The objective of thisarticle was to identify factors that pose risks particularly associated witheating disorders in adolescences. Doctor Rhode, Stice and Marti conducted alongitudinal study in which participants were to complete self-reportinterviews and assessments regarding body image and dissatisfaction. The studyexamined six different variables and the period of developmental change knownas adolescence. In attempt to understand the theoretical relationship betweenthe two, they concluded that five out of the six predictors impacted the rateof early onset eating disorders. After reviving this article, I’ve decided thatsome of the methodological decisions made by the researchers could be improved.
This experiment includes flaws regarding external validity and interraterreliability but does a really good job with establishing construct validity. The first factor that I decided was an issue was thedemographic of the sample obtained. Almost nearly 70% of the participants were Caucasian,and the researchers only sampled students from major private schools. Thedesign of this experiment wasn’t necessary an inappropriate choice for whatthey were trying to asses but the researchers could’ve better controlled forbias during the sampling and collection of data. For a large majority of thepeople who do attend private schools are usually upper-class Caucasian studentsand for that reason, I don’t agree with that being an accurate representationof the population.
Let alone the population of people that suffer from eatingdisorders. I feel like the researchers could’ve randomly sampled students from severalpublic schools which would’ve given them a sample that was much more representativeof middle schoolers. Threats to externalvalidity make us question a studies design, results and applicability, so themore demonstrative our sample is the more we can generalize to the greaterpopulation.
The second flaw I noticed was that the experiment did notdemonstrate consistency when obtaining scores. Interrater reliability was notestablished throughout the course of the experiment. The study used assessmentsand interviews to collect data and scores based on mental health and bodydissatisfaction. The scores were not perfectly consistent with that of theother observers and as mentioned in the study, researchers were unable to obtainsome scores at the beginning of the experiment. Although there is no such thingas getting “perfect” scores each time you conduct an experiment, it isimportant that your scores are relatively consistent no matter who ismeasuring.
I would suggest improving interrater reliability by establishingclear cut guidelines and questions to improve the quality of assessments and interviews.I believe the researchers could’ve also had better coders as well as more ofthem. The last methodological factor I am going to discuss isnot considered a flaw, in fact it is a perfect example of how well theresearchers established construct validity. Concepts and variables were accuratelydefined which allowed them to properly assess body image, the pressure to bethin and mental health.
They used Likert scales that measured and assessed thedifferent ways in which participants felt about, thought about, and sawthemselves which were key indicators in determining onset eating disorders.Each type of assessments evaluated the relationship between these disorders andage concluding that body dissatisfaction, perceived pressure to be thin,thin-ideal internalization, dieting and negative affectivity each have animpact on the development of eating disorders for women ages 13-15. Ultimately,the scales that were used were reflective of what they were researching.
Overall this experiment was well designed and appropriatefor understanding the theoretical relationship between the different stages ofdevelopment and when these risk factors first present themselves. I believethat the researchers could’ve picked a larger and fairer sample that would’veminimize bias and would’ve better control for generalizability. The more representativea sample is, the more significant your results will be. Also, I’d suggest that the researchers could’veprovided clearer instructions for the coders and interviews. The more consistentyour scores are, the more statistically valid the findings are. These recommendedimprovements would not only change the results of the study but they would alsoimprove the validity and reliability of the entire experiment making it almostnearly a perfect one.