Chapter 4 Methodology
4.3 Research Strategy and Design
4.3.2 Instruments
The quantitative questionnaire (see Appendix B) was divided into three parts, the first being self-report demographic variables, including gender, age, ethnicity/cultural background and SES. The second asked questions pertaining to the students’ health and lifestyle, including body image and self-perception questions. The third section assessed respondent’s worries and/or concerns in environments such as school or the home. Questions were based on the research literature in the area of mental health and wellbeing programmes, adolescents’ health, including adolescent anxiety and depression, and protective factors such as resilience and self-esteem (Abraham, O’Dea & Kefford, 1999; Allen & Rapee, 2009; Allen, Rapee & Sandberg, 2012;
Lyneham et al., 2011; O’Dea, 1998, 1999, 2009; O’Dea & Wilson, 2006; Stunkard, Sorensen & Schlusinger, 1983; Van Hoorn, Kefford, O’Dea, Richardson & Abraham, 1999). The questionnaire had been previously pilot tested and administrated to children and adolescents aged 6 to 18 years in two national school-base studies (O’Dea & Dibley, 2010; O’Dea & Wilson, 2006). Internal consistency reliability was assessed, by Cronbach’s alpha, for each outcome measure (Waltz, Strickland & Lenz,
2005). For internal reliability, a Cronbach’s alpha of approximately .70 and above is shown to be acceptable (Carmines & Zeller, 1979; Field, 2009). Validity and reliability has been reported as follows for individual sections of the questionnaire where available.
4.3.2.1 Measurement of demographic variables
Information regarding age, gender, and family cultural/ethnic background was provided by means of a questionnaire with closed respond questions with categorical options.
4.3.2.2 Measurement of SES and ethnicity/cultural background
Information regarding age, gender, and family background was provided by means of a questionnaire with closed questions. The participants’ ethnicity was self-reported and included categories of Caucasian/Northern European (52.8% (n=84), 65.0%
(n=26)), Southern European (23.9% (n=38), 30.0% (n=12)), Asian (3.1% (n=5), 2.5% (n=1)), Middle Eastern (8.8% (n=14), 2.5% (n=1)) and Other (11.3% (n=18), 2.5% (n=1)) for both intervention and control groups respectively.
The SES of the schools was categorised as low, middle or high SES according to parental income and education level, as well as confirmation by the principal of each school’s SES according to the majority of students attending.
4.3.2.3 Measurement of self-perception
The Physical Self-Esteem Score (Abraham et al., 1999), also known as the Body Appearance Rating (Van Hoorn et al., 1999), was used to assess perception of physical appearance using a self-perception rating score from zero to 10 (10 being
‘perfect’) for how participants rate their own body appearance (self) and their perceptions of how ‘other people’, ‘people of the opposite sex’, ‘mother’ and ‘father’
would rate them. This component of the instrument has been successfully validated (O’Dea, 2009) against several scales on the Eating Disorder Inventory, including the Body Dissatisfaction and Drive for Thinness Scales (Garner, Olmstead & Polivy, 1983) with significant (p<.001) negative Spearman correlation coefficients of between .55 and .67. In addition, adolescents have been found to be able to accurately report
their anxiety (Brooks & Kutcher, 2003), depression and self-esteem using this basic 10-point scale (O’Dea, 2009; O’Dea & Abraham, 2000). Such self-reports have also been verified by clinicians as being reasonably accurate in identifying these issues (Rey et al., 2002; Timbremont, Braet & Roelofs, 2008). The self-perception subscale of the questionnaire appeared to have high internal reliability, α=.85. All items appeared to be worthy of retention, and correlated with the total scale to a good degree.
While there is a surfeit of self-report measures on anxiety and depression available for adolescents, these instruments need to be low cost and easily accessible, so that the school will continue to use them, while still providing adequate psychometric properties of reliability and validity (Campbell, 2004). Using this instrument would also enable schools and health workers to similarly and simply initiate early identification and intervention systems in the school support network, and efficiently and appropriately refer adolescents (O’Dea, 2009).
4.3.2.4 Measurement of body image
In the current study, body image was measured using two major categorical Likert items: ‘Do you think you are too thin (1), about right (2) or too fat (3)?’ and desired body weight was measured by asking, ‘Would you like your body weight to be: A lot heavier (1); a little heavier (2); same as at present (3); a little lighter (4); a lot lighter (5)’. Desired weight was measured by asking yes/no questions: ‘Have you ever tried to lose weight?’ and ‘Are you currently trying to lose weight?’ (O’Dea, 1998, 1999).
The body image subscale of the questionnaire appeared to have good internal consistency, α=.73.
Figural drawing scales are often employed to measure aspects of body disturbance, dissatisfaction and body ideals (Fingeret, Gleaves & Pearson, 2004; Gardner &
Brown, 2010; Hill, 2011). This method involves the presentation of body silhouettes ranging from very thin (score 1) to very large (score 9) (Fingeret et al., 2004), with participants being asked to select figures in response to a variety of questions. A higher score represents a greater desired body size (O’Dea, 1999) (rating of 1–9, 1=emaciated, 9=obese). Research has noted several limitations to using such instruments, including the fact that the number of available figures can be misleading
as participants usually only select from a small portion of the available figures (Gardner & Brown, 2010), and participants may be influenced by the sequential nature of the figures and questions (Fingeret et al., 2004). Nevertheless, Gardner and Brown (2010) report the test-retest reliability of the Stunkard Figure Rating Scale (Stunkard et al., 1983) for use in adolescents for current size (c=0.87) and ideal size (I=0.83), and the concurrent validity (CO=0.61–0.75). In the current study, the Stunkard Figure rating scale (Stunkard et al., 1983) was used to measure body size ideals, specifically ‘current self’, ‘ideal self’, ‘ideal female’ and ‘ideal male’. The Stunkard Score subscale of the questionnaire had a high internal reliability, α=.81.
4.3.2.5 Measurement of BMI
Participants’ height and weight were measured using standardised anthropometric procedures. The measurements were taken using the protocol of International Society for the Advancement Kinanthropometry (ISAK) (Marfell-Jones, Olds, Stewart &
Carter, 2006). Height was measured to the nearest 0.5 cm using a portable stadiometer. Weight was measured to the nearest 0.1kg using portable Soenle digital scales with a range of 0–200 kg. Body Max Index (BMI) was calculated from the students’ height and weight (Keys, Fidanza, Karvonen, Kimura & Taylor, 1972). BMI ranges were calculated using the Children and Adolescent Syntax in SPSS recommended and defined by the International Obesity Taskforce (IOTF) (Cole, Bellizzi, Flegal & Dietz, 2000; Cole, Flegal, Nicholls & Jackson, 2007). The BMI of adolescents in the current study ranges from 12 to 40, in keeping with the normal range findings from other studies (O’Dea & Wilson, 2006) and indicate normal growth over the 12 months (de Leonibus et al., 2013).
4.3.2.6 Measurement of students worries/stresses
In the current study, the Child and Adolescent Survey of Experiences (CASE-C/P) was used to assess student worries as it provides a measure of stressful life experiences of relevance to young people (Allen & Rapee, 2009; Allen et al., 2012).
Items for the CASE were designed to parallel a standard clinical interview measure of life events, the Psychosocial Assessment of Child Experiences (PACE) (see Sandberg et al., 1993). It consisted originally of 40 items- stressors/experiences, but several items were excluded from the scale due to their confusing meaning and their
irrelevance to the school context. The Children’s Anxiety Life Interference Scale (CALIS-C) is designed to assess life interference attributed to fears and worries from child and parent perspectives (Castaneda, McCandless & Palermo, 1956). The CALIS-C is a 10-item self-report questionnaire about the impact of fears and worries on an adolescent’s quality of life, self-efficacy and wellbeing (Lyneham et al., 2011).
These instruments were found to be reliable and valid (Allen et al., 2012; Lyneham et al., 2011).
The CALIS-C also demonstrates sensitivity to change (Lyneham et al., 2011). Items for the current research survey were chosen and modified in collaboration with the school principal and the school counsellor as a means to assess student stress of school-aged children from a number of sources. Students’ worries and/or stresses were primarily assessed using a three-point Likert scale. For each of the questions, respondents indicated their level of worry or concern each one evoked (ranging from 1 ‘worry a lot’ to 3 ‘do not worry’). Participants were also asked to list their top three stressors/worries and to report any of the strategies they use to help reduce anxiety and worry using an open-response design. Cronbach’s alphas for the nine relationship and seven social interaction items were .87 and .85 respectively. The school worries subscale consisted of five items (α=.76), and the self-confidence worries subscale consisted of five items (α=.76). The worries and stress inventory was found to be highly reliable (26 items, α=.92) in the current data.
4.3.2.7 Measurement of high-risk intervention students
The high-risk intervention group were assessed at baseline, post-intervention and at the 16-month follow-up the same measures as described above. As a result of these findings they also received ongoing assessment, support, and referral that was elicited by the school counsellor.