• No se han encontrado resultados

3.3.1 Community sample

If the statistics of the Health Survey for England 1999 (Department of Health, 2001) are taken to offer a true reflection of the level of overweightness in the general population (58%), then the percentage oven^/eight or obese in the Community Study (36%) is considerably lower than that in the general population. Similarly, the percentage of the Community Study that was obese (9%) is lower than that reported in the 1999 survey of the general population (18%).

One explanation for this difference could be that the age, education and weight of the sample population in the Community Study is truly different to the UK population. We know from studies on the prevalence of overweight and obesity that a lower weight usually coexists with the characteristics of this sample (i.e., lower BMI with higher education, higher SES and younger age - especially for women; Bajekal, Boreham, Brens, Falaschetti, Hirani, Primatesta, Prior & Tait, 2000; Colhoun & Prescott-Clarke, 1996.).

A second reason for this difference in findings could be that there was erroneous or biased reporting. For instance, the respondents may be representative of the target population and general population, but they may have underreported their age and weight (and/or over-reported height), and overreported their educational level. This may have been motivated by social desirability bias. Support for this explanation comes from a large- scale community study by Flood et a!., (2000) who found that those who were overweight and obese significantly underreported their weight, a finding that was particularly pronounced in women. In fact, this biased reporting has been shown to be most pronounced in people with higher educational qualifications and who have professional jobs (Jeffery, 1996). Also, the Health Survey for England 1999 used interview as the method of data collection which would make it more difficult to give inaccurate information, especially about weight and height, than in the self-report method used in this research.

A third explanation could be that the differences in reported findings are as a consequence of selection bias. As the questionnaires were sent to a random selection of people registered on two GP lists it is possible that those who elected to return the questionnaire were different from the target population in terms of their age, educational level and weight. Indeed, this scenario receives support from knowledge that non­ responders usually differ considerably from responders in terms of their educational level (Barker, Pistrang & Elliot, 1994).

The difference in age, educational level and weight between the Health Survey for England 1999 and the Community Study sample are quite probably a product of all three

reasons. In other words, the differences reflect actual differences between the general population and target population, between respondents and non-respondents, and the self-selection bias. It is not possible to estimate with any real accuracy the relative contribution of these reasons. Nevertheless, they are likely to make a significant contribution to the reported differences.

3.3.2 Clinical sample

A strength of the Clinical study is that the sample included a significant proportion of respondents who are classed in the highest BMI category (WHO Expert Committee, 1995). Twenty-two percent were morbidly obese (BMI > 40kg/m^). This is important, given criticisms made of other research that few previous studies have included people who fall into this category of obesity (e.g., Williamson, 1995). This means that my research has gathered beliefs about genes and obesity from each category of obesity and is therefore likely to be more representative.

The Clinical Study sample is likely to be highly representative of the target population - i.e. patients attending the hospital dietetic clinic for weight loss. This is because all suitable patients who met the inclusion criteria (i.e., those who attended, who were referred for help with weight reduction and who were aged 18-65 years) were approached. Out of these patients, 86% took part in the study. Therefore, only a relatively small proportion of the potential sample failed to be included.

The sample did not, however, include some patients who cancelled or who failed to attend their appointment. Some of these patients attended a subsequent appointment during the ten weeks of data collection and were included in the study. It is possible that the group of patients who were not included may have expressed different beliefs to those who participated in the study, but the number of patients in this category is relatively small.

The non-inclusion of this group does not influence conclusions drawn about patients attending \he dietetic clinic for help with weight reduction.

A high percentage of participants in the Clinical Study were females (85%). Given the high response rate, this was representative of clinic attendees. The Dietetic Service manager confirmed that this high proportion of females was representative of the patients seen by the service. While the beliefs of females may be different from males in the clinical population, this was not evident in the Community Study, it still is a risk.

3.3.3 Experimental sample

The Experimental sample is highly representative of the target population - undergraduate students reading an arts subject at a London university. This is because all 125 attendees completed the questionnaire, and the questionnaires of only two (1.6%) attendees were excluded because they were not sufficiently complete. However, the sample will not be representative of the UK general population.

3.3.4 Reliability sample

The Reliability sample is probably not very representative of the general population. This is because the sample is relatively small (N=32) and respondents were sampled from two work sites. However, the central purpose of the study was not to provide beliefs from a highly representative sample of the population, but rather to administer the questionnaire to people on two occasions, with a specified interval of time between administrations, and under similar conditions. These aims were realised.

Documento similar