Baseline Methodology and Descriptive Statistics
4.1 Methodology
4.2.9 Dieting and Weight Control
All participants were asked, ‘Have you ever done anything to try to control your weight (to get thinner or stop yourself getting fatter)? Those who said that they had been involved in weight control in the past were questioned further about the specific activities that they had tried. Those who responded that they had not were asked to skip the remaining questions in this section.
4.2.9.1 Specific weight control activities
In order to assess weight control activities, a new checklist was developed, comprising 17 items examining lifetime participation in a range o f dietary and non-dietary weight control behaviours. Items related to intake of foods from specific nutritional groups, meal and eating patterns, lifestyle changes, purging, and participation in formal weight loss programmes either through a GP or dietician or independently. The items were generated with reference to those used by other researchers (e.g Neumark Sztainer et al 1999; French Perry Leon and Fulkerson 1995; Middleman et al 1998), and modified in response to comments by pilot groups of women and adolescent girls. Participants were asked whether they had ever used each method specifically to try to control their weight
4.2.9.2 Dieting Intensitv Rating
Weight controlling participants were asked to provide a numerical estimate of how many times they had ‘been on a diet’ in the past year, referred to here as the ‘frequency question’. They were also asked which o f a number o f time periods represented the length of time their diets had usually lasted during the past year from "less than one day’, ‘‘between one day and one w eek\ ‘‘between one week and one month' and ‘more than one month', referred to as the ‘duration question’. These two questions were combined to produce a ‘dieting intensity’ variable, which provided a rough estimate of the number o f days in the past year that each individual had been dieting. In order to do this an estimated number of
Chapter 4. Baseline methodology and descriptive statistics
days relating to each of the dieting periods from the duration question was multiplied by the number of dieting episodes reported in the frequency question.
Deriving estimated number o f 'dieting days per d iet’ fo r each response to the diet duration question. Participants who claimed that their diets usually lasted for less than one day were given a multiplier of one as a rough estimate of the number of days on which they were dieting per dieting episode. For those who claimed that their diets had lasted between a day and a week , the span of this time period (six days) was divided by 2 in order to obtain the mid-point of the time period. This resulted in an multiplier of three, representing three dieting days per episode. A similar procedure was carried out for those who indicated that their diets lasted for more than a week but less than a month. The range of the period between a week and a month is equivalent to 23.4 days (taking into account the differing number of days in different months). This number was divided by 2 in order to obtain the mid-point of the time period and the resulting number (11.7) rounded up to an estimated 12 dieting days per episode. The highest duration category (more than a month) presented a problem as it had no upper limit to allow a calculation of the mid-point o f the time period. Although the midpoint between one and 12 months could have been used, it was felt that this would weight the resulting scale too far towards dieting duration at the expense of dieting frequency. On the other hand some of the dieters who diet for more than a month are likely to be dieting for considerable periods o f time since a study of British adolescents in the North-west o f England reported that 6% of dieters were dieting ‘most of the time’ (Roberts et al 1999). An estimated dieting period of two months (60 days) was used as the estimated dieting duration of those who reported that their diets lasted more than a month. These conversions of dieting periods to estimated average dieting length in days is shown in Table 4.9.
Chapter 4. Baseline methodology and descriptive statistics
Table 4.9 Estimations of the number o f dieting days per diet for each of the dieting duration periods
Response categories Estimated no o f dieting days
Less than one day 1 day
Between one day and one week 3 days Between one week and one month 12 days
More than one month 60 days
Combining the duration question and the frequency question to estimate the number o f dieting days in the past year. When the frequency question and the duration question were combined using the multipliers shown above, a small number of girls (6% of the dieters) were classified as having dieted for more than the possible 365 days in the past year. In addition, the resulting dieting intensity scale was strongly positively skewed since the majority of participants reported dieting levels towards the lower end of the scale.(Figure 4.l)ln order to minimise the effects of this deviation from normality in analyses using dieting intensity scores, the score was converted into a dieting intensity rating, dividing the participants into four groups. The first group was composed o f individuals who reported never having dieted at all in the last year, including those who have never controlled their weight. Participants who reported dieting at all were split into tertiles depending on their reported dieting intensity (occasional, moderate and frequent dieting groups). Boundaries for the tertiles fell at 20 and 60 days, such that dieters were placed in the 'occasional dieting' group if they were estimated to have dieted less than 20 days in total over the past year, moderate dieters were defined as those who had dieted between 20 and 60 days in total, whilst frequent dieters were those who had dieted more than 60 days. Any participants who had failed to give a numerical estimate of the number of times they had been dieting but claimed to diet ‘all the time’ were included in the frequent dieting category.
Chapter 4. Baseline methodology and descriptive statistics
Figure 4.1 D ecile values for num ber o f dieting days in the past year (dieters only), show ing tertile cut points.
D ietin g d a y s 250 200 150 100 50 0 33.3% 66.6% — ---1- ■==--r ■=! .-C Il _ ---
,
10 20 30 40 50 60 P e r c e n t i l e 70 80 90V alidating the dieting intensity ra tin g The m ethod for deriving the dieting intensity rating discussed above is a procedure inevitably likely to introduce error to the resulting scale, since the total num ber o f dieting days is derived from estim ates o f diet duration. For this reason it w as im portant to assess w hether the dieting intensity rating had levels o f convergent validity at least equivalent to the two scales from w hich it w as derived. Three m easures know n to be closely linked to dieting were selected as validation criteria for the dieting intensity rating: the D EBQ restraint subscale (Van Strien et al 1986), the ChEat instrum ent for assessing sym ptom s o f anorexia nervosa (M aloney et al 1988) and a m easure o f body dissatisfaction, the self-ideal discrepancy m easured using a figure rating scale (details in section 4.2.8.1). Spearm an correlations betw een these three m easures and the four dieting intensity groups, the four dieting duration categories and dieting frequency quartiles can be seen in table 4.10.
Chapter 4. Baseline methodology and descriptive statistics
Table 4.10 Spearman correlations between three measures of dieting and three validation measures: dietary restraint, anorexic symptoms and body dissatisfaction.
N=615 Dieting intensity Dieting duration Dieting frequency
DEBQ Restraint .38 .30 .34
ChEat Anorexic symptoms .43 .31 .41
Body dissatisafaction (current- .29 .24 .18 ideal)
All correlations significant at p<0.001
The dieting intensity rating was found to be somewhat more closely correlated with the three validation measures than either of the measures from which it was derived. The rating was strongly associated with dietary restraint, anorexic symptoms and body dissatisfaction, and on this basis it was considered to be a useful and acceptable alternative to using frequency or duration alone.