Keeping Contact with Children: Assessing the Father/Child Post-separation Relationship from the Male Perspective

II. DATA AND METHODS

A. Limits of the 1995 General Social Survey on the Family

Cycle 10 of the General Social Survey was carried out by Statistics Canada in 1995. For the first time in Canada, this survey collected information on the frequency of contact between fathers and children not only from separated mothers but also from fathers (though not from both parents of the same child). More than 10,000 men and women aged eighteen and over replied to questions concerning all the children they had given birth to or raised during their life. Separated parents were also asked to state how much time they had spent with each of their children during the year preceding the survey and how much contact each child had had with the other parent. The information reported by fathers enabled us to measure directly the time they spent with their children and to construct a socio-demographic profile of these fathers according to the frequency of father/child contact.

Exploiting the 1995 GSS data was not problem-free. First, an earlier study had shown that the quality of information provided by men concerning children from previous unions was inferior to that supplied by women. In theory, one would expect separated mothers and fathers participating in the survey to declare approximately the same number of biological children as mothers and fathers who were still together at the time of the survey had done (see Juby and Le Bourdais, 1999). However, our study showed this to be far from the case: of the children whose parents were living apart at the time of the survey, almost two-thirds were declared by female respondents, i.e. by their mother. In other words, male respondents not living with the mother of their children, i.e. separated fathers, reported 40 percent fewer children than separated mothers. These findings support those found in other studies (Furstenberg, 1988; Poulain et al., 1991; Rendall et al., 1997; Seltzer and Brandreth, 1994) and result from the fact that, on the one hand, surveys have more difficulty contacting separated fathers than mothers and, on the other, separated fathers are more likely than separated mothers to under-report the number of children they have had in the past. Given this difference in reporting according to the sex of the responding parent, we were unable to create a single sample of children by combining reports from fathers and mothers to analyze the frequency of father/child contact without introducing an important bias into the results.

In addition, the comparison of reports by fathers and mothers demonstrated that under-reporting particularly affected fathers with little or no contact with their children, and who paid no child support, while those with a regular contact were much more likely to declare them (Juby and Le Bourdais, 1999). As long as this limitation is kept in mind, however, it is possible to extract from the analysis of data provided by fathers some very valuable information on the contact they have with their children.

The process of data collection itself is the source of the second problem encountered in the exploitation of GSS data. Given the complexity of pathways that had to be followed in the GSS questionnaire, some questions were not put to particular groups of respondents. The questions on whether or not custody arrangements had been settled in court, for instance, were omitted for children declared to have spent the entire year prior to the survey with the other parent. This made difficult, if not impossible, some of the analyses that we had intended to carry out; we will return to these problems later on.

B. Creating the samples of fathers and children

This study concerns only biological or adopted children, declared by male or female respondents, who were aged 0 to 17 years at the time of survey and whose parents (biological or adoptive) were not living together at the time of the survey; thus, any children from an earlier union of the respondent's current partner, and children aged eighteen or over in 1995, were excluded from the sample.

An analysis of the frequency of father/child contact could adopt either the fathers' perspective, and be based on a sample of fathers, or the children's, using a sample of children. Selecting the father as the unit of analysis avoids attributing too high a weight to the characteristics of fathers with several children. However, a cursory examination of the data showed that fathers often spend different amounts of time with their various children. Custody arrangements may vary among siblings: a father might, for instance, have his adolescent son living with him full time while a younger child might only visit every other weekend. In addition, children reported by a father may not all have the same mother, in which case they are very likely to have different custody arrangements. To take these variations in the levels of father/child contact into consideration, we first constructed a sample of children.

Table 1: Number of biological or adopted children aged 0-17 years whose parents were living apart at the time of the survey, by sex of the responding parent
Description Respondent
Father Mother
Total number of children a 477 832
Exclusions:
- Other parent deceased 15 45
- Does not know if other parent alive 6 33
- Missing 9 9
- Other (child seldom with either parent) 4 13
Number of children whose parents are alive but not together 443 732
Number of parents of the children in the sample 311 449
Other exclusions:
- Children excluded because of missing data 25 56
For analyses based on the time spent with father:
Number of children 418 676
Number of parents 291 414

Source: Statistics Canada, General Social Survey (GSS), Cycle 10: the Family, 1995.
a Weighted data, brought back to the original sample size.

To do this, we proceeded in the following way (see Table 1). From the information provided by the male and female respondents, we first selected all biological or adopted children aged from 0 to 17 years whose biological or adoptive parents were not living together at the time of survey (477 children reported by fathers, 832 by mothers), and retained only those with both parents still living. Thus, among the children declared by fathers, 15 children whose mother had died were ineligible, as were six others for whom the mothers' fate was unknown. Thirteen additional children were excluded from the sample: 9 for whom no information on the mother was available, and four others with discrepancies in the information collected. Overall, 34 children declared by fathers and 100 by mothers were excluded. The remaining sample of children reported by fathers equalled 443 children declared by 311 fathers, compared with 732 children reported by 449 mothers.

In the course of the following analyses, we will at times use the child sample, at others the father sample. When the variable being examined is based on information available for each child (such as the amount and type of contact, the child's age at the time of the survey and at separation, the level of satisfaction with the custody arrangements), the child sample will be used. Analyses of fathers' characteristics at the time of the survey (age, education, income and perception of the paternal role), however, will be carried out on the father sample.

The distribution of mothers and fathers according to the number and age of their children, presented in Table 2, makes clearer the link between the two samples. Three age categories were defined to describe the children declared by the same parent: 1) the respondent has at least one preschool-aged child (0-5 years); 2) the parent has no child of preschool age, but at least one child of primary school age (6-11 years); 3) all children are of secondary school age (12-17 years).

Table 2: Distribution of respondents living apart from the other biological parent of their children at the time of the survey, by sex and age group, and by the number and age of their children
  Father Mother
Age group Age group
< 30 30-39 40-49 50+ N a % < 30 30-39 40-49 50+ N a %
Number 50 120 115 26 311 --- 113 210 115 11 449 ---
% 16.2 38.5 37.0 8.3 --- 100 25.1 46.8 25.7 2.4 --- 100
Number of children aged 0-17 years:
1 82.4 54.1 61.6 77.0 197 63.4 64.3 44.4 45.3 86.2 227 50.6
2 13.3 40.7 33.1 23.0 99 31.9 26.2 38.8 46.6 13.8 166 371
3 or more 4.3 5.2 5.3 --- 15 4.7 9.5 16.8 8.1 --- 56 12.3
Total 100 100 100 100 311 100 100 100 100 100 449 100
Age group of children:
At least 1 child aged 0-5 years 83.6 22.2 9.2 --- 79 25.5 79.5 26.9 4.1 --- 151 33.6
At least 1 child aged 6-11 years (none aged 0-5 ) 16.4 52.0 36.2 14.6 116 37.3 20.5 41.6 35.3 9.6 152 33.9
All children aged 12-17 years --- 25.8 54.6 85.4 116 37.2 --- 31.5 60.6 90.4 146 32.5
Total 100 100 100 100 311 100 100 100 100 100 449 100

Source: Statistics Canada, General Social Survey (GSS), Cycle 10: the Family, 1995.
a Weighted data, brought back to the original sample size.

A first observation is that mothers not living with the father of their children are younger than fathers in the same situation. While three-quarters of fathers and mothers are aged between 30 and 49 years, one mother in four is under thirty 30 compared with one father in six. This is undoubtedly due in part to the age difference traditionally observed between spouses. Mothers also report a greater number of children than fathers. Almost two-thirds (63.4 percent) of fathers only reported one child under the age of 18 at the time of the survey, compared with half the mothers; in contrast, less than 5 percent of fathers declared three or more children as against 12.3 percent of mothers. The age of parents is closely linked to the number and age of reported children. Take, for example, the higher percentage of fathers aged 30 to 49 years with two or more children; one can assume that fathers under 30 have scarcely begun their paternal career, while some of the children of those aged over 50 will already have reached the age of majority. Finally, while a substantial majority of fathers (75 percent) have no child under six, almost all the young fathers (under 30) had a child of this age.

C. Defining the frequency of father/child contact

Within the framework of the GSS on the Family, respondents living apart from their children's other parent were asked to specify the number of hours, days, weeks or months that they spent with each of their children in the course of the 12 months preceding the survey; they were also asked for an estimate of the amount of time each child had spent with the other parent. This information, recoded in the micro-data file produced by Statistics Canada, as a number of days permitted us to classify the children according to the frequency of contact with their father;[1] 25 children for whom this information was missing had to be excluded from the analysis, reducing our sample to 418 children (and 291 fathers; see Table 1).

Table 3 presents the distribution of children reported by their fathers, according to the number of days, grouped into eight categories, that they spent together in the 12 months preceding the survey. Three children in ten (30.4 percent[2]) lived at least five months with their father, with around one child in eight spending 10 months or more with him; at the other extreme, one child in six spent no time at all with their father in the previous year. It is important to remember that our child sample overestimates the amount of father/child contact, as children spending less than two months per year with their father are significantly under-represented (Juby and Le Bourdais, 1999). The real proportion of children with little contact with their father is likely to be substantially higher than that observed here, as is suggested by the information given by mothers. According to their reports, one child in four had no contact with his father in the twelve months preceding the survey, and only 16.8 percent of children (half of the proportion declared by the fathers) spent five months or more with him.

Table 3: Distribution of biological or adopted children aged 0-17 years reported by their father or mother, according to the number of days spent with their father in the twelve months preceding the survey
Number of days spent with the father Respondent
Father a Mother b
%
None 16.8 24.7
1-6 days (less than 1 week) 6.6 9.4
7-29 days (less than 1 month) 8.2 11.6
30-59 days (1 to 2 months) 14.6 17.0
60-149 days (2 to 5 months) 23.4 20.5
150-209 days (5 to 7 months) 13.5 6.7
210-299 days (7 to 10 months) 4.0 3.3
300-365 days (10 months and over) 12.9 6.8
Total 100  100 
N c 418  676 

Source: Statistics Canada, General Social Survey (GSS), Cycle 10: the Family, 1995.

In the following analyses, we have grouped the children in four categories, according to the time spent with their father. The first category contained those children who had, for all practical purposes, lost contact with their father, grouping together those who had spent fewer than seven days in the previous year with him. At the opposite extreme, we sought to isolate the children who had maintained a close relationship with their father, including in the same category children who spent five months (150 days) or more with him in the 12 months preceding the survey.[3] Children spending one week to two months (7-59 days) and two to five months (60-149 days) with their father formed the two intermediate categories. The distribution of children at the centre of almost all the ensuing analyses can be summarized as follows:

  1. less than one week: little or no contact (23.4 percent of children);
  2. between one week and two months: low level of contact (22.8 percent of children);
  3. between two and five months: regular contact (23.4 percent) of children);
  4. five months or more: close contact (30.4 percent of children).

One criticism that might be levelled against the approach taken here is that spending time with someone is not the only way to maintain links. Making a telephone call or writing a letter could also be effective means of keeping contact, particularly when distance prevents frequent visits. Using the GSS data, we attempted to measure the extent to which these means could serve as an alternative to visiting. The findings, presented in Table 4 and Figure 1, reassured us immediately as to the risk of underestimating the level of father/child contact that might be incurred by restricting our measurement of this contact to the amount of time fathers and children spent in each other's company.

In the course of the 1995 GSS, separated respondents were asked how often, in the previous 12 months, they had contacted each child by letter or telephone when the child was not with them; they were also asked for an estimate of this type of contact with the other parent. The first question was not asked for children declared to be living full-time with the respondent; as a result, 67[4] of the 418 children reported by their father have been excluded from the following analysis.

Table 4: Distribution (%) of children aged 0-17 years not living full-time with their father, according to the time spent with him and the frequency of letter and telephone contact with him during the year preceding the survey
  Time spent with father
Frequency of contact
by letter or telephone a
Less than
1 week
1 week to
2 months
2 to 5 months 5 months
and over
Total
Daily 5.6 4.0 10.8 31.3 11.4
At least once a week 17.3 56.3 67.1 47.3 46.9
At least once a month 22.9 22.8 13.2 8.0 17.4
Less than once a month 17.4 9.1 2.2 4.5 8.6
Never 36.8 7.8 6.7 8.9 15.7
Total 100 100 100 100 100
N b 98 92 95 66 351

2 = 111.717, p = 0.001, missing cases: 1

Source: Statistics Canada, General Social Survey (GSS), Cycle 10: the Family, 1995.

Figure 1: Frequency of contact by letter or telephone according to time spent with father

Figure 1: Frequency sof contact by letter or telephone according to time spent with father

[ Description ]

A close look at Figure 1 clearly shows that contact by letter or telephone is far from being used as a substitute for visits by fathers who rarely see their children. The opposite situation seems to be the case: the less often fathers see their children, the less likely they are to communicate with them by letter or telephone. Almost 80 percent of children with frequent contact with their father (five months or more) communicate at least once a week with him by letter or telephone when living with their mother, compared with less than a quarter of children who scarcely ever see their father (less than one week). In fact, more than half the children who spent less than a week a year with their father were not even in contact by letter or telephone on a monthly basis. It is also of interest that children who rarely see their father but who have regular contact with him by letter or telephone do not appear to communicate in this way because of distance separating the parents' households. In almost half the cases, the children lived less than 10 kilometres from their father and very few lived more than 100 kilometres away (data not presented). It seems likely, therefore, that the variable constructed as an indicator of the frequency of father/child contact adequately represents the extent of the contact that separated fathers had with their children in the previous year.

As mentioned earlier, certain analyses will be based on the father rather than the child sample. As our aim is to characterise the fathers according to the degree of contact they had with their children, using the father sample raised the question of how to classify fathers who did not have the same level of contact with all their children? Of the 291 fathers of the children in the sample (see Table 1), 110 had more than one child; of these, 40 percent reported spending a different number of days with each child. In 60 percent of cases, the frequency of contact varies from one child to another even when the number of days is classified into four categories. How should a father's level of contact be best represented, therefore--should preference be given to the time spent with the oldest child, or with the youngest? After a detailed examination of the data, we concluded that calculating the average number of days spent by the father with his children produced the best overall measure of contact. The father was then classified into the appropriate group of the four-category father/child contact variable according to this estimated average number of days. A comparison of the two samples--fathers and children--distributed according to the time spent with the father, is presented in Table 5. It is interesting to note the similarity between the two series of percentages.

Table 5: Number of children and fathers according to the frequency of father/child contact

Number of children
  Frequency of father/child contact
Less than 1 week 1 week to
2 months
2 to 5 months 5 months
and over
Total
N a 98 95 98 127 418
% 23.4 22.8 23.4 30.4 100

Number of fathers
  Frequency of father/child contact
Less than 1 week 1 week to
2 months
2 to 5 months 5 months
and over
Total
N a 67 69 68 87 291
% 23.1 23.8 23.3 29.8 100

Source: Statistics Canada, General Social Survey (GSS), Cycle 10: the Family, 1995.
aWeighted data, brought back to the original sample size.


Date modified: