TRHG_14.2_finaltext.x_TRHG_14.2

150

Petersen, I., Martinussen, T., McGue, M., Bingley, P. & Christensen, K. (2011). Lower marriage and divorce rates among twins than among

singletons in danish birth cohorts 1940–1964.

Twin Research and Human Genetics

, 2, 150–157. DOI 10.1375/twin.14.2.150

In 1976 Zazzo compared the propensity to marry in a

group of French twins and singletons. The study was based

on official registries of marriage in the municipality of

Paris. He found that the proportion of married twins

among 283 pairs from birth cohorts 1883–1902 was sub-

stantially lower than the proportion among 500 singletons.

He did not include any statistical tests to support his

claims but, nonetheless, stated that (Zazzo, 1976):

… the figures seem to confirm the hypothesis that I held

when observing so often in adult twins their hostility or

at least their indifference to marriage, and despair of

some of them when their partner decided to get married.

Initial findings from the Finnish Twin Registry demon-

strated that at all ages fewer twins than singletons were

married in 1976 and that the differences were more pro-

nounced for females than males (Kaprio et al., 1979). In a

small study conducted by Pearlman of 60 twins and 30

RECEIVED 18 October, 2010; ACCEPTED 18 November, 2010.

ADDRESS FOR CORRESPONDENCE: Inge Petersen, Institute of Public

Health, Epidemiology, University of Southern Denmark, Odense,

J. B. Winslows Vej 9B, st.tv., 5000 Odense C, Denmark. E-mail:

[email protected]

singletons from US, the author did not find any evidence

of a lower frequency of marriage among twins compared

with singletons (Pearlman, 1990). Neither did the author

find any statistically significant differences between twins

and singletons with respect to number of years being

married, whether having previously been married,

number of previous marriages, or number of years since

divorce. Johnson et al. analyzed social closeness in a

sample of twins, their spouses, siblings, offspring, and

parents recruited from the Minnesota Twin Register

(7,169 twins and 5,802 relatives). They found higher social

Lower Marriage and Divorce Rates Among

Twins Than Among Singletons in Danish

Birth Cohorts 1940–1964

Inge Petersen,

Torben Martinussen,

Matthew McGue,

1,3

Paul Bingley

and Kaare Christensen

1,5

The Danish Twin Registry, Epidemiology, University of Southern Denmark, Denmark

Department of Basic Science and Environment, University of Copenhagen, Denmark

Department of Psychology, University of Minnesota, United States of America

The Danish National Centre for Social Research, Denmark

Department of Clinical Biochemistry and Pharmacology and Department of Clinical Genetics, Odense University Hospital,

Odense, Denmark

Few studies have examined differences of civil status of twins and singletons and the conclusions are contradic-

tory. In the present study, based on a linkage between the Danish Twin Register, a random 5% sample of the

total Danish population, and administrative register databases, the authors compare rates of marriage and

divorce in a sample of 35,975 twins and 81,803 singletons born 1940–1964. Cox-regressions are used in order

to control for potential confounders. We find that compared with singletons twins have significantly lower mar-

riage rates: (males: 15–19 years: Hazard Ratio (HR) = 0.66 (95%CI: 0.58–0.76); 20–24 years: 0.85 (0.82–0.88); 25

years or more: 0.96 (0.93–0.98) and females: 15–19 years: 0.70 (0.67–0.75); 20–24 years: 0.83 (0.80–0.85); 25

years or more: 0.94 (0.91–0.97)). There is no difference in divorce rates for males, but a significantly lower

divorce rate for female twins compared with singletons (HR=0.87, 95%CI: 0.83–0.90). These differences offset

each other, thus 57% of both populations remain in their first marriage until censoring. The interpretation may

be that since twins have a partner from birth, they do not have the same need for marriage as singletons but

have more experience in maintaining a relationship if they do marry.

■

Keywords: twins, singletons, marriage rate, divorce rate, Denmark, register study

ARTICLE AVAILABLE ONLINE

Twin Research and Human Genetics

Volume 14

Number 2

pp. 150–157

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

closeness in twins compared with singletons, but the

authors stated that this might not necessarily be an obsta-

cle when it comes to the quality of the twins’ social bonds

to others (Johnson et al., 2002):

It is interesting to speculate, nonetheless, whether the

twins’ greater Social Closeness extends to others around

them or is primarily limited to the twin pair bond.

The effect of social closeness within twin pairs on their

relationships to other peers was investigated in a study of

107 Canadian twin pairs where the authors concluded that

(Foy et al., 2001):

… this study does not support the conclusion that

having an intimate twin relationship interferes with the

formation of close relationships with others during late

adolescence or throughout adulthood.

Middeldorp et al. examined the frequency of divorce and

break-up of a relationship in a large sample of adolescent

and young adult Dutch twins and their singleton siblings

(Middeldorp et al., 2005). The authors found that in one

female group (mean age: 27) monozygotic twins were less

likely to be involved in a romantic relationship, dizygotic

twins were intermediate, and singletons were most likely

to be involved in a relationship. Otherwise no significant

differences between the groups were demonstrated.

Contrary to this finding Lykken et al. used a large sample

of twins from the Minnesota Twin Register to compare

the marriage and divorce rates of 4,855 twins from birth

cohorts 1936–55 at age 34–54 (Lykken et al., 1990). They

did not find any significant differences between monozy-

gotic and dizygotic twins.

Risk factors of divorce were examined in a study of

54,178 Norwegian first marriages. This study demon-

strated that couples with an age-difference of more than 4

years, couples married at younger ages (less than 25 years),

without children, and low educational level had a higher

divorce risk (Lyngstad, 2004).

In the present register study we compare marriage and

divorce rates of 35,975 twins with that of 91,803 single-

tons from Danish birth cohorts 1940–1964 controlling for

several potential confounders: birth cohort, age at mar-

riage, age difference within couples, presence of children

within marriage, and educational level.

Material and Methods

Data

The data material was obtained through a linkage of the

Danish Twin Register with administrative databases held at

Statistics Denmark. The Danish Twin Register comprises

more than 75,000 twin pairs born in Denmark since 1870

(Skytthe et al., 2006). Zygosity of same-sex pairs has been

classified by means of a mail-back questionnaire consisting

of four standard questions on physical similarity, a method

with less than 5% misclassification (Christiansen et al.,

2003). Furthermore, the study included a 5% random

sample of the total Danish population from each birth year

from 1870–2002 either alive at the start of the Danish Civil

Person Register (CPR) April 2, 1968 or born thereafter.

Linkage between registries is feasible due to a unique 10-

digit personal identification number given to all residents

of Denmark alive at the establishment of the CPR-register,

at immigration, or birth if born after this date (Pedersen et

al., 2006). Birth cohorts earlier than 1940 were excluded in

order to minimize the risk of not observing first marriage

while birth cohorts later than 1964 were excluded due to

short follow-up time and therefore a low prevalence of

divorce. Date of marriage among immigrants is not regis-

tered in Danish records unless the marriage took place in

Denmark; therefore we restricted the sample to individuals

born in Denmark.

Confounders

The CPR-register holds event-based information on

marital status (single, married, divorced or widowed), and

linkage to spouses as well as children. At the start of the

CPR-register information on the present marital status,

including date of the last registered change in marital

status, was retrieved from local registries in April 1968.

Age at first marriage as well as the age difference within

the married couple for the sample was calculated from

these data. Furthermore, we obtained information on

whether the married couple had any children in common.

The linkage between parents and children is not complete

for children born in the period from 1960–1968 (approxi-

mately 95% of all children have a link to their father

whereas approximately 99% have a link to their mother)

but approximately 100% for all children born after April 2,

1968 have a link. Hence, we expect that more than 95% of

the children of our study sample are identified.

From the Integrated Database for Labor Market

Research (IDA; Emerek et al., 1990) we have information

on the highest educational qualification obtained. The

IDA database begins in 1980, we therefore used the regis-

tered education from 1980 for birth cohorts 1940–49 and

for the subsequent birth cohorts we used the educational

code for the year of their 30th birthday. From the

International Standard Classification of Education code

(ISCED) (Unesco, 2006) we created a three level educa-

tional code (< 10 years (pre-primary, primary and

secondary), 10–12 years (upper secondary, post-secondary

non-tertiary education), and 13+ years (tertiary)) for the

sample along with their spouses and assigned an ISCED

code for the couple consisting of the maximum educa-

tional level within the pair. We added a fourth category

consisting of those with unknown educational level (edu-

cational level was unknown for 1,803 twins (5.1%) and

4,966 singletons (5.6%)).

The sample comprised 35,975 twins and 91,803 single-

tons from birth cohorts 1940–64 of which 28,495 (79.2%)

twins and 75,276 (82.0%) singletons were ever married.

Marriage and Divorce Rates in Danish Twins and Singletons

151

TWIN RESEARCH AND HUMAN GENETICS APRIL 2011

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

Inge Petersen, Torben Martinussen, Matthew McGue, Paul Bingley and Kaare Christensen

152

APRIL 2011 TWIN RESEARCH AND HUMAN GENETICS

We had information on date of marriage on 27,675 twins

(97%) and 89,253 singletons (97%), information was

missing either because they were divorced/widowed prior

to April 2 1968 or due to errors in the registration.

Statistical Methods

The frequency of marriage, mean and standard deviation

of age at first marriage as well as the frequency of divorces

among married individuals was calculated separately for

each sex and twin/singleton status as well as by zygosity.

Kaplan-Meier plots for marriage and divorce separately

were generated and Cox regression was used to estimate

the hazard rates (HR) comparing divorce and marriage

rates of twins and singletons. Entry of study was age 15 for

marriage rates and date of marriage for divorce rates,

while exit of study was event (first marriage and divorce,

respectively) or censoring due to death, widowhood or

end of study (January 1, 2003) whatever came first.

Regressions were performed separately for each sex and

controlled for potential confounders. Estimation of HR

for marriage was controlled for educational level (in 1980

or at age 30 for birth cohorts 1950–64) and 5-year birth

cohort; while for divorce rates we controlled for quintiles

of age at first marriage, age difference within the married

couple (four groups: 0–4 years, 4–9 years, 10–19 years, and

20+ years), birth cohort (5-year intervals), educational

level (couples ISCED code), and whether the couple had

any children in common. Educational level, birth cohort

and quintiles of age at marriage and difference of age

within couple were modeled using dummy variables. To

avoid violations of the proportional hazard assumption

we modeled time-varying covariate effects by splitting the

data at specified time points. We used Aalens additive

models (Hosmer & Royston, 2002) to choose the time

points for which the effects of covariates were allowed to

change. Thus, covariate effects were allowed to change at

time points 5 and 10 years for the Cox analyses of mar-

riage rates while for divorce rates we allowed changes at 1

year and 10 years. We clustered the observations on twin

pairs (using Stata’s cluster option on twin pair id) to take

account for potential correlation within pairs. Individuals

with missing information on date of marriage were

excluded from the regression analyses.

We repeated the analyses including a dummy variable

for zygosity in order to investigate the effect of self-

reported zygosity. Unknown zygosity twins were further

divided into 2 subgroups: twins whose zygosity was unre-

solved due to contradicting responses to the questionnaire

and a second group where zygosity was unknown due to

non-response. Furthermore we have performed analyses

comparing same-sex twins with opposite-sex twins.

Stata 10.1 was used for all analyses.

Results

Table 1 gives the overall frequencies of ever marriage as

well as divorce from first marriage, along with summary

of age at first marriage and age of survivors on January

1, 2003 stratified by twin/singleton status and sex. The

results show that age at first marriage for both males and

females is higher for twins compared with singletons

(overall age difference is 0.8 years (p < .001) for males and

1.3 years (p < .001) for females) and fewer twins marry

compared with singletons; the overall marriage frequency

for males was 78.1% for singletons and 76.0% for twins,

the numbers for females were 86.1% respectively 82.9%.

The results for divorce show that fewer twins than single-

tons are divorced (29.6% vs. 30.4% (p = .09) for males and

27.3% vs. 31.3% (p < .001) for females).

The Kaplan-Meier plots for marriage are pictured in

Figure 1. The plots demonstrate a higher marriage rate

among singletons compared with twins for males as well

as females. The differences occur at the beginning of the

observation period and persist at a virtually constant level

throughout the observation period. Figure 2 depicts the

Kaplan-Meier plots for divorce. These plots demonstrate

TABLE 1

Mean Age at First Marriage and Overall Frequencies of Ever Marriage and Divorce on January 1, 2003 Among 35,975 Twins and 91,803

Singletons from Danish Birth Cohorts 1940–1964

Males Females

Twins Singletons Twins Singletons

Total number 19,291 47,067 16,684 44,736

Alive on January 1, 2003,

(%)

18,366 (95.2) 44,850 (95.3) 16,258 (97.4) 43,663 (97.6)

Age on January 1, 2003, mean (

)

50.7 (7.2) 50.4 (7.1) 50.3 (7.4) 50.5 (7.2)

Married,

(%)

14,662 (76.0) 36,742 (78.1) 13,833 (82.9) 38,534 (86.1)

Divorced,

(%)

4,346 (29.6) 11,168 (30.4) 3,780 (27.3) 12,076 (31.3)

Divorced or widowed before April 2 1968, n (%)

132 (0.7) 380 (0.8) 173 (1.0) 567 (1.3)

Error in registration,

(%)

251 (1.3) 689 (1.5) 264 (1.6) 914 (2.0)

Study base,

(%)

18,908 (98.0) 45,998 (97.7) 16,247 (97.4) 43,255 (96.7)

Age at 1st marriage, mean (

)

31.9 (10.3) 31.1 (10.0) 28.6 (10.0) 27.3 (9.5)

Note:

, standard deviation;

Percentage of total number;

Mean age of individuals alive on January 1, 2003;

Percentage of married individuals;

Age at first

marriage of study base

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

Marriage and Divorce Rates in Danish Twins and Singletons

153

TWIN RESEARCH AND HUMAN GENETICS APRIL 2011

that female twins have a lower divorce rate, and the differ-

ence is increasing over time, compared with male twins,

and male as well as female singletons but only small differ-

ences between these three latter groups.

The frequencies of marriage and divorce according to

zygosity groups are reported in Table 2. The general

picture is that twins with unknown zygosity (UZ) have

lower frequency of marriage and higher frequency of

divorce compared with the twins of known zygosity; the

group of non-responders has the lowest marriage and the

highest divorce rates of all groups. The three groups of

known zygosity exhibit small differences with respect to

frequency of marriage and divorce. The trend for both

sexes is that the monozygotic (MZ) twins have the highest

frequency of marriage and the lowest frequency of

divorce, while opposite-sex (osDZ) twins have the lowest

frequency of marriage (only males) and the highest fre-

quency of divorce.

Comparison — without taking censoring into consid-

eration — of the age at first marriage and frequencies of

marriage and divorce of same-sex with opposite-sex twins

proved quite small, and statistically insignificant, differ-

ences for males (ever married: 76.0% of same-sex and

76.1% of opposite-sex twins (p = .86); mean age at first

marriage: 27.4 years for same-sex and 27.5 years for oppo-

site-sex twins (p = .85); divorce: 29.4% for same-sex and

30.1% for opposite-sex twins (p = .37)). For females we

observed a significantly (p = .04) higher frequency of mar-

riage among opposite-sex twins (83.6%) compared with

same-sex twins (82.4%), opposite-sex twins married on

average slightly younger than same-sex twins (24.7 years

and 25.1 years (p < .001)), and same-sex twins had a lower

frequency of divorces than opposite-sex twins (26.4% vs.

28.7% (p < .01)).

Regressions

Raw and adjusted hazard ratios (HR) for marriage

obtained through Cox regression using time-varying

covariate effects (15–19 years, 20–24 years, and more than

25 years) are reported for males and females (Table 3). The

FIGURE 2

Kaplan-Meier plots for divorce from first marriage. Danish twins

(

= 27,675) and a 5% random sample (

= 72,725) of each birth

cohort from 1940–1964.

0.6

0.7

0.8

0.9

1.0

0 10 20 30 40 50

Years since first marriage

Twins - males Random sample - males

Twins - females Random sample - females

FIGURE 1

Kaplan-Meier plots for first marriage. Danish twins (

= 53,155)

and a 5% random sample (

= 89,253) of each birth cohort from

1940–1964.

0.0

0.2

0.4

0.6

0.8

1.0

15 25 35 45 55 65

Age

Twins - males Random sample - males

Twins - females Random sample - females

TABLE 2

Marital Status on January 31, 2003 (or at Death) of 35,975 Twins from Danish Birth Cohorts 1940–64

Marriage Divorce

Males Females Males Females

Number

Percent Number

Percent

Number

Percent

Opposite-sex 6,905 76.1 6,690 83.6 5,253 30.1 5,595 28.7

All same-sex 12,386 76.0 9,994 82.4 9,409 29.4 8,238 26.4

Monozygotic 3,650 78.7 3,248 83.7 2,873 28.6 2,719 23.9

Same-sex dizygotic 7,185 76.8 5,791 82.7 5,520 28.7 4,788 26.2

Unknown zygosity 1,551 65.5 955 76.5 1,016 35.3 731 36.5

Unresolved

498 69.1 349 78.2 344 32.6 273 27.8

Non-response

1053 63.8 606 75.6 672 36.8 458 41.7

Note:

Total Sample;

Number of married individuals of the total sample;

Percent of married;

Twins with missing zygosity information due to conflicting

answers on zygosity questionnaire;

Twins with missing zygosity information due to non-response on zygosity questionnaire

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

Inge Petersen, Torben Martinussen, Matthew McGue, Paul Bingley and Kaare Christensen

154

APRIL 2011 TWIN RESEARCH AND HUMAN GENETICS

tables include results for the pooled twin sample as well as

for the four zygosity groups (MZ, ssDZ, osDZ, and UZ) as

well as for a further subdivision of UZ twins (conflicting

answers and non-response). Except for MZ males and

females, who on average stayed unmarried until 25 years

of age, we found lower marriage rate for twins compared

with singletons — most of them highly statistically signifi-

cant. The results reveal that the HR’s converges to 1 with

increasing observation time for all groups of twins except

for the twins of unknown zygosity where the largest HR is

observed in the first time period (corresponding to 15–19

years of age); this opposite trend is mainly due to twins

with unknown zygosity due to non-response. These results

partly reflect that fewer twins marry compared with single-

tons and partly that twins marry at older ages (especially

twins of known zygosity) compared with singletons.

The results of linear Aalen regression showed that there

was no interaction between divorce and observation time

for either sex. Thus we did not model time-varying effects

of the outcome (divorce) in these analyses (but time-

varying effects of the confounding covariates were

included in the models). The results of the regressions are

shown in Table 4. Statistically significant results for divorce

rates are obtained for females (HR: 0.91 (95%CI: 0.87–

0.94)) but not for males (HR: 0.99 (95%CI: 0.96–1.03)).

Table 4 also reports the HR’s of zygosity with singletons as

reference group; the unknown zygosity demonstrates the

highest divorce rates. While for females the higher divorce

rates among twins of unknown zygosity was solely due to

non-responders, we observed high divorce rates for both

groups of male twins of unknown zygosity.

Discussion

In the present register-based study we found a lower, time-

dependent (HR’s are converging to one with increasing

observation time) marriage rate among twins than among

singletons and those twins that do marry are on average

older at first marriage compared with singletons. We

found non-significant hazard rates (HR) for divorce

among males but we observed that female twins have sig-

nificantly lower divorce rates compared with their

singleton counterparts.

Comparison With Other Studies

In accordance with the French study (Zazzo, 1976) com-

paring the propensity of twins and singletons in Paris to

marry, we demonstrated a lower percentage of ever-

married twins compared with singletons, both for males

(76% vs. 78%) and for females (83% vs. 86%). In contrast,

Zazzo found a larger difference, 65% vs. 78% among 272

traceable twins and singletons. Zazzo’s study concerns

earlier birth cohorts (1882–1902) than the present study

(1940–1964); his results are based on a small selected

sample (566 twins of which 90 were traceable, and 500

randomly selected singletons of which 182 were traceable),

and he did not take account of sex and age in his analyses,

nor did he report any statistical tests. Furthermore, our

study is supported by a cross-sectional study of Finnish

twins by Kaprio et al. that found fewer twins of all ages

TABLE 3

Overall Raw as well as Confounder Adjusted Hazard Ratios and 95% Confidence Intervals in Three Age Periods: 15–19 Years, 20–24 Years,

25+Years for 1st Marriage Among 35,155 Twins and 89,253 Singletons from Danish Birth Cohorts 1940–64 Using Singletons as Reference Group

— Censoring at Death or End of Study (January 1, 2003)

Raw 15–19 years 20–24 years 25+ years

HR 95%CI HR

95%CI HR

95%CI

Males

Twins (pooled) 0.93

***

0.91–0.95 0.66

***

0.58–0.76 0.85

***

0.82–0.88 0.96

0.93–0.98

Monozygotic 0.97 0.93–1.01 0.66

0.48–0.89 0.87

***

0.80–0.93 1.06

1.00–1.12

Same-sex dizygotic 0.95

0.92–0.98 0.66

***

0.53–0.82 0.90

***

0.86–0.94 0.96

0.92–1.00

Opposite-sex dizygotic 0.93

***

0.90–0.95 0.63

***

0.51–0.78 0.82

***

0.78–0.86 0.97 0.93–1.00

Unknown zygosity 0.73

***

0.68–0.79 0.82 0.55–1.20 0.77

***

0.69–0.86 0.68

***

0.62–0.75

Unresolved

0.72

***

0.64–0.81 0.47 0.17–1.24 0.63

***

0.50–0.79 0.78

***

0.68–0.89

Non-response

0.74

***

0.67–0.81 0.93 0.61–1.42 0.83

0.74–0.93 0.62

***

0.55–0.71

Females

Twins (pooled) 0.87

***

0.85–0.88 0.70

***

0.67–0.75 0.83

***

0.80–0.85 0.94

***

0.91–0.97

Monozygotic 0.86

***

0.82–0.90 0.68

***

0.59–0.78 0.82

***

0.77–0.88 1.01 0.95–1.08

Same-sex dizygotic 0.86

***

0.83–0.88 0.72

***

0.66–0.79 0.82

***

0.78–0.86 0.93

0.88–0.97

Opposite-sex dizygotic 0.90

***

0.87–0.92 0.69

***

0.64–0.75 0.87

***

0.83–0.90 0.96 0.91–1.00

Unknown zygosity 0.76

***

0.69–0.83 0.79

0.65–0.97 0.66

***

0.58–0.74 0.68

***

0.60–0.79

Unresolved

0.77

***

0.67–0.88 0.82 0.59–1.15 0.72

0.58–0.88 0.82 0.66–1.02

Non-response

0.75

***

0.67–0.85 0.78

0.61–0.99 0.63

***

0.54–0.73 0.60

***

0.50–0.71

Note: CI, Confidence Intervals; HR, Hazard Ratio;

p <

.05;

p <

.01;

***

p <

.001

Hazard ratios for marriage are adjusted for education and 5-year birth cohort.

Twins with missing zygosity information due to conflicting answers on zygosity questionnaire

Twins with missing zygosity information due to non-response on zygosity questionnaire

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

Marriage and Divorce Rates in Danish Twins and Singletons

155

TWIN RESEARCH AND HUMAN GENETICS APRIL 2011

were married in 1976 compared with census data from

1970, and that the differences were smaller for males than

females (Kaprio et al., 1979). The results in the present

study contradicts the findings by Pearlman (1990) and

Middeldorp et al. (2005), both reporting insignificant

association between single-status and twin-singleton

status. The conclusion drawn by Pearlman is based on a

small sample of 60 twins and 30 singletons, and thus study

size might explain why the author did not find significant

results. Though statistically insignificant, the results of a

Dutch study by Middeldorp et al. (2005) reports the same

trend as found in the present study. Furthermore the

authors of that study find — although still statistically

insignificant — that the proportion of MZ twins having a

spouse is lower than that of DZ twins. In the present study

we find the opposite trend (Table 2). In the study per-

formed by Zazzo (1976) a higher marital frequency was

found among male same-sex twins (25%) compared with

opposite-sex twins (15%), but no difference for females.

Our study does not support this finding. For males we

found that 76% of both same-sex and opposite-sex twins

were ever married, while for females we found the oppo-

site trend, that is, 82% of same-sex twins were married vs.

84% of opposite-sex twins. Contrary to the studies by

Pearlman (1990) and Middeldorp et al. (2005) the data in

our study demonstrate a lower frequency of divorces

among twins compared with singletons. The lack of signif-

icance in previous studies might be due to small sample

sizes and follow-up intervals that were too short. We have

not been able to detect any previous studies comparing

the age at first marriage of twins with that of singletons.

Previous studies have demonstrated that twins often

are more closely attached to each other than non-twin sib-

lings (Foy et al., 2001; Johnson et al., 2002). Zazzo (1976)

observed this closeness within twin pairs and felt con-

vinced that the tight bonds were causing the low marriage

frequency among 283 twin pairs compared with the one in

500 singletons born in the municipality of Paris 1883–

1902. The present study is based on register data. Hence,

the interpretation of the results is based on pure conjec-

ture. A possible interpretation is that the close bonds

within twin pairs reduce their desire of bonding to a

partner — thus reducing the marriage rate of twins. The

time-dependent nature of the marriage rates reveals that

the difference is mainly due to a lower teenage marriage

rate. Hence, the twin-bonding may be loosened in adult-

hood, thereby giving room for evolving romantic

relationships. This interpretation is furthermore justified

by the slightly higher average age at marriage among twins

compared with singletons. The interpretation of the lower

divorce rates of female twins compared with singletons

may be that twins are used to show consideration for their

co-twin from the very beginning thereby training their

adaptability of bond making.

Mortality

Several studies have established an association between

mortality and marital status (Ben-Shlomo et al., 1993;

Ebrahim et al., 1995; Hu & Goldman, 1990; Kotler &

Wingard, 1989; Rosengren et al., 1989; Shaikh, 1990)

probably both due to selection into marriage, that is, those

with poor health and/or low socio-economic status stay

single as well as health promoting effect of marriage.

Previous studies comparing mortality of twins with that

of the total Danish population did not demonstrate any

differences (Christensen et al., 1995; 2001). Thus, it seems

that single status is not as hazardous for twins as for sin-

gletons. This may be because the lower marriage rate of

twins is counterbalanced by their lower divorce rate.

Consequently, 53.5% of male twins and 54.3% male sin-

gletons remain in their first marriage until the end of the

study, death or widowhood. The same numbers for

TABLE 4

Raw as well as Confounder Adjusted Hazard Ratios and 95% Confidence Intervals for Divorce (From 1st Marriage) of 27,675 Twins (14,279 Males

and 13,396 Females) and 72,725 Singletons (35,673 Males and 37,052 Females) from Danish Birth Cohorts 1940–64 using Singletons as

Reference Group — Censoring at Death, Widowhood, or End of Study (January 1, 2003)

Males Females

Raw Adjusted

HR 95%CI HR 95%CI HR 95%CI HR 95%CI

Twins (pooled) 0.97 0.94–1.01 0.99 0.96–1.03 0.87

***

0.83–0.90 0.91

***

0.87–0.94

Monozygotic twins 0.99 0.91–1.07 1.00 0.92–1.08 0.79

***

0.73–0.87 0.81

***

0.74–0.89

Same-sex dizygotic twins 0.93

0.88–0.98 0.94

0.89–1.00 0.84

***

0.79–0.89 0.88

***

0.82–0.94

Opposite-sex twins 0.98 0.93–1.04 1.01 0.95–1.07 0.89

***

0.84–0.94 0.94

0.89–0.99

Unknown zygosity 1.18

1.06–1.32 1.18

1.05–1.32 1.14 0.99–1.32 1.20

1.04–1.38

Unresolved

1.17 0.95–1.45 1.20 0.97–1.47 0.92 0.72–1.19 0.94 0.72–1.22

Non-response

1.19

1.04–1.36 1.17

1.03–1.34 1.27

1.08–1.51 1.36

***

1.15–1.61

Note: CI, Confidence Interval; HR, Hazard Ratio;

p <

.05;

p <

.01;

***

p <

.001

Hazard Ratios are adjusted for couple education, 5-year birth cohort, quintiles of age at marriage, age difference within couple (0–4 years, 5–9 years,

10–19 years, and 20+ years), and whether the couple had any children in common.

Twins with missing zygosity information due to conflicting answers on zygosity questionnaire

Twins with missing zygosity information due to non-response on zygosity questionnaire

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

females are 60.3% and 59.1%, respectively. Hence it is

unlikely that lower marriage rate among twins is caused by

poorer health of twins compared with that of singletons.

Unknown Zygosity

The unknown zygosity twins are a heterogeneous group

consisting of identified twins who either didn’t respond

to the questionnaire on physical similarities or whose

answers were inconsistent. In a previous register-based

study of school achievements of Danish adolescent twins

from birth cohorts 1986–1989 it was demonstrated that

the parents of unknown zygosity twins had lower educa-

tional level and were slightly younger, and the UZ twins

had significantly worse school achievements compared

with twins of known zygosity (Petersen et al., 2009). In the

present study we demonstrated that the unknown zygosity

twins had lower marriage rates and higher divorce rates

compared with twins of known zygosity, and the disad-

vantage was highest among non-responders. Hence, most

likely it is the general picture that the twins whose zygosity

is not accessible — especially those who didn’t respond to

the questionnaire — are disadvantaged compared to twins

with known zygosity.

Limitations

Our study only covers legal marriages and not couples

living together without being married. However, since

86% of all individuals from birth cohorts 1940–64 were or

had been legally married in 2002 and furthermore we have

no reason to believe that twins who live together are more

likely to be unmarried than singletons living together, it is

unlikely to introduce a bias to our study.

A second concern is the erroneous registration of date

of marriage. Approximately 2.5% of our sample was

excluded from the Cox analyses, either because they had

ended their first marriage before the start of the CPR-reg-

ister or due to registration errors. We found an uneven

distribution of these exclusions in that more females than

males were excluded and more singletons than twins were

excluded. Still, the twin-singleton difference in exclusion

rate is very small for both males (twins: 2.0%; singletons:

2.3%) and for females (twins: 2.6%; singletons: 3.3%),

thus we do not expect that this causes any substantial bias

in the results.

The Danish Twin Register holds approximately 70% of

the twins born in the period from 1940–64, the remaining

twins are mainly lost due to infant death of one or both of

the twins prior to the start of the CPR register in April

1968 (Skytthe et al., 2002). Since we only include individu-

als surviving to age 15 the loss due to incompleteness is

quite a lot lower: mainly twins surviving to age 15 without

a surviving co-twin are missing in our sample. This might

introduce a bias in our study if infant death is associated

with marital status of the surviving twin.

Due to either non-response or conflicts of the answers,

zygosity was missing for 11.2% of same-sex twins. This

group has, as demonstrated in the present study, lower

marriage as well as higher divorce rates compared with

same-sex twins of known zygosity. Hence, the present

results for MZ and same-sex DZ twins are slightly over-

estimated.

We have limited the study to first marriage. There may

have been a few individuals who were married at a young

age, and divorced and remarried before April 2, 1968.

Since the oldest in our study are turning 28 in 1968 this

group must be small and therefore we do not expect this

to distort the results of our study.

Strengths

The size of the study (35,975 twins and 91,803 singletons)

combined with the register-based results are the major

strengths of the study. No prior study has examined such

a large sample with minimal selection bias and loss to

follow up.

Conclusion

We conclude that twins are married at slightly older ages

compared with singletons, that twins have lower marriage

rates compared with singletons, and that female twins

have lower divorce rates compared with singletons, but no

difference is observed for divorce rates among males. The

results might indicate that twins do not have the same

need for marriage as singletons but female twins are better

at maintaining the marriage thereby avoiding divorce.

References

Ben-Shlomo, Y., Smith, G. D., Shipley, M., & Marmot, M. G.

(1993). Magnitude and causes of mortality differences

between married and unmarried men. Journal of

Epidemiology and Community Health, 47, 200–205.

Christensen, K., Vaupel, J. W., Holm, N. V., & Yashin, A. I.

(1995). Mortality among twins after age 6: Fetal origins

hypothesis versus twin method. BMJ, 310, 432–436.

Christensen, K., Wienke, A., Skytthe, A., Holm, N. V., Vaupel,

J. W., & Yashin, A. I. (2001). Cardiovascular mortality in

twins and the fetal origins hypothesis. Twin Research, 4,

344–349.

Christiansen, L., Frederiksen, H., Schousboe, K., Skytthe, A.,

von Wurmb-Schwark, N., Christensen, K. et al. (2003).

Age- and sex-differences in the validity of questionnaire-

based zygosity in twins. Twin Research, 6, 275–278.

Ebrahim, S., Wannamethee, G., McCallum, A., Walker, M., &

Shaper, A. G. (1995). Marital status, change in marital

status, and mortality in middle-aged British men.

American Journal of Epidemiology, 142, 834–842.

Emerek, R., Hansen, P. V., & Leth-Sorensen, S. (1990).

Documentation of the Danish Database for Labor Market

Research (Danish). Retrieved from http://dst.dk/

upload/hovedrapport.pdf

Foy, A. K., Vernon, P., & Jang, K. (2001). Examining the

dimensions of intimacy in twin and peer relationships.

Twin Research, 4, 443–452.

Inge Petersen, Torben Martinussen, Matthew McGue, Paul Bingley and Kaare Christensen

156

APRIL 2011 TWIN RESEARCH AND HUMAN GENETICS

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press

Marriage and Divorce Rates in Danish Twins and Singletons

157

TWIN RESEARCH AND HUMAN GENETICS APRIL 2011

Hosmer, D. W. & Royston, P. (2002). Using Aalen’s linear

hazards model to investigate time-varying effects in the

proportional hazards regression model. The Stata Journal,

4, 331–350.

Hu, Y. R. & Goldman, N. (1990). Mortality differentials by

marital status: An international comparison. Demography,

27, 233–250.

Johnson, W., Krueger, R. F., Bouchard, T. J., Jr., & McGue, M.

(2002). The personalities of twins: Just ordinary folks.

Twin Research, 5, 125–131.

Kaprio, J., Koskenvou, M., Artimo, M., Sarna, S., & Rantasalo,

I. (1979). The Finnish Twin Registry: Baseline characteris-

tics; Section 1: Materials, methods, representativeness and

results for variables special to twin studies. Helsinki.

Kotler, P. & Wingard, D. L. (1989). The effect of occupa-

tional, marital and parental roles on mortality: The

Alameda County Study. American Journal of Public

Health, 79, 607–612.

Lykken, D. T., Bouchard, T. J., Jr., McGue, M., & Tellegen, A.

(1990). The Minnesota Twin Family Registry: Some

initial findings. Acta Geneticae Medicae et Gemellologiae,

39, 35–70.

Lyngstad, T. H. (2004). The impact of parents’ and spouses

education on divorce rates in Norway. Demographic

Research, 10, 121–142.

Middeldorp, C. M., Cath, D. C., Vink, J. M., & Boomsma, D.

I. (2005). Twin and genetic effects on life events. Tw in

Research and Human Genetics, 8, 224–231.

Pearlman, E. M. (1990). Separation-individuation, self-

concept, and object relations in fraternal twins, identical

twins, and singletons. Journal of Psychology, 124, 619–628.

Pedersen, C. B., Gotzsche, H., Moller, J. O., & Mortensen, P.

B. (2006). The Danish Civil Registration System. A cohort

of eight million persons. Danish Medical Bulletin, 53,

441–449.

Petersen, I., Jensen, V. M., McGue, M., Bingley, P., &

Christensen, K. (2009). No evidence of genetic mediation

in the association between birthweight and academic per-

formance in 2,413 Danish adolescent twin pairs. Tw in

Research and Human Genetics, 12, 564–572.

Rosengren, A., Wedel, H., & Wilhelmsen, L. (1989). Marital

status and mortality in middle-aged Swedish men.

American Journal of Epidemiology, 129, 54–64.

Shaikh, K. (1990). Marriage and mortality: A life table analy-

sis. Journal of Biosocial Science, 22, 53–61.

Skytthe, A., Kyvik, K., Bathum, L., Holm, N., Vaupel, J. W., &

Christensen, K. (2006). The Danish Twin Registry in the

new millennium. Twin Research and Human Genetics, 9,

763–771.

Skytthe, A., Kyvik, K., Holm, N. V., Vaupel, J. W., &

Christensen, K. (2002). The Danish Twin Registry: 127

birth cohorts of twins. Twin Research, 5, 352–357.

Unesco (2006). United Nations Educational,Scientific and

Cultural Organization. Available at http://www.uis.unesco.

org/template/pdf/isced/ISCED_A.pdf

Zazzo, R. (1976). The twin condition and the couple effects

on personality development. Acta Geneticae Medicae et

Gemellologiae, 25, 343–352.

https://doi.org/10.1375/twin.14.2.150 Published online by Cambridge University Press