![[ Visit Client Website ]](images/banner.gif)
Purpose: The purpose of this study was to assess the fidelity and cultural appropriateness of the Child Feeding Questionnaire (CFQ) by Birch et al. (2001) and the Child Eating Behavior Questionnaire (CEBQ) by Wardle et al. (2001) in Saudi Arabia. The CFQ is a 31-item self-report Likert questionnaire that measures maternal feeding styles and it’s association with child’s adiposity. The CFQ consists of seven factors including perceived parent weight, perceived responsibility, perceived child weight, concern about child weight, restriction, pressure to eat and monitoring (Birch et al., 2001). Higher scores indicate higher controlling maternal feeding practices yielding to increased risk of developing obesity (Birch et al., 2001). Response options varied according to the measured factor and these included, never to always; markedly underweight to markedly overweight; unconcerned to very concerned and disagree to agree.
While the CEBQ is a 35-item self-report Likert questionnaire that measures mother’s assessment of their child’s eating behaviors. The CEBQ consists of eight dimensions including food responsiveness, fussiness, emotional eating (over and under), enjoyment of food, satiety responsiveness, slowness in eating and desire to drink (Wardle, Guthrie, Sanderson, & Rapoport, 2001). Higher scores indicate a higher rate of that particular eating behavior construct (e.g. food responsiveness, emotional overeating, etc.); responses to the CEBQ were four options and these included, never, rarely, sometimes, often and always (Haycraft, Farrow, Meyer, Powell, & Blissett, 2011; Wardle et al., 2001).
Methods:
The procedures taken for this study undertook six steps and these include, translation, synthesis, back-translation, expert committee review, cognitive interviewing and pilot testing.
Results: The time to complete the six procedures was approximately four months. Results have demonstrated the following for each procedure:
I. Translationof the CFQ and CEBQ was completed by two Arabic-speaking translators; they were recruited from the Saudi community producing two versions of each questionnaire. During this phase both translators were informed about the study aims, research questions and studied concepts to ensure accuracy of translation. One translator was from Saudi Arabia and had years of experience in translation and was a PhD faculty in academics. The second translator was from Jordan and was also a PhD prepared researcher that had years of experience translating questionnaires. One translation was received via e-mail communications and the other translation was received via face-to-face. Both versions were compared for the next phase.
II. Synthesiswas conducted by the Principal Investigator (PI), where she integrated the two Arabic translated versions of the CFQ to one version and similarly one version from the two Arabic translated versions of the CEBQ. The PI is fluent in both Arabic and English as she has lived half of her life in an Arabic-speaking country and the second half of her life in an English-speaking country. The technique used during this phase considered the five equivalences recommended by Streiner and Norman (2008), which include, conceptual, item, semantic and operational equivalences. The final equivalence, measurement equivalence, will be tested later because it requires a larger sample to measure factor analysis.
III. Back-translation, where two English-speaking translators were recruited from Elance to back-translate the synthesized version of the CFQ and CEBQ from Arabic back to English, then the English-back-translated versions were sent to the PI’s dissertation chairperson to compare it with the original CFQ and CEBQ English version. During this phase both back-translators were not informed about the purpose of the study, research questions and studied concepts to avoid any bias in translation (Guillemin, Bombardier, & Beaton, 1993). The English-back-translated versions of CFQ and CEBQ were found to be closely similar to the original English version of the CFQ and CEBQ as judged by the PI’s dissertation chairperson.
IV. Expert committee review(n=3) were researchers in or from Saudi Arabia studying childhood obesity, recruited via e-mail communications. The researchers were asked to rate each item from the Arabic version of the CFQ and the CEBQ on a 4-point Likert scale (highly relevant, quite relevant, somewhat relevant, not relevant) to the studied concepts of the PI’s aims and research questions. Data were collected via Qualtrics, Provo, UT (2015), version 1.107s. Agreement scores between the expert committee were calculated to assess content validity index (CVI) for both the CFQ and CEBQ. An adequate Scale-Content Validity Index (S-CVI) score for the CFQ and CEBQ was found consecutively (0.99 and 0.92). However, the Item-Content Validity Index (I-CVI) was not adequate because not all the items were a perfect one. In the CFQ, one item was 0.67. In the CEBQ, several items were 0.67. Since, the S-CVI showed to have an adequate score with what the literature supports (Lynn, 1986; Polit, Beek, & Owen, 2007). The PI continued to the cognitive interviewing phase and focused on the items that did not show a perfect one.
V. Cognitive interviews(n=5) were conducted using a convenience sample from the Saudi community to assess understanding and cultural appropriateness of the Arabic versions of the CFQ and CEBQ and to detect problems before the questionnaire was pilot tested and administered (Willis, 2015). Each interview was individually conducted in the mother’s home, audio-taped, transcribed for analysis. The PI used the think-aloud technique to detect any misunderstandings and recommended analysis method described in the literature (Abubakar, Dimitrova, Adams, Jordanov, & Stefenel, 2013; Knafl et al., 2007). A few items were modified after all the interviews were completed. Issues such as, word choices and vague sentences helped the PI in modifying the Arabic version.
VI. Pilot test (n=10) to estimate the test-retest reliability and to calculate the time spent to fill out the questionnaire. Participants included a convenience sample from the Saudi community and were communicated through Qualtrics e-mail server. Seven to ten days was the gap between the first and second administration of the questionnaires. Data for this phase were collected through Qualtrics, Provo, UT (2015), version 1.107s. Test-retest reliability estimates showed acceptable alpha co-efficient. For the CEBQ alpha coefficients were acceptable for all the concepts (α= 0.86 to 0.94). For the CFQ alpha ranged from α= 0.61 to 0.90. The lowest alpha coefficient (α= 0.61) was the concept ‘concern about child weight’, but was kept in the questionnaire after consultation with the PI’s chairperson and an expert in psychometrics and instrument development. Since this was a small pilot test, further testing was recommended on a larger sample. The time spent time to fill out the questionnaire ranged from 13 to 60 minutes.
Conclusion:
The Arabic version of the CFQ and CEBQ has shown acceptable agreement scores as assessed by a group of Saudi researchers on the topic of childhood obesity and calculations showed acceptable scale-content validity index scores. Moreover, acceptable internal consistency has been demonstrated through pilot testing for all the CEBQ concepts and all except one concept in the CFQ, which was kept in the questionnaire as recommended by a psychometrics expert. This study has demonstrated preliminary and acceptable findings of the Arabic versions of the CFQ and CEBQ. Moreover, this study is part of a mixed methods study on childhood feeding and growth in Saudi Arabia. The following phase will be a quantitative phase (QN) studying the association between child’s adiposity and how it correlates with maternal feeding styles, child’s eating behavior and family socioeconomic factors in Saudi Arabia. After that, a qualitative (QL) phase will follow by interviewing selected mothers from the QN phase to explore the concepts within the Saudi Arabian culture. Finally the mixed phase will explore how the QL findings further explain the QN results.