Research Quarterly for Exercise and Sport
ISSN: 0270-1367 (Print) 2168-3824 (Online) Journal homepage: http://www.tandfonline.com/loi/urqe20
Familial Aggregation in Physical Activity Patty S. Freedson & Sherrie Evenson To cite this article: Patty S. Freedson & Sherrie Evenson (1991) Familial Aggregation in Physical Activity, Research Quarterly for Exercise and Sport, 62:4, 384-389, DOI: 10.1080/02701367.1991.10607538 To link to this article: https://doi.org/10.1080/02701367.1991.10607538
Published online: 08 Feb 2013.
Submit your article to this journal
Article views: 29
Citing articles: 99 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=urqe20
Ruearch QUlrterly for Exercise andSport tel 1991 bythe American Alliance for Health,
Physical Education, Recreation and Dance Vol. 62. No. 4, pp. 384-389
Familial Aggregation in Physical Activity PattyS. Freedson andSherrie Evenson The purposes ofthis study were to (a) examine the stability and consistency ofthe Galtrac accelerometer (Hemokinetics, Madison, WI) and an activity record to assess physicalactivity in children and adults (Experiment 1), and (b) to determine if there is a relationship between parents and their children in physicalactivity level(Experiment 2). Thirty 5-9-year-old children and their biological parents wore Galtrac accelerometers for three consecutive days (including one weeJumd day). At the same time, parents completed a Caltrac Activity Record (CAL REG)for themselves and theirchild.Dependentvariables were counts per dayfor the Galtrac and minutes oflight activity and activityfor the CAL REG. Between-day correlations for the Galtrac rangedfrom r =. 73 to .87 for the parents (p < .001) andfromr = .38 (p < .04) to. 79 (p < • 001) for the children. An analysis ofvariance with repeated measures indicated no significant differences for the Galtrac between daysfor parents and children. Between-day correlations for CAL REG rangedfrom r = .67 to. 91 (p < .05) for parents and r = .36 to . 72 (p < .05) for children, and there were no significant differences between days. In Experiment 2, X2 analyses were used to examinefamilial resemblance in physical activity. Using the Galtrac, familial resemblance occurred in 67% (father and child) and 73 % (mother and child) ofthe families. Using the CAL REG,familial aggregation was present in 70 % (fatherand child) and 66 % (motherand child) of the
families. Thus, children of active and less activeparents exhibited physicalactivitypatterns similar to theirparents. Key words: physical activity, familial resemblance, children, Caltrac accelerometer
F
am ilial aggregation, the combination of genetic and environmental influences within a family, has been studied in relation to many factors associatedwith health. These factors include blood pressure (Laskarzewski et al., 1981), blood lipids (Morrison et al., 1982), and body composition (Charney, Goodman, McBride, Lyon, & Pratt, 1976; Gam & Lavelle, 1985; Savard, Bouchard, LeBlanc, & Tremblay, 1983). Familial "clustering" of multiple coronary artery disease (CAD) risk factors is also well recognized (Feinleib et al., 1977; Laskarzewski et al., 1981; Snowden et al., 1982). Physical inactivity has been documented as a risk factor associated with the incidence of CAD (Morris, Chave, Adam, Sirey, & Epstein, 1973; Paffenbarger & Hale, 1975; Paffenbarger, Hyde, Wing, & Hsieh, 1986; Paffenbarger, Laughlin, Gima, & Black, 1970; Paffenbarger, Wing, & Hyde, 1978; Snowden etal., 1982). Research byPaffenbargeretal. (1970,1975,1978,1986) and Blair et al. (1989) has consistently suggested inverse relationships between physical activity or fitness and
Patty S. Freedson is affiliated with the Exercise Physiology Laboratory at the University of Massachusetts. Sherrie Evenson is affiliated with the Cardiopulmonary Rehabilitation Institute in Seattle, WA. Date submitted: July~ 1990 Revision accepted: May4 1991
384
coronary disease. Moreover, research by Strong and McGill (1969) suggests that the pathology of heart disease begins early in life and that children with high levels ofrisk track or retain this relative position into adulthood (Montoye, 1985). Researchers, however, have only recently begun to examine the familial similarities in physical activity, despite its potential role in CAD. For example, a study by Sallis, Patterson, Buono, Atkins, and Nader (1988) using an observation method of assessment on 33 preschool children during free-play found that selected familial variables contributed to 39% ofthe variance in children's moderate activity. Sallis et al, (1988) later used 7-day activity record on 206 families with 5th--6th grade children and observed significant familial associations between mothers and both their younger and older children for time spent in physically demanding leisure activities (r= .20 and .50, respectively, p< .05). Significant familial resemblance in physical activity between parents and their offspring has been reported (Moore et al., 1991; Perusse, LeBlanc, & Bouchard, 1988; Perusse, Tremblay, LeBlanc, & Bouchard, 1989). One reason research in this area has been limited is the lack ofvalid and reliable methods to assess physical activity in both adults and children. Because physical activity cannot be directly measured in a laboratory, much variability exists between and within studies attempting to quantify physical activity (LaPorte et al., 1982). Ifthe relationships ofphysical activitywithin families are to be examined, methods ofassessing physical activity must be improved and standardized. One of the most
ROES: December 1991
Freedson andEvenson
practical and widely used methods ofmeasuring physical activity is the self-reportassessment. Though it is believed that self-report is useful in distinguishing inactive from active individuals (Baranowski et al., 1984; Taylor et al., 1984; Washburn, Adams, & Haile, 1987), there are problems associated with this subjective form of measurement. These include interference, accuracy ofrecall, and definitions of work (LaPorte et al., 1982). Accuracy of recall is a special problem with children, as children may give restricted and inaccurate information about their activity until about 10 to 12yearsofage (Baranowskietal., 1984; Saris, 1986; Telama et al., 1985). Thus, activity assessment techniques in adults and children need to be improved to examine potential familial associations. The present investigation had two purposes: (a) to examine an objective and a subjective method of physical activity assessment in children and adults, and (b) to explore the familial aggregation of physical activity patterns between parents and their children.
EXPERIMENT 1 Method Subjects Thirty 5-9-year-old children (13 males, 17 females) and their biological parents volunteered to participate in this study. Both parents must have raised the child together from the child's birth and must have currently been living at the same residence as the child. Only one child was used from a family. No family member had physically limiting handicaps, and the child had no history ofhyperactivity. An informed consent documentwas signed by the parents and the child. Table 1 presents pertinent descriptive characteristics of the parents and children.
Table 1. Descriptive characteristics of parentsand children Variable
Age (yrs)
M
SO range
Height (em)
M
SO range
Mass (kg)
M
SO range
Fathers
Mothers
Children
(n=3O)
(n=3O)
(n=3O)
37.4 4.99 25.:H8.o
36.2 5.51 24.1-49.2
7.0 1.22 5.3-9.5
176.45 8.424 148.59-189.23
162.54 6.200 153.04-175.26
121.55 9.367 104.14-149.86
79.2 11.57 63.6-113.6
61.6 11.00 46.4-85.0
228 3.55 15.G--3o.9
RDES: December 1991
Materials andProcedure To objectively assess physical activity, each family member wore a Caltrac accelerometer on the nondominant hip for three consecutive 12-hour days (two weekdays and one weekend day). The Caltrac, designed to estimate energy expenditure in kilocalories, was used solely as a physical activity counter in the present investigation. Therefore, standard constants of personal data (height = 36 in, weight = 25 lb, age = 99 years, gender = male) were used for all family members. The function of these specific constants was to eliminate the estimation ofbasal metabolic rate so that only movement was registered on the Caltrac. The Caltrac activity score (counts/day) (CAL CT) for parents and children was recorded at the end of each 12-hour day by the parent. Parents simultaneously completed a Caltrac Activity Record (CAL REC) for themselves and their child. The CAL REC provided information on the frequency, intensity, duration, and types of activities in which the family members participated each day. The CAL REC was divided into three major time blocks: morning, afternoon, and evening. Parents reported duration and intensity of both their own and their child's activities based on lists of standard reference activities (Passmore & Durnin, 1955). Parents were instructed to question children about their activity during periods the children were not in the presence ofthe parents (i.e., before recess, during recess, and after recess). These activities were recorded by the parentwhen the child came home from school. Activities were categorized into four levels of intensity modified by Tayloretal. (1984): inactive/light (IN/LT = < 3.5 METS) , moderate (MD = 3.6-5.5 METS) , hard (HD == 5.6-9 METS), and very hard (VHD = 9.1 + METS) , and minutes spent in each category were computed.
Data analysis All measures of physical activity were normally distributed. Thus, an analysis of variance with repeated measures (REANOVA) was used to identify the stability of the two physical activity assessment methods across days. Separate REANOVAswere performed for the mothers, fathers, and children for CAL CT (counts per day) and CAL REC (min of light activity and activity). The correlations between the 3 days for CAL CT and CAL REC variables (i.e., minutes spent in IN/LT, MD, HD, and VI-! activities) and the relationship between the two methods were described by Pearson product moment correlations.
Results The REANOVA did not detect significant differences in the means of the CAL CT or CAL REC variables
385
Freedson andEvenson
across days for any family member. Correlations between days for the CAL CT were highly significant for parents, ranging from r (58) = .73 to .87 (P< .001). The associations for children ranged from r (28) = .38 (P < .05) to r (28) = .79 (P < .001). The nonsignificant mean differences between days justified the use of the three-day mean to represent the CAL CT (counts per day) (M ± SD = 77 ± 43.8, 81 ± 40.3 and 116 ± 25.4 cts for fathers, mothers, and children, respectively). The children's CAL CT, however, was significantly higher (P < .05) than the fathers' andmothers',F(3,82) = 7.5,p< .01 (sphericity assumption satisfied). Correlations between days for the CAL REC variables ranged from r (58) = .67 to .91 (P< .05) for the parents and r (28) = .36 to .72 (P < .05) for the children. The mean of the three days for CAL REC was used for further analysis. The mean time spent in IN/LT intensity activities was referred to as light activity (LTACT), and the mean time spent in activities above LTACT intensity (MD, HD, and VH) was summed and referred to as activity (ACT). There were no significant differences in the amount of time parents and children spent in ACT (M ± SD = 229 ± 149.1, 250 ± 133.5 and 232 ± 117.4 min for fathers, mothers, and children, respectively). The correlations between the two methods ofassessment were statistically significant for both parents and children: r (28) = -.30, -.47, and -.39 (P < .05) between CALCTandLTACT,andr(28) = .32,.50,and.35 (P< .05) between CAL CT and ACT for fathers, mothers, and children, respectively.
Discussion No significant differences were present for mothers, fathers, and children across days for CAL CT (counts) and CAL REC (min ofLTACT and ACT). Further, the correlations between days for CAL CT were significant between all three days for parents (r[58] = .73 to .87, P< .001), which compare favorably with the test-retest reliability on the Caltrac in a study on adult males and females by Washburn and LaPorte (1988) (r= .70 and r= .87 for normal and fast-paced walking, respectively). These results imply that activity ofthe parents across days was consistent and the Caltrac was a reliable measure of movement. The mean CAL CT was 68% higher for the children compared to the parents (P< .05), suggesting the children were more physically active than their parents. However, the children did not appear to be as consistent as parentsacross days (r[28] =.38to.79,p< .05). SimonsMorton, O'Hara, Simons-Morton, and Parcel (1987) have stated that children's physical activity tends to be sporadic, which may explain the lower consistency across days. Thus, it may be necessary to obtain several days of
386
Caltrac activity measures to characterize daily activity in children. The correlations between days for the CAL REC variables (IN/LT, MD, HD, VH) were variable for both parents and children, ranging from r= .36 to .91. There were no significant differences in the amount of time fathers, mothers, and children spent in ACT (32%,35%, and 33%, respectively). The CAL REC was designed to reduce problems associated with memory in self-reported physical activity. Baranowski et aI. (1984) reported that daily self-report was associated with fewer memory difficulties than when activity was collected by recall. Telama et al. (1985) have stated that the use of a daily diary method is probably more valid than a questionnaire when assessing a child's activity level. Baranowski et al, (1984) showed that the accuracy in a child's perception of time was increased when the day was segmented into functional time components. Further, due to potential ambiguity in defining physical activity intensity, summing up the time spent in MD, HD, and VH activity levels (ACT) was an attempt to reduce differences in categorizing physical activity. The results suggest these two methods of physical activity assessment may be providing measures of different dimensions of activity (r [28] = -.30 to .50, P< .05, between CAL CT and CAL REC for children and adults). Although the correlations are significant, the moderate common variance suggests these two methods provide information on different aspects of physical activity. The CAL CT represents the mechanical frequency of movement and appears to best measure walking-type activities (Washburn et aI., 1987) and activities performed in the vertical plane. The Caltrac, therefore, may neglect to account for activities performed in the nonvertical plane, as well as intense activities (Klesges & Klesges, 1987). The CAL REC describes the minutes spent in activity and light activity. Though this is a subjective account of the subjects' physical activity, the information provided may detect activity not reflected by the mechanical monitor. Further, Taylor et al, (1984) have reported that participants most accurately recall conditioning activities. If these activities were intense, performed in a nonvertical plane, or did not involve torso movement, the Caltrac score may not have correctly represented physical activity. It should also be noted that a certain amount of error may be introduced to the CAL REC, which uses caloric expenditure estimates to place activities into one of the four intensity categories. The energy expenditure tables ofPassmore and Durnin (1955) were used for these caloric estimates. Since parents reported both their own and their child's activities, the CAL REC may have been biased. Nevertheless, the significant correlations between CAL REC and CAL CT suggest that if reporting of the children's activities by the parents was biased, the total amount of activity reported on the CAL REC was appropriately categorized.
ROES: December 1991
Freedson andEvenson
Therefore, due to the difficulties encountered in assessing physical activity, the independent use of objective and subjective techniques may be less relevant than regarding their association. Thus, assessing physical activity may be more meaningful when the combination of these methods is employed, particularly when assessing the physical activity of children.
higher activity. Subsequently any family member with a value below the 50th percentile was defined as LOACT, and any family memberwith a value greater than or equal to the 50th percentile was defined as HIACT. This was then repeated for the CAL REC assessmentmethod using the 50th percentile for ACT time (sum of moderate, hard, and very hard activities) as the boundary separating LOACT and HIACT.
Data Analysis
EXPERIMENT 2 The family members were sorted into a joint frequency distribution presented in 2 x 2 contingency tables (active and inactive fathers, mothers, and children) and analyzed using X2•
Method Subjects The sample of30 families discussed in Experiment 1 was then examined to determine whether children of higher and lower activity parents had activity patterns similar to their parents.
Results
To classify lower and higher activity, the 50th percentile Caltrac activity score was identified for fathers, mothers, and children (70,71,117 cts, respectively). Based on this value, parents and children were divided into two categories: LOACT, which contained family members of lower activity, and HIACT, which contained those of
Tables 2 and 3 present the XZ analyses for children and fathers and children and mothers for Caltrac activity assessment (see Table 2) and activity record assessment (see Table 3). Familial aggregation of physical activity based on CAL CT between father and child occurred in 20 families, or 67% of the sample, which approached statistical significance (XZ[l, N= 30] = 3.45, p « .06). The occurrence of LOACT FATHERS with a HIACT CHILD was 13% and 20% for HIACT FATHERS with a LOACT CHILD.
Table 2. 1,1 Analysis for familial aggregation in physical activity:
Table 3. X2 Analysis forfamilial aggregationin physical activity:
Caltrac
Caltrac record
Materials andProcedures
low Activity Child low Activity
High Activity
10 33.30/0 4
13.3%
Fathers High Activity
6
low Activity Child low Activity
20.0% 10 33.3%
x III = 3.45, P< .06
"
5 16.7%
3
11 36.7%
36.7%
10.0%
X2 (1I = 6.72, p« .01
ROES: December 1991
4
10 34.5%
13.8%
Mothers High Activity low Activity
High Activity
4 13.8%
X2 (1I =5.81, p « .01
2
Child low Activity
"
37.9% High Activity
Fathers High Activity
Mothers low Activity High Activity Child low Activity High Activity
10 34.5%
5 17.2%
4
10 34.5%
13.8% X2 (11 =4.21, p < .04
381
Freedson andEvenson
Familial aggregation of physical activity based on CAL REC between father and child occurred in 21 families, or 70% of the sample, which was highly significant (X2 [1, N = 30] = 5.81, P< .05). Twenty-eight percent of the families showed no association. Familial aggregation of physical activity based on CAL CT between mothers and their child occurred in 22 families, or 73% of the sample, which was highly significant (X2[l, N= 30] = 6.72, .05). There was less occurrence of LOACT mothers with a HIACT child (10%) compared to HIACT MOTHERS with a LOACT CHILD (17%). CAL REC analysis revealed that 20, or 66%, of the mothers had a similar activity classification as their children. The percentage of inactive children when (a) neither parent was active, (b) one parent was active, and (c) both parentswere active was also examined. These results are presented in Table 4. These data suggest the presence of a gradient across categories where the largest proportion of inactive children are in families where neither parent is active. Moreover, if both parents were in the HIACT category, nearly all children (93-97%) were HIACT.
«
ISSN: 0270-1367 (Print) 2168-3824 (Online) Journal homepage: http://www.tandfonline.com/loi/urqe20
Familial Aggregation in Physical Activity Patty S. Freedson & Sherrie Evenson To cite this article: Patty S. Freedson & Sherrie Evenson (1991) Familial Aggregation in Physical Activity, Research Quarterly for Exercise and Sport, 62:4, 384-389, DOI: 10.1080/02701367.1991.10607538 To link to this article: https://doi.org/10.1080/02701367.1991.10607538
Published online: 08 Feb 2013.
Submit your article to this journal
Article views: 29
Citing articles: 99 View citing articles
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=urqe20
Ruearch QUlrterly for Exercise andSport tel 1991 bythe American Alliance for Health,
Physical Education, Recreation and Dance Vol. 62. No. 4, pp. 384-389
Familial Aggregation in Physical Activity PattyS. Freedson andSherrie Evenson The purposes ofthis study were to (a) examine the stability and consistency ofthe Galtrac accelerometer (Hemokinetics, Madison, WI) and an activity record to assess physicalactivity in children and adults (Experiment 1), and (b) to determine if there is a relationship between parents and their children in physicalactivity level(Experiment 2). Thirty 5-9-year-old children and their biological parents wore Galtrac accelerometers for three consecutive days (including one weeJumd day). At the same time, parents completed a Caltrac Activity Record (CAL REG)for themselves and theirchild.Dependentvariables were counts per dayfor the Galtrac and minutes oflight activity and activityfor the CAL REG. Between-day correlations for the Galtrac rangedfrom r =. 73 to .87 for the parents (p < .001) andfromr = .38 (p < .04) to. 79 (p < • 001) for the children. An analysis ofvariance with repeated measures indicated no significant differences for the Galtrac between daysfor parents and children. Between-day correlations for CAL REG rangedfrom r = .67 to. 91 (p < .05) for parents and r = .36 to . 72 (p < .05) for children, and there were no significant differences between days. In Experiment 2, X2 analyses were used to examinefamilial resemblance in physical activity. Using the Galtrac, familial resemblance occurred in 67% (father and child) and 73 % (mother and child) ofthe families. Using the CAL REG,familial aggregation was present in 70 % (fatherand child) and 66 % (motherand child) of the
families. Thus, children of active and less activeparents exhibited physicalactivitypatterns similar to theirparents. Key words: physical activity, familial resemblance, children, Caltrac accelerometer
F
am ilial aggregation, the combination of genetic and environmental influences within a family, has been studied in relation to many factors associatedwith health. These factors include blood pressure (Laskarzewski et al., 1981), blood lipids (Morrison et al., 1982), and body composition (Charney, Goodman, McBride, Lyon, & Pratt, 1976; Gam & Lavelle, 1985; Savard, Bouchard, LeBlanc, & Tremblay, 1983). Familial "clustering" of multiple coronary artery disease (CAD) risk factors is also well recognized (Feinleib et al., 1977; Laskarzewski et al., 1981; Snowden et al., 1982). Physical inactivity has been documented as a risk factor associated with the incidence of CAD (Morris, Chave, Adam, Sirey, & Epstein, 1973; Paffenbarger & Hale, 1975; Paffenbarger, Hyde, Wing, & Hsieh, 1986; Paffenbarger, Laughlin, Gima, & Black, 1970; Paffenbarger, Wing, & Hyde, 1978; Snowden etal., 1982). Research byPaffenbargeretal. (1970,1975,1978,1986) and Blair et al. (1989) has consistently suggested inverse relationships between physical activity or fitness and
Patty S. Freedson is affiliated with the Exercise Physiology Laboratory at the University of Massachusetts. Sherrie Evenson is affiliated with the Cardiopulmonary Rehabilitation Institute in Seattle, WA. Date submitted: July~ 1990 Revision accepted: May4 1991
384
coronary disease. Moreover, research by Strong and McGill (1969) suggests that the pathology of heart disease begins early in life and that children with high levels ofrisk track or retain this relative position into adulthood (Montoye, 1985). Researchers, however, have only recently begun to examine the familial similarities in physical activity, despite its potential role in CAD. For example, a study by Sallis, Patterson, Buono, Atkins, and Nader (1988) using an observation method of assessment on 33 preschool children during free-play found that selected familial variables contributed to 39% ofthe variance in children's moderate activity. Sallis et al, (1988) later used 7-day activity record on 206 families with 5th--6th grade children and observed significant familial associations between mothers and both their younger and older children for time spent in physically demanding leisure activities (r= .20 and .50, respectively, p< .05). Significant familial resemblance in physical activity between parents and their offspring has been reported (Moore et al., 1991; Perusse, LeBlanc, & Bouchard, 1988; Perusse, Tremblay, LeBlanc, & Bouchard, 1989). One reason research in this area has been limited is the lack ofvalid and reliable methods to assess physical activity in both adults and children. Because physical activity cannot be directly measured in a laboratory, much variability exists between and within studies attempting to quantify physical activity (LaPorte et al., 1982). Ifthe relationships ofphysical activitywithin families are to be examined, methods ofassessing physical activity must be improved and standardized. One of the most
ROES: December 1991
Freedson andEvenson
practical and widely used methods ofmeasuring physical activity is the self-reportassessment. Though it is believed that self-report is useful in distinguishing inactive from active individuals (Baranowski et al., 1984; Taylor et al., 1984; Washburn, Adams, & Haile, 1987), there are problems associated with this subjective form of measurement. These include interference, accuracy ofrecall, and definitions of work (LaPorte et al., 1982). Accuracy of recall is a special problem with children, as children may give restricted and inaccurate information about their activity until about 10 to 12yearsofage (Baranowskietal., 1984; Saris, 1986; Telama et al., 1985). Thus, activity assessment techniques in adults and children need to be improved to examine potential familial associations. The present investigation had two purposes: (a) to examine an objective and a subjective method of physical activity assessment in children and adults, and (b) to explore the familial aggregation of physical activity patterns between parents and their children.
EXPERIMENT 1 Method Subjects Thirty 5-9-year-old children (13 males, 17 females) and their biological parents volunteered to participate in this study. Both parents must have raised the child together from the child's birth and must have currently been living at the same residence as the child. Only one child was used from a family. No family member had physically limiting handicaps, and the child had no history ofhyperactivity. An informed consent documentwas signed by the parents and the child. Table 1 presents pertinent descriptive characteristics of the parents and children.
Table 1. Descriptive characteristics of parentsand children Variable
Age (yrs)
M
SO range
Height (em)
M
SO range
Mass (kg)
M
SO range
Fathers
Mothers
Children
(n=3O)
(n=3O)
(n=3O)
37.4 4.99 25.:H8.o
36.2 5.51 24.1-49.2
7.0 1.22 5.3-9.5
176.45 8.424 148.59-189.23
162.54 6.200 153.04-175.26
121.55 9.367 104.14-149.86
79.2 11.57 63.6-113.6
61.6 11.00 46.4-85.0
228 3.55 15.G--3o.9
RDES: December 1991
Materials andProcedure To objectively assess physical activity, each family member wore a Caltrac accelerometer on the nondominant hip for three consecutive 12-hour days (two weekdays and one weekend day). The Caltrac, designed to estimate energy expenditure in kilocalories, was used solely as a physical activity counter in the present investigation. Therefore, standard constants of personal data (height = 36 in, weight = 25 lb, age = 99 years, gender = male) were used for all family members. The function of these specific constants was to eliminate the estimation ofbasal metabolic rate so that only movement was registered on the Caltrac. The Caltrac activity score (counts/day) (CAL CT) for parents and children was recorded at the end of each 12-hour day by the parent. Parents simultaneously completed a Caltrac Activity Record (CAL REC) for themselves and their child. The CAL REC provided information on the frequency, intensity, duration, and types of activities in which the family members participated each day. The CAL REC was divided into three major time blocks: morning, afternoon, and evening. Parents reported duration and intensity of both their own and their child's activities based on lists of standard reference activities (Passmore & Durnin, 1955). Parents were instructed to question children about their activity during periods the children were not in the presence ofthe parents (i.e., before recess, during recess, and after recess). These activities were recorded by the parentwhen the child came home from school. Activities were categorized into four levels of intensity modified by Tayloretal. (1984): inactive/light (IN/LT = < 3.5 METS) , moderate (MD = 3.6-5.5 METS) , hard (HD == 5.6-9 METS), and very hard (VHD = 9.1 + METS) , and minutes spent in each category were computed.
Data analysis All measures of physical activity were normally distributed. Thus, an analysis of variance with repeated measures (REANOVA) was used to identify the stability of the two physical activity assessment methods across days. Separate REANOVAswere performed for the mothers, fathers, and children for CAL CT (counts per day) and CAL REC (min of light activity and activity). The correlations between the 3 days for CAL CT and CAL REC variables (i.e., minutes spent in IN/LT, MD, HD, and VI-! activities) and the relationship between the two methods were described by Pearson product moment correlations.
Results The REANOVA did not detect significant differences in the means of the CAL CT or CAL REC variables
385
Freedson andEvenson
across days for any family member. Correlations between days for the CAL CT were highly significant for parents, ranging from r (58) = .73 to .87 (P< .001). The associations for children ranged from r (28) = .38 (P < .05) to r (28) = .79 (P < .001). The nonsignificant mean differences between days justified the use of the three-day mean to represent the CAL CT (counts per day) (M ± SD = 77 ± 43.8, 81 ± 40.3 and 116 ± 25.4 cts for fathers, mothers, and children, respectively). The children's CAL CT, however, was significantly higher (P < .05) than the fathers' andmothers',F(3,82) = 7.5,p< .01 (sphericity assumption satisfied). Correlations between days for the CAL REC variables ranged from r (58) = .67 to .91 (P< .05) for the parents and r (28) = .36 to .72 (P < .05) for the children. The mean of the three days for CAL REC was used for further analysis. The mean time spent in IN/LT intensity activities was referred to as light activity (LTACT), and the mean time spent in activities above LTACT intensity (MD, HD, and VH) was summed and referred to as activity (ACT). There were no significant differences in the amount of time parents and children spent in ACT (M ± SD = 229 ± 149.1, 250 ± 133.5 and 232 ± 117.4 min for fathers, mothers, and children, respectively). The correlations between the two methods ofassessment were statistically significant for both parents and children: r (28) = -.30, -.47, and -.39 (P < .05) between CALCTandLTACT,andr(28) = .32,.50,and.35 (P< .05) between CAL CT and ACT for fathers, mothers, and children, respectively.
Discussion No significant differences were present for mothers, fathers, and children across days for CAL CT (counts) and CAL REC (min ofLTACT and ACT). Further, the correlations between days for CAL CT were significant between all three days for parents (r[58] = .73 to .87, P< .001), which compare favorably with the test-retest reliability on the Caltrac in a study on adult males and females by Washburn and LaPorte (1988) (r= .70 and r= .87 for normal and fast-paced walking, respectively). These results imply that activity ofthe parents across days was consistent and the Caltrac was a reliable measure of movement. The mean CAL CT was 68% higher for the children compared to the parents (P< .05), suggesting the children were more physically active than their parents. However, the children did not appear to be as consistent as parentsacross days (r[28] =.38to.79,p< .05). SimonsMorton, O'Hara, Simons-Morton, and Parcel (1987) have stated that children's physical activity tends to be sporadic, which may explain the lower consistency across days. Thus, it may be necessary to obtain several days of
386
Caltrac activity measures to characterize daily activity in children. The correlations between days for the CAL REC variables (IN/LT, MD, HD, VH) were variable for both parents and children, ranging from r= .36 to .91. There were no significant differences in the amount of time fathers, mothers, and children spent in ACT (32%,35%, and 33%, respectively). The CAL REC was designed to reduce problems associated with memory in self-reported physical activity. Baranowski et aI. (1984) reported that daily self-report was associated with fewer memory difficulties than when activity was collected by recall. Telama et al. (1985) have stated that the use of a daily diary method is probably more valid than a questionnaire when assessing a child's activity level. Baranowski et al, (1984) showed that the accuracy in a child's perception of time was increased when the day was segmented into functional time components. Further, due to potential ambiguity in defining physical activity intensity, summing up the time spent in MD, HD, and VH activity levels (ACT) was an attempt to reduce differences in categorizing physical activity. The results suggest these two methods of physical activity assessment may be providing measures of different dimensions of activity (r [28] = -.30 to .50, P< .05, between CAL CT and CAL REC for children and adults). Although the correlations are significant, the moderate common variance suggests these two methods provide information on different aspects of physical activity. The CAL CT represents the mechanical frequency of movement and appears to best measure walking-type activities (Washburn et aI., 1987) and activities performed in the vertical plane. The Caltrac, therefore, may neglect to account for activities performed in the nonvertical plane, as well as intense activities (Klesges & Klesges, 1987). The CAL REC describes the minutes spent in activity and light activity. Though this is a subjective account of the subjects' physical activity, the information provided may detect activity not reflected by the mechanical monitor. Further, Taylor et al, (1984) have reported that participants most accurately recall conditioning activities. If these activities were intense, performed in a nonvertical plane, or did not involve torso movement, the Caltrac score may not have correctly represented physical activity. It should also be noted that a certain amount of error may be introduced to the CAL REC, which uses caloric expenditure estimates to place activities into one of the four intensity categories. The energy expenditure tables ofPassmore and Durnin (1955) were used for these caloric estimates. Since parents reported both their own and their child's activities, the CAL REC may have been biased. Nevertheless, the significant correlations between CAL REC and CAL CT suggest that if reporting of the children's activities by the parents was biased, the total amount of activity reported on the CAL REC was appropriately categorized.
ROES: December 1991
Freedson andEvenson
Therefore, due to the difficulties encountered in assessing physical activity, the independent use of objective and subjective techniques may be less relevant than regarding their association. Thus, assessing physical activity may be more meaningful when the combination of these methods is employed, particularly when assessing the physical activity of children.
higher activity. Subsequently any family member with a value below the 50th percentile was defined as LOACT, and any family memberwith a value greater than or equal to the 50th percentile was defined as HIACT. This was then repeated for the CAL REC assessmentmethod using the 50th percentile for ACT time (sum of moderate, hard, and very hard activities) as the boundary separating LOACT and HIACT.
Data Analysis
EXPERIMENT 2 The family members were sorted into a joint frequency distribution presented in 2 x 2 contingency tables (active and inactive fathers, mothers, and children) and analyzed using X2•
Method Subjects The sample of30 families discussed in Experiment 1 was then examined to determine whether children of higher and lower activity parents had activity patterns similar to their parents.
Results
To classify lower and higher activity, the 50th percentile Caltrac activity score was identified for fathers, mothers, and children (70,71,117 cts, respectively). Based on this value, parents and children were divided into two categories: LOACT, which contained family members of lower activity, and HIACT, which contained those of
Tables 2 and 3 present the XZ analyses for children and fathers and children and mothers for Caltrac activity assessment (see Table 2) and activity record assessment (see Table 3). Familial aggregation of physical activity based on CAL CT between father and child occurred in 20 families, or 67% of the sample, which approached statistical significance (XZ[l, N= 30] = 3.45, p « .06). The occurrence of LOACT FATHERS with a HIACT CHILD was 13% and 20% for HIACT FATHERS with a LOACT CHILD.
Table 2. 1,1 Analysis for familial aggregation in physical activity:
Table 3. X2 Analysis forfamilial aggregationin physical activity:
Caltrac
Caltrac record
Materials andProcedures
low Activity Child low Activity
High Activity
10 33.30/0 4
13.3%
Fathers High Activity
6
low Activity Child low Activity
20.0% 10 33.3%
x III = 3.45, P< .06
"
5 16.7%
3
11 36.7%
36.7%
10.0%
X2 (1I = 6.72, p« .01
ROES: December 1991
4
10 34.5%
13.8%
Mothers High Activity low Activity
High Activity
4 13.8%
X2 (1I =5.81, p « .01
2
Child low Activity
"
37.9% High Activity
Fathers High Activity
Mothers low Activity High Activity Child low Activity High Activity
10 34.5%
5 17.2%
4
10 34.5%
13.8% X2 (11 =4.21, p < .04
381
Freedson andEvenson
Familial aggregation of physical activity based on CAL REC between father and child occurred in 21 families, or 70% of the sample, which was highly significant (X2 [1, N = 30] = 5.81, P< .05). Twenty-eight percent of the families showed no association. Familial aggregation of physical activity based on CAL CT between mothers and their child occurred in 22 families, or 73% of the sample, which was highly significant (X2[l, N= 30] = 6.72, .05). There was less occurrence of LOACT mothers with a HIACT child (10%) compared to HIACT MOTHERS with a LOACT CHILD (17%). CAL REC analysis revealed that 20, or 66%, of the mothers had a similar activity classification as their children. The percentage of inactive children when (a) neither parent was active, (b) one parent was active, and (c) both parentswere active was also examined. These results are presented in Table 4. These data suggest the presence of a gradient across categories where the largest proportion of inactive children are in families where neither parent is active. Moreover, if both parents were in the HIACT category, nearly all children (93-97%) were HIACT.
«
