Abstract
The purpose of this study was to examine the internal structure of the Children Helping Out: Responsibilities, Expectations, and Supports (CHORES), an assessment of household task participation for children. Rasch analysis was used to examine patterns of item response and scale structure with data collected from caregivers of 132 children and youth ages 6–14 yr with and without disabling conditions. Internal consistency was strong for the total measure and the subscales. The items in both subscales fit the measurement model, and the item difficulty order matched the expected pattern from harder to easier household task performance and degree of caregiver assistance. The sample distribution in the hierarchical continuum showed that younger participants and those with physical disabilities tended to score lower. Some inconsistencies in rating scale use suggest a need for further clarification of the scoring criteria for measurement coherence.
Keywords
Participation is a key predictor of quality of life and preparation for independent living (Kanne et al., 2011; Law et al., 2006). Household task participation warrants assessment because daily living activities are predictive of community living and greater independence for people of all ability levels (Kanne et al., 2011; Smith, Maenner, & Seltzer, 2012). Chores are common tasks that take place frequently in all homes, and participation affords a way to examine skills and social participation (Goodnow, 1996; Paradise & Rogoff, 2009). In addition, parents have reported that participation in domestic tasks is an important activity for their children (Adolfsson, Malmqvist, Pless, & Granuld, 2011).
Engagement in household tasks helps prepare children for future living in the community; such preparation occurs largely within the home (Anderson-Moore, Chalk, Scarpa, & Vandivere, 2002; Paradise & Rogoff, 2009; Reid et al., 2011). Thus, active involvement in domestic routines fosters children’s learning of household tasks (Goodnow, 1996; Larson, 2004; Rogoff, 2003). The learning may be tacit, such as by observing others perform household tasks, or more explicit, such as with direct instruction and supervision.
Caregivers expect children to assume more responsibility for household tasks as they age (Hofferth & Sandberg, 2001; Rogoff, 2003). Changes in responsibility are evident in the transfer of accountability for task completion from the caregiver to the child (Rogoff, 2003). Documenting these changes leading to children’s independence helps families adjust expectations and identify strategies to involve and engage children in daily tasks. In addition, exploration of children’s engagement in these tasks helps increase occupational therapy practitioners’ understanding of how to prepare children of all abilities for future independent or interdependent living.
Much variation occurs among families with regard to the household tasks that children perform because of cultural values, the children’s ages, the children’s abilities, and available time (Dunn, Coster, Orsmond, & Cohn, 2009; Goodnow, 2006; Hofferth & Sandberg, 2001; Larson, 2004). Many caregivers value household task participation as a way to help children learn to manage their future home; others (especially mothers) value the development of responsibility and the sense of being a part of the family (Anderson-Moore et al., 2002; Dunn et al., 2009; Reid et al., 2011).
Adaptive behavior measures, such as the Vineland Adaptive Behavior Scale (VABS; Sparrow, Cicchetti, & Balla, 2005), are considered the gold standard to assess household task performance. The VABS is a normative measure, however; it does not provide continuity in measuring the same tasks over time, nor does it provide a way to assess progress or change in assistance needed. A normative measure also may not be appropriate to document changes in the performance of tasks that are culturally laden. Families differ in expectations and satisfaction and in their basis for distributing household responsibilities, so a measure to target this construct needs to include tasks that are culturally sensitive and represent a range of complexity. Measurement of the tasks children perform and the amount of assistance they require or typically receive provides a way to examine independence in home management. In addition, engagement in everyday tasks supports the social structuring of the family environment (Anderson-Moore et al., 2002; Dunn et al., 2009; Goodnow, 2006; Rogoff, 2003).
Assessment of responsibility for household tasks provides a means to examine independence and its changes over time. Household tasks are routines that occur frequently and are common in all homes; the learning of and active involvement in such tasks constitute a way to measure acquisition of independence among children and youth of various ability levels within and across families. Dunn (2004) developed the Children Helping Out: Responsibilities, Expectations, and Support (CHORES) measure to examine participation in the home in common household tasks. The CHORES assesses domestic task engagement with two subscales: Self-Care and Family Care. Self-Care subscale items examine involvement of the child or youth in organizing and caring for his or her belongings in his or her own space, whereas Family Care subscale items examine the involvement of the child or youth in looking after the needs and belongings of family members in common spaces in the home.
Preliminary analysis of the psychometric properties of the CHORES (Dunn, 2004) showed good test–retest reliability (intraclass correlation coefficient [ICC] = .93) and strong internal consistency (r = .94). Occupational therapy practitioners and parents have supported the content validity of the CHORES (Dunn, 2004). The CHORES has differentiated performance and assistance levels for children with cognitive and behavioral disabilities such as attention deficit hyperactivity disorder (ADHD; Dunn et al., 2009), physical disabilities (Dunn & Gardner, 2013), and autism spectrum disorder (Dunn & Gardner, 2013). Further examination of the internal structure and validity of the two subscales, Self-Care and Family Care, would strengthen the psychometric properties of the CHORES. Thus, the purpose of this study was to examine the internal structure and dimensions of the CHORES using Rasch analysis, including principal-components analysis.
Method
Research Design
A cross-sectional design with a convenience sample was used for this mailed survey study. Louise Dunn received ethics approval from internal review boards at Boston University and the University of Utah.
Participants
The first author (Dunn) collected data from the caregivers of 132 children and youth ages 6–14 yr with and without disabilities from different regions of the United States. The children formed three groups: (1) those developing typically, (2) those with cognitive or behavioral disorders (ADHD, Asperger syndrome, or high-functioning autism), and (3) those with physical disabilities (cerebral palsy, spina bifida, orthopedic impairments, amputation). Caregivers were mailed a packet of questionnaires to complete on their own and a stamped, addressed return envelope to return the questionnaires.
Participants were recruited through multiple strategies, including ads posted on email lists, at recreation centers, and in specialty pediatric clinics and through colleagues and occupational therapy students. To be included in the study, the targeted child had to have caregiver-reported intellect in the normal range, and the family had to include between 2 and 4 children. Exclusion criteria included caregiver report of the child’s intellectual disability, seizures, sensory impairments (e.g., blind, deaf), or inability to use the upper extremities and presence of any sibling who was not able to participate in activities with same-age peers.
During recruitment, the first author sought participants who had children in the three groups. Group members were matched by child’s age, gender, and ethnicity. The sample was further divided into two age groups—younger children ages 6–10 yr and older children ages 11–14 yr.
Instrument
The CHORES is a 34-item questionnaire designed to assess participation in household tasks (Performance scale) and help needed to do household tasks (Assistance scale). The Performance scale has a dichotomous yes–no response format. Total performance is calculated by summing all the yes responses. Higher performance scores indicate greater extent of age-expected participation. The response format of the Assistance scale is a 7-point Likert scale with 0 = child is not expected to perform task, 1 = child cannot perform task, 2 = child performs task with a lot of help, 3 = child performs task with some help, 4 = child performs task with supervision, 5 = child performs task given verbal cue, and 6 = child performs task on own initiative more than 50% of the time. CHORES Assistance scores are calculated by summing all applicable item responses and dividing this number by the total possible score from all applicable items; the result is multiplied by 100 to conform to a 100-point scale. Higher Assistance scores indicate greater independence with household tasks.
In addition, the CHORES has two subscales: Self-Care household tasks, in which item content primarily affects the child (e.g., Cleans up after own play), and Family Care household tasks, in which item content affects other members of the family (e.g., Prepares a cold meal for family). These subscales illustrate the conceptualization that household participation includes family division of labor and that members’ take responsibility on a routine basis for tasks related to self-maintenance and family maintenance (Gager, Sanchez, & Demaris, 2009; Goodnow, 1988; Hofferth & Sandberg, 2001). The CHORES has strong stability (test–retest, ICC = .94) and internal consistency (r = .93; Dunn, 2004). Parents and occupational therapists had 80% agreement on the CHORES items, supporting its content validity (Dunn, n.d.).
Procedure
The first author contacted the participants, obtained consent, and conducted a brief screening. During the screening, she asked participants about their child’s classroom placement, communication abilities, and additional concerns (e.g., seizures) to support their reports of intellectual functioning. Participants completed the questionnaires and mailed them to the first author in the stamped, addressed envelopes provided. Participants received a gift certificate in compensation for their time and assistance. The CHORES subscales were used in multiple studies, and the data were combined to conduct the Rasch analyses of the instrument.
Data Analysis
To investigate the measurement qualities of the CHORES, two different approaches were used: (1) Cronbach’s α was used to test the internal consistency of the subscales, and (2) the Rasch model was applied to conduct a more detailed item analysis to examine the structure of the CHORES subscales. The Rasch model derives from item response theory and has been used in the construction and validation of several rehabilitation- and health-related measures (Bedell, 2009; Haley, Coster, Ludlow, Haltiwanger, & Andrellos, 1992; Russell et al., 2000). Rasch analysis is based on a probabilistic model that converts ordinal scores obtained from rating scales into interval-level measures. The model is based on the principle that only two attributes are needed to determine participants’ response to test items: the ability of the participant and the difficulty of the test item, each expressed as estimates of the underlying latent trait. However, this principle holds only if the items measure a unidimensional construct.
Based on these principles, it is possible to construct a measure, like a ruler, that is defined by items hierarchically organized from easy to difficult. Because participants and items are calibrated along the same interval metric, expressed in logits and distributed in hierarchical order, it is possible to examine the extent to which items are of appropriate difficulty for the sample. The logic of the distribution of indicated item and person hierarchies can be used as evidence of construct validity (Chien & Bond, 2009).
Rasch analysis using the Winsteps program (Version 3.72.0; Linacre, 2013) was used to examine whether CHORES items combine to measure a single unidimensional construct. Analysis using the rating scale model indicated that increments in percentage of response in each score category of the CHORES rating scale varied, suggesting irregularity in the difficulty intervals across items. Therefore, we proceeded with a partial-credit model. CHORES subscales’ unidimensionality was examined using two methods: (1) principal-components analysis, with variance explained by the first dimension >50% and eigenvalue of the unexplained variance of the first residual factor <2, and (2) goodness of fit of items and participants to the Rasch model, as expressed by the residuals derived from the interval-level scaling (<5% of both the infit and outfit mean square [MnSq] values were ≥1.4 with associated t ≥ ±2.0).
Results from a preliminary principal-components analysis helped guide further data analysis. The CHORES items fell into at least two groups or dimensions that resembled the natural division of the measure into Self-Care and Family Care subscales (Dunn, 2004). Because only a couple of items had been misplaced by the loadings, we decided to keep the two original dimensions (subscales) of the CHORES because these dimensions made theoretical sense. Results indicated that although variance explained for each scale was >50%, the eigenvalues were >2.0 (i.e., 2.2 for the Self-Care and 2.9 for the Family Care subscale), implying extra dimensions; however, these extra dimensions did not make either theoretical or clinical sense, and when analyzed separately, the reliability values fell below .80. Thus, we decided to analyze the full subscales to identify misfitting items that were not contributing to defining the measure.
Items that fit the model are those that present a pattern of response that conforms to the expectations of the hierarchical model: Children with higher ability should do more chores and, therefore, score higher on most items, easy or hard, and children with lower ability should score lower on the difficult items. The Winsteps program provides fit statistics that indicate the fitness of the pattern of responses to the model. Items with infit and outfit statistics ≥1.4 (MnSq) and with associated t ≥ ±2.0 (Z standardized distribution value) are considered to misfit and should be flagged for revision and further analysis. If no more than 5% of the items meet the above criteria for misfitting, the subscale is considered valid.
The infit statistics and children pattern of responses to the model, along with analysis of the logic of the hierarchy of item difficulty levels, were used as an indicator of construct validity. Inspection of the value of the mean person measure and of the item–person maps was conducted as a means to examine the relationships among item difficulty, the ability of the children (item targeting), and the spread of the items over the range of ability levels.
The Winsteps program also calculates the reliability for item calibration and corresponding person measure, with values above .80 considered an indication of the stability of the measures. The error for each child measure and item calibration was calculated. The reliability and error values are used to calculate the separation index, or number of ability levels the items divide the sample into, which should be at least two levels (Bond & Fox, 2007). Cronbach’s α was set at >.80 to indicate the subscales’ internal consistency.
The structure of the Assistance 7-point scoring system was also investigated to verify whether the scoring criteria were behaving according to the principles of the Rasch model. The 0 scores were excluded because these scores indicate that the caregivers did not expect their children do the tasks. The response pattern of each scoring category for each item was examined to identify weaknesses and points that needed to be revised or clarified.
Results
Participants
Caregivers of 132 children of all abilities ages 6–14 yr participated in this study. Of the children, 49.2% (n = 65) were in the typically developing group, 33.3% (n = 44) were in the cognitive or behavioral disorders group, and 17.4% (n = 23) were in the physical disabilities group. Approximately 60% (n = 79) of the children were in the younger age group. The numbers of male (49%, n = 64) and female children were approximately equal. Table 1 provides demographic information on the sample and each group’s mean scores on the CHORES.
Demographic and Clinical Characteristics of Caregivers and Children and CHORES Scores
Note. CHORES = Children Helping Out: Responsibilities, Expectations, and Supports; M = mean; POMP = percent of maximum possible; SD = standard deviation.
Internal Consistency
Both the Self-Care and Family Care subscales showed evidence of strong internal consistency (αs = .96 and .98, respectively).
CHORES Self-Care Subscale
Rasch Analysis—Scalability and Dimensionality.
Regarding calibration, all 13 items from the CHORES Self-Care subscale had infit and outfit statistics within acceptable Rasch measurement ranges, and only 6 children (4.5%) misfitted the model. Item and person reliability were .88 and .95, respectively, and the items separated the children into 2.45 ability levels. Table 2 presents item calibration and fit statistics for the CHORES Self-Care subscale.
Item Calibration and Fit Statistics for the Self-Care and Family Care Subscales of the CHORES
Note. CHORES = Children Helping Out: Responsibilities, Expectations, and Supports; MnSq = mean square; SE = standard error; Z = standardized distribution value.
Number (percentage) of children (N = 132) that were scored on the item; scores of 0 (child is not expected to perform task) were deleted from the analysis.
Rasch Analysis—Hierarchical Ordering and Test Targeting.
Item difficulty calibration spanned 1.37 logits. The easiest was Item 17, Puts own laundry in hamper (logit = −0.71), and the most difficult was Item 9, Makes self a hot meal (logit = 0.66). The mean measure fell 1.09 logits above the mean item calibration, indicating that, on average, the CHORES tasks were easy relative to the children’s ability levels. The number of respondents (n) in Table 2 indicates the relevance of each task to the parents; items considered irrelevant (score = 0) by specific parents were dropped from the analysis. Although most parents (98.5%, n = 130) considered Puts own laundry in hamper (Item 17) to be relevant, fewer (60.6%, n = 80) considered Item 23, Dusts own room, to be relevant. Considering the scoring criteria, most children performed the Self-Care items with a verbal cue (score of 5; 33.3%, n = 44) or on their own initiative (score of 6; 37.1%, n = 49); very few were considered unable to do specific tasks (score of 1; 3.0%, n = 4).
Figure 1 shows the task difficulty hierarchy, on the right, and the distribution of children according to ability level on the Self-Care subscale, on the left. The children are organized in a normal-like distribution, dislocated somewhat upward (1.09 logits) in relation to the item distribution.
CHORES Family Care Subscale
Rasch Analysis—Scalability and Dimensionality.
As seen in Table 2, all items on the CHORES Family Care subscale had infit statistics within acceptable limits (MnSq < 1.4), and 17 of 21 items had acceptable outfit statistics values (MnSq < 1.4); however, no item presented values for both infit and outfit outside the criteria for misfitting items. Only 4 children (3.0%) misfit the model. Item and person reliability were .92 and .87, respectively, and the items separated the children into 2.57 levels of ability.
Rasch Analysis—Hierarchical Ordering and Test Targeting.
Item difficulty calibration spanned 2.06 logits and ranged from −1.01 (easiest: Item 27, Feeds pet) to −0.99 (most difficult: Item 21, Runs washer/dryer). The mean measure was 0.54 logits above the mean item calibration, and as shown in Figure 2, the Family Care items were more targeted to the distribution of the children’s ability to do family care tasks than the Self-Care items. However, more Family Care items were considered beyond expectations; for example, Item 20, Puts laundry away for family, was scored by only 27.3% (n = 36) of the parents. The scoring system showed the same characteristics as the Self-Care subscale, with the score of 1 (Child cannot perform task) being used very rarely (6.1%, n = 8) and the score of 5 (Child performs task given verbal cue) being the most frequent score (50.8%, n = 67).
Discussion
This study expands on the earlier examination of test stability, internal consistency, and content validity of the CHORES, a measure of household task participation for children of all abilities ages 6–14 yr (Dunn, 2004). In this study, we used Rasch analysis to examine the internal structure of the CHORES. The CHORES is a unique tool; it is one of the few measures that examines the participation of children in common home activities, including social engagement as a family member, which may be predictive of future independent living.
Findings from the Rasch analysis substantiate the unidimensionality of the Self-Care and Family-Care subscales thus supporting the use of summed subscale scores rather than a total score when measuring children’s assistance with household tasks. In addition, these findings support the internal consistency of the CHORES’ Assistance ratings, with a larger sample.
The sample distribution in the hierarchical continuum suggested a good fit of children by ability and age; younger participants and those with physical disabilities tended to score lower. However, the distribution of items and children along the hierarchical continuum was distinct for the two subscales. Although the Self-Care subscale presented a ceiling effect, with 2 children from the older group receiving maximum scores, items from the Family Care subscale were more difficult, with maximum item difficulty exceeding the ability level of the participants (Figures 1 and 2). The values of the mean measure for the children in each subscale were consistent with this pattern; children scored 1.09 logits above the mean item calibration on the Self-Care subscale and 0.54 logits on the Family Care subscale, indicating that the Self-Care subscale items, on average, were easier for the sample.
Although the Rasch model is structured to deal with missing data in the analysis, it is noteworthy that parents scored more tasks on the Family Care subscale as 2 (Child is not expected to perform task) than on the Self-Care subscale. As shown in Table 2, more parents responded to the Self-Care items (ranging from 63.7% to 99.3%) than to the Family Care items (ranging from 27.3% to 96.3%). These results reflect what we have observed clinically: Most parents require children to do self-care chores, but assignment of family care chores varies, is more flexible, and depends on cultural values and expectations (Goodnow, 1996, 2006).
Hierarchies of tasks fit the expected pattern of a continuum of difficulty for household tasks. Most of the parents reported that their children performed easier tasks, such as Cleans up after own play, many independently or with a verbal cue. Fewer children performed the more difficult tasks, such as Prepares part of a hot meal for family, and often with assistance. The availability of a difficulty hierarchy for household tasks is helpful to guide clinical reasoning and to inform caregivers about the degree of involvement in household tasks to expect of children.
The rating scale presents some concerns with respect to a contracted degree of difficulty for both CHORES Self-Care subscale and Family-care subscale items. Specifically, the rating scale includes a combination of two criteria—that is, two ratings to identify why the child does not do a specific task and five ratings to identify the type and amount of assistance frequently made available in the family daily routine. To examine the unidimensionality of the CHORES, these criteria were combined into one rating scale for the scoring of each item. Although the presence of two criteria enriches the items scoring, it may pose confusion to parents who fill out the questionnaire. Clarification of each scoring criteria and rating, with a brief explanatory sentence may help parents make more precise choices, especially considering that this instrument could be completed online. For example, performing task with a verbal prompt (rating 5) and performing task with supervision (rating 4) might be clearer if presence of caregiver during the task was defined. Here, presence of family member throughout the child’s performance of the task would describe with supervision and performance of the task without family member would be the rating for the verbal prompt. Bourke-Taylor, Law, Howie, and Pallant (2009) provided assistance ratings for leisure activities that would be a helpful resource for refining the rating criteria for the CHORES.
Comparing the response pattern on the two scales, 51% percent of the scores on the Family-Care subscale were the given verbal cue rating, whereas only 33% of the Self-Care subscale items received this score. Such a distinction on the same rating across subscales evidences the greater complexity of skills needed to initiate and complete Family Care tasks. In addition, caregivers reported that they did not expect their children to do the Family Care tasks more often than Self-Care tasks, suggesting that families of this cultural background may value chores on these two subscales distinctly. The extent to which culture influences children and adolescent engagement in self-care and family care tasks remains to be investigated.
Parents of children of all abilities have reported that participation in home activities, including household tasks, was important to them (Adolfsson et al., 2011; Reid et al., 2011). Parents of children with special needs have reported that participation in household tasks was important but often was not addressed by current service delivery models (Bedell, Khetani, Cousins, Coster, & Law, 2011; Law et al., 2013). Household task participation warrants attention from occupational therapy practitioners because parents perceive that it contributes to their family routines, is a way to help their children develop responsibility, and prepares their children for the future (Dunn et al., 2009; Dunn & Gardner, 2013; Reid et al., 2011).
These concerns support the utility of a valid and reliable measure of household task participation such as the CHORES. The psychometric properties of the CHORES are adequate to make it useful in measuring children’s participation in household tasks. Incorporation of the CHORES into the assessment protocol offers a springboard to promote discussion with caregivers and their children about engaging in family routines and preparing for independent living. Addressing concerns by analyzing task difficulty and environmental supports and barriers to participation can inform practice and help practitioners and families prepare children of all abilities for future independent living.
Limitations and Future Research
The study participants were from a relatively homogeneous population, mostly White and from the United States. Initial data collection did not include representation of children with intellectual disabilities. We used a mail survey to gather information and interviews with some participants to confirm their understanding of the ratings. Development of an electronic version of the CHORES warrants consideration, given the ease of access and completion that technology affords caregivers. A CHORES website could also offer opportunities for providing links to resources and a broader discussion network for supporting and engaging children of all abilities in household tasks. Studies with larger numbers of children of varying abilities would further support the construct and discriminative validity of the CHORES.
Implications for Occupational Therapy Practice
The findings of this study have the following implications for occupational therapy practice:
The CHORES provides practitioners with a tool that is psychometrically sound, easy to administer, relatively brief, and clinically relevant to measure children’s participation in and degree of assistance needed to participate in household tasks.
The CHORES measures a valued aspect of home participation that contributes to children’s roles in the family and to their preparation for independent living.
The CHORES provides a way to measure some aspects of preparation for adult roles for children.
The CHORES provides information about relevant household tasks with a range of complexity.
The internal structure of the CHORES discriminates household task participation among children by ability and age.
Footnotes
Acknowledgments
We appreciate the participation and contributions of the caregivers to our study. We also appreciate the review by Tina McNulty and recruitment assistance from practitioners, community agencies, and graduate students.