80 BEHAVIORAL AND BRAIN SCIENCES(1992)15:1
Kenrick Keefe:Age preferences in mates data from Study I are inherently limited,however.For MARRIAGES PHOENIX 1986 one thing,preferences expressed in singles advertise- ments may not reflect preferences in the larger popula- 20 tion.For another,such preferences may not transfer into actual reproductive behaviors because they may be lim- ited by a number of constraints.Although older males might be interested in females 15 or more years younger than themselves,for example,they must still deal with female preferences.The choice of a marriage partner is, after all,mutual,rather than unilateral;people may not get what they want.In addition,other factors,including physical attractiveness,personality,and preferences for similarity on other dimensions could all act to weaken the importance of initial age preferences in actual mate selection. 10120139440s50601 04030404 MALE'S AGE FEMALES AGE Figure 3. Actual age differences in marriages in Phoenix,AZ 4.Study 2:Marriage age statistics (aggregated for January and May 1986),plotted according to husband's and wife's ages. Because of the inherent limitations on advertised prefer- ences,we examined age differences in actual mate choice,as reflected in marriage statistics.Our first sample Once again,the same sex difference obtains.As pre- included all the marriages (n =753)for the month of dicted by our evolutionary model,the age of a man's wife January 1986 in a large city in the western United States significantly decreased in relation to his own age as he got (Seattle,WA).We first analyzed the data using the male's older (F,5,435 30.81,p <.001).An analysis using age as the predictor,and his partner's age as the depen- female's age as predictor again produced a statistically dent variable.We next analyzed the same data using significant,but much smaller F value (F,5,435=2.25,p female's age as predictor.The results are depicted in <.05).As examination of the figure indicates,women Figure 2. usually married men who were just a little older than they Those results are consistent with the evolutionary were.As men grew older,they tended to marry women model,and they fall cleanly within the preference ranges who were younger and younger relative to their own age. found in Study 1.Using age of male as predictor variable, In summary,data from two distinct geographical regions the change over the decades yields an F(5,752)=41.90,p of the United States suggest that the age preferences <.001.A comparable analysis using female age also expressed in singles advertisements reflect general popu- yields a significant F,but it is much smaller in magnitude lation preferences that translate into reproductive behav- (3.42). ior.Marriage patterns from the two cities fall squarely To guard against the possibility that spurious factors within the preference limits specified in Study 1. might have influenced the data for the month of January 1986 in Seattle,WA,we also collected data for the same 5.Study 3:Cross-generational analysis month,and from May 1986 from another large western U.S.city (Phoenix,AZ).The overall Phoenix population From an evolutionary perspective,human mate prefer- of cases was larger,and we randomly sampled 436 mar- ences reflect selection pressures that have occurred over riages from the two months.Because preliminary analysis several million years.If the sex differences described in revealed that the two months did not differ from one the above studies reflect past evolutionary constraints, another,they are collapsed for presentation in Figure 3. they should be relatively impervious to changing norms such as those that have affected sex roles in the United States during the latter half of this century.This is a MARRIAGES SEATTLE 1986 relatively conservative prediction because evolutionary theorists assume that behaviors influenced by human 20 genetic predispositions are modifiable by features of the physical and cultural environment(Lumsden Wilson 1981).In fact,natural selection can result in variable behavior patterns for several reasons.If an animal's an- cestors encountered a variable environment,that could select for a facultative strategy in which certain environ- ments lead to certain behavioral strategies,and other environments lead to different such strategies (see Crawford Anderson,1989,for a more extensive discus- sion).Environmental factors that might lead to facultative 20 20 variations in the magnitude of age discrepancies are 20304制6 1070430445060 discussed below.We would expect the general sex dif- MALE'S AGE FEMALES AGE ference in age preference to be a robust effect,however, Figure 2. Actual age differences in marriages for Seattle,WA, because the selection pressure on males to prefer in- January 1986,plotted according to husband's and wife's ages. creasingly younger females is likely to have been consis- BEHAVIORAL AND BRAIN SCIENCES(1992)15:1 81
Kenrick Keefe:Age preferences in mates MARRIAGES PHOENIX 1923 103044s4 0n39:你5040 MALE'S AGE FEMALE'S AGE MALE'S AGE FEMALE'S AGE Figure 4.Actual age differences in marriages in Phoenix,AZ. 1923,plotted according to husband's and wife's ages. Figure 5.Age preferences stated in German singles advertise- ments,plotted as minimum and maximum differences from target's age. tent over our evolutionary history.The persisting dif- ferences in the reproductive life histories of males and newspapers provide a somewhat more distinct source. females would serve to anchor the age discrepancies in Although Indian culture also has connections to Euro- pean culture,Indian marital advertisements indicate mate preferences against short-term variations in the cultural milieu. very different cultural criteria from those seen in Ameri- Assuming a historically unidirectional selection pres- can advertisements.Most advertisements include strict sure on males to prefer females of reproductive age,we limitations by caste and religion,and they commonly request horoscope information.For instance: would predict a similar age-discrepant pattern in mating ages during our grandparents'generation.To test this Wanted:a non-Bharadwaj smart good-looking preferably em- prediction,we analyzed a sample of 100 marriages re- ployed Kerala lyer girl below 25 for a Kerala lyer boy 29. corded in Phoenix,AZ,for 1923(the earliest year for Chemical engineer.Contact with horoscope. which well-organized records were available).The results (Times of India,Bombay,Sunday,Jan.29,1989) are depicted in Figure 4. As earlier,we selected advertisements that included These data conform reasonably well to the pattern of the advertiser's age,and that stipulated a minimum the data from the 1980s.Once again,the age of a man's and/or a maximum age for preferred partners.Unfortu- partner decreases directly with his age(F,5,99=8.07,p nately,we found no advertisements for women above age <.001).The changes in women's age preferences were 40 in our sample.As an Indian informant explained to us, not significantly different across decades(F<1). Indian women who do not marry by their 40s are not Thus,the sex differences in age choices have shown considered marriageable.Nevertheless,the gender dif- consistency across generations despite changes in sex- ferences in the pattern of the data are similar to those role norms during the intervening historical period. found in the American sample (Figure 7). These advertisements indicate that,as they age,males prefer increasingly younger females,whereas females 6.Study 4:Mating advertisements in continue to prefer males slightly older than themselves. different cultures The analysis for a gender x decade interaction yields an F An evolutionary model would also lead us to predict that these sex differences in the preferred age of a partner HOLLAND (Tibur 1989 would be consistent across cultures.As we noted earlier, facultative mechanisms could have evolved to vary the size of the sex discrepancy to fit varying ecological condi- tions.Several of these factors are discussed in a later section,but again note that the consistent evolutionary pressures for age-related changes in male preference should lead to some consistency in the pattern of sex discrepancies across cultures. We analyzed personal advertisements from Germany and Holland and found that the pattern in both countries is essentially the same as the pattern in American adver- tisements(see Figures 5 and 6).3 It could be argued that European countries are not the 23345 best test of the cross-cultural robustness of this phe- MALES AGE FEMALE'S AGE nomenon,because American society is deeply rooted in European culture.Matrimonial advertisements in Indian Figure 6. Age preferences stated in Dutch singles advertisements. 82 BEHAVIORAL AND BRAIN SCIENCES(1992)15:1
Kenrick Keefe:Age preferences in mates ●Youngest preferred 20 10 0 10 MALES AGE FEMALE'SAGE Figure 8.Actual age differences in 1,511 marriages on the 1 island of Poro(1913-1939),plotted according to husband's and 20 2s30s4050 2043040n503 wife's ages,respectively. MALE'S AGE FEMALE'S AGE Figure 7.Age preferences stated in Indian matrimonial advertisements. We also separately analyzed the 278 marriages re- corded between 1913 and 1920,to provide the most (4,106)=8.03,p<.001 for the minimum age prefer- historically distant test.The pattern was exactly the same ences,and an F(4,176)=2.51,p<.08 for the maximum for that earliest sample.As men aged,they showed a age preferences.Analyzed separately by gender,males' consistent increase in the tendency to marry relatively preferences changed significantly across decades for both younger women(F,5,278 =35.91,p<.001).Women minimum (F,3,57=28.89,p<.001)and maximum age married slightly older men at all ages(3.37 years,on preferences(F,3,96=12.65,p <.001).Females'age average),and there was no significant effect of the wom- preferences did not differ significantly across decades for an's decade ofage (F,5,278=1.47,p>.20).These data either minimum (F,2,46 =2.36,p>.10)or maximum argue strongly against the sociological explanation of age age preferences (F,2,83 =0.24,p >.78). discrepancies that attributes them to norms based in modern“cultural images”of“an advertising culture' (Neal,cited in Winegar 1989). 7.Study 5:Marriage ages on a small Philippine island,.1913-1939 8.Gender differences in economic resources As an additional test of the cross-cultural and cross- historical robustness of the age discrepancies in mate As we noted earlier,social psychological models some- choice,we examined marriage ages on the island of Poro times attribute the gender difference in emphasis on between the years 1913 (the earliest year for which youth and attractiveness to culturally bound differences records were available)and 1939.Poro is a small isolated in economic resources.As Margolin and White (1987) island in the Visayas region(approximately 500 kilo- note: meters southeast of Manila),with its own distinct dialect. One social norm clearly related to this asymmetry is the The natives of Poro had little contact with Europeans or long-standing tradition for beautiful women to trade Americans during the years studied.The economy was their looks for economic status in cross-gender rela- based on fishing and agriculture.Because ofthe relatively tionships....For centuries women have relied on the small size of the community,we were able to examine the commodity value of physical appearance to achieve total sample of marriages during these years(n =1511).4 upward mobility because,in comparison to men,they As in Studies 2 and 3,we analyzed the data using male age have had little else to bargain with(p.22). as the independent variable and the discrepancy between As we noted earlier,Rosenblatt(1974)speculated that his age and his wife's age as the dependent variable.We such gender differences would be less pronounced in then analyzed the data using the wife's age as the inde- societies in which women had more wealth and power.It pendent variable.Results are depicted in Figure 8. is also consistent with an evolutionary perspective that When analyzed using male age as the dependent vari- wealth might influence age preferences in a mate.As we able,the pattern was consistent with that found in the noted above,wealth can affect an individual's ability to studies discussed above-a consistent increase in age make indirect contributions to offspring survival.As dis- discrepancy as the male aged(F,5,1510 124.76,p< cussed below,it might also have an indirect influence on .001).Once again,the youngest males married women the aging process itself.Nevertheless,other research has close to their own age(0.69 years discrepancy for teens), indicated that traditional sex differences in mate prefer- whereas the oldest males married women much younger ence persist even with increases in social status and than them(20.75 years discrepancy for men in their 60s). wealth (Townsend 1989).That is,high status men con- Women's preferences also showed changes over the life- tinue to weight physical attractiveness more highly than span (F,5,1510=13.94,p<.001),but women at all ages do high status women,and high status women continue to married men slightly older,and the changes across de- weight status and wealth relatively more highly than do cades did not manifest the consistent downward trend high status men.Thus,high status American women and shown in the male data. men continue to demonstrate the same differences that BEHAVIORAL AND BRAIN SCIENCES (1992)15:1 83
Kenrick Keefe:Age preferences in mates Buss(1989)found across numerous cultures.We would therefore not expect social status to erase the pattern of WASHINGTONIAN,1989 sex differences found in the other data sets reported above. Actually,some segments of modern American society may provide a reasonable test of these speculations. Compared to women in hunter-gatherer and agrarian societies,contemporary urban professional women are much more equivalent to men in wealth and status. Although this increase in power has not occurred for all women in our society,it is possible to examine women whose personal socioeconomic status exceeds not only the status of most males in other cultures and historical epochs,but also that of the average American man Certain newspapers and magazines,like the New York 450 Review of Books and Washingtonian regularly include MALES AGE FEMALES AGE personal advertisements from women of higher socioeco- Figure 9.Age preferences stated in Washingtonian magazine. nomic status.Consider the following advertisement: with separate plotting for those specifying high SES. Wealthy attractive DWF 49,brown-eyed brunette,5'4"120 Ibs.,wants tall handsome blue-eyed monogamous unmarried ing no differences between the social class groups for the gentleman,no dependents,50-65 for long term relationship. highest age preferred.5 (Washingtonian,June 1989) In summary,the interaction between gender and age We analyzed advertisements in Washingtonian maga- persists even among those at the highest levels of income. zine that indicated a minimum and/or a maximum age Wealth does have a weak effect on preferences for youth. preferred in a partner and that specified the age of the That effect is independent of gender,so that the differen- advertiser.In addition,we coded the ads in terms of tial gender pattern remains among wealthy individuals information related to the advertiser's wealth and social Unlike older men,wealthy older women are still in- status(hereafter SES).The analysis divided advertisers terested in men their own age and older,but as they grow into those who specified that:(1)they were wealthy or had older,wealthy people of both sexes show a relatively a professional occupation(physician,lawyer,etc.),or(2) lower "floor"in the ages in which they are interested. no information or specific mention that they were not wealthy or in a relatively low-paying occupation (e.g., 9.Discussion clerical or secretary).Relatively more males than females listed themselves as wealthy or professional.Out of 182 men,129 reported themselves as high SES,versus 84 out The data we have presented indicate that age preferences are more complex than earlier social psychological mod- of 184 women.Results including this variable are de- els led us to expect.Earlier studies suggested a simple picted in Figure 9. relationship:Males seek younger females and females As indicated in this figure,the basic sex difference seek older males.Our results are consistent with half of found in the earlier studies also obtained in this study. that generalization;females tend to seek males who are Specifically,there was the predicted gender x decade interaction for both maximum (F,3,365 18.42,p slightly older than they are.For males,however,the preference for younger females is weak or nonexistent .001)and minimum age differences (F.4,369 12.07,p during early years,but becomes increasingly pronounced <.001).There was also a two-way interaction of wealth with age and decade(F,3,365=3.083,p<.05).This interaction Previous authors,although failing to note the develop- indicated that the two SES groups showed an increasing mental variations in the gender discrepancy in age prefer- discrepancy as they aged.This is consistent with the idea ence,observed that the basic age discrepancy seemed that wealthier people indicate more interest in relative somewhat irrational economically.As Cameron et al. youth as they grow older.A three-way interaction would (1977)note,"since the average age of the women adver- have indicated that the sex x decade interaction was tisers was over 9 years older than that desired by male different for high SES versus low/no info subjects.There advertisers,a serious problem of matching is evident." was no three-way interaction for either maximum or Bolig et al.(1984)likewise note,"The men who placed minimum age differences preferred (both Fs<1.30,ps> profiles in this magazine were not looking for the women 25).For exploratory purposes,we nevertheless exam- who placed profiles (nor were the women looking for ined the data separately by gender and social class. these men)."The preferences ofolder females indicate an Neither women nor men showed a significant interaction interest in males who are themselves interested in much of wealth and decade for either the high or low preferred younger women.Given the general tendency for males to difference.Consistent with the two-way interaction re- prefer younger females,however,the older females may ported above,however,there was a similar marginal tendency toward a decade by wealth interaction for both be following a somewhat"rational"strategy.On the other hand,if the goal of these advertisements is simply to men(F,4,170=2.10,p<,09)and women(F,3,191= maximize the probability of acquiring a partner,males 2.39,p<.08)on the lowest age preferred.The analogous F's for the high differences were both less than 1,suggest- would fare better by expressing an interest in females in their own age range,as well as in younger females.Given BEHAVIORAL AND BRAIN SCIENCES (1992)15:1