5. Análisis de resultados
5.3. FASE 3: La voz de los expertos
i. Overview of measurement and variable derivation
HFA was assessed using a 17-item questionnaire about the frequency of availability (never, rarely, sometimes, mostly, always) of fruits, vegetables, savory snacks, sweets, milk, and sugar- and artificially-sweetened beverages within the home over the previous 30 days (180). The questionnaire was adapted from the Active Where? parent-child survey (35,173). Only items identified as obesogenic in Table 3.10 were used in our analyses. The divisions were used in previous implementations of this questionnaire (33,35) and are based on nutrient values and caloric density. HFA of obesogenic items was defined as the sum of the Likert rating of frequencies of these nine obesogenic items.
ii. Questionnaire development and validation
Information on the questionnaire’s development is not available. Validity and reliability of the questionnaire were assessed in three multiethnic samples: 171 adolescents aged 12-18 years (10% Hispanic) and 116 children aged 5-11 years (8% Hispanic) in a study by Ding et al. (35), and 699 children aged 6-11 years (17% Hispanic) in a study by Couch et al. (33). The questionnaire showed moderate to good consistency for obesogenic items (184) in both children and adolescents based on Cronbach’s alpha coefficient (33,35). The questionnaire also showed moderate to substantial test-retest reliability (27 days between HFA questionnaire
administrations) according to ICC values (157) (Table 3.11.).
Construct validity for HFA questionnaires can be determined based on whether
availability of foods/beverages in the home is associated with dietary intake of the corresponding items (34,207). Construct validity was assessed in both studies using different measures of
dietary intake. Ding et al. assessed fruit and vegetable intake with two questions regarding the number of servings of fruits and vegetables the child or adolescent ate in a typical day (ranging from 0-≥4 servings/day) (35). Couch et al. assessed dietary intake using three 24-hour dietary recalls, with the parent and child jointly reporting the diet if the child was <8 years of age, and the child reporting their diet with parental assistance if he/she was ≥8 years of age (33). Comparisons of HFA and dietary intake in both studies showed good construct validity of this HFA measure in children and adolescents. Specifically, significant associations between HFA and dietary intake were generally in the expected directions (Table 3.12.).
Although construct validity as indicated by both the correlations and beta coefficients could be stronger, other HFA tools with similar validity (208) have also been able to identify significant associations between HFA and dietary intake (33,36,158–160). Given the high reliability and consistency of this tool and its ability to detect associations with dietary intake, it was deemed appropriate for the purpose of assessing whether HFA is a potential modifier of the association between food parenting practices and dietary intake in SOL Youth.
b. Parenting style
i. Overview of measurement and variable derivation
Parenting style was assessed using the demandingness (7-item) and responsiveness (9- item) scales from the 16-item Authoritative Parenting Index (180,209). The Authoritative Parenting Index was reworded for SOL Youth for parents. Though parenting styles are typically studied by classifying individuals into the four parenting styles proposed by Maccoby and Martin (i.e., authoritative, authoritarian, permissive, or uninvolved) (210), we examined the scales separately using continuous scores for each scale as explained in Chapter 6.
ii. Questionnaire development
The Authoritative Parenting Index was originally developed for a study of alcohol and tobacco use in 1236 4th and 6th grade students (83% White) in central North Carolina. The Authoritative Parenting Index was designed to assess the two dimensions of authoritative parenting: demandingness and responsiveness (209). The initial version of the Authoritative Parenting Index included twenty items developed based on qualitative research of authoritative parenting (209,211–216). Items were pilot-tested in 3rd and 4th grade students to ensure that younger children could comprehend the Authoritative Parenting Index (209). Each item provides a statement to which individuals respond whether it sounds just like, a lot like, sort of like, or not like their mother/step-mother (209). To check that the index assessed demandingness and
responsiveness, researchers conducted exploratory factor analyses in the aforementioned sample and in a sample of 1490 9th and 10th grade students (80% White) (209). Both factor analyses showed that the Authoritative Parenting Index measured a third factor representative of
permissive parenting, but this factor had an eigenvalue <1.0 (209). Thus the three items loading on this factor in addition to another item with a low cross-loading were dropped from the
Authoritative Parenting Index, with the final version and corresponding factor loadings shown in Table 3.13. (209).
iii. Questionnaire validity and reliability
Internal consistency for each dimension of parenting was assessed in the aforementioned study samples, stratified by race/ethnicity and sex, using Cronbach’s alpha coefficient.
Coefficients for the responsiveness and demandingness subscale ranged from 0.71 to 0.90 and 0.65 to 0.83, respectively, indicating good internal consistency (Table 3.14.) (184,209). Although there were no differences by sex, coefficients tended to be lower among Blacks versus Whites.
Construct validity was assessed in both aforementioned samples and in 224 7th and 8th grade students (75% White) (209). Because the association of the Authoritative Parenting Index with measures of dietary intake has not been examined, validity was based on whether
associations of authoritative parenting style were in the hypothesized direction with indicators of social and academic competence (e.g., self-esteem) and child risk behaviors (e.g., tobacco use) (209). Responses to the two subscales (demandingness and responsiveness) were divided into tertiles, with individuals falling into the high/low tertiles being assigned to one of four styles: authoritarian (high demandingness/low responsiveness), authoritative (high demandingness/high responsiveness), neglectful (i.e., uninvolved; low demandingness/low responsiveness), and indulgent (i.e., permissive; low demandingness/high responsiveness) (209). The Authoritative Parenting Index varied as predicted with indicators of social and academic competence and risk behaviors. Adolescents whose parents were defined as authoritative had significantly higher self- esteem, self-control, school adjustment, peer acceptance, resistance to peer influence, and conflict resolution ability; significantly lower anger proneness and alienation; and significantly lower odds of reporting substance use and violence-related behaviors than peers whose parents were defined as neglectful (209). Comparisons with parents defined as indulgent or authoritarian were also consistent with prior research providing additional support for the Authoritative Parenting Index’s construct validity (209). Based on the high consistency and validity of the Authoritative Parenting Index, the tool is considered an appropriate measure of parenting styles for the purpose of assessing the potential modifying role of parenting styles in the association between food parenting practices and dietary intake.