This analysis was developed in 2007 based on voluntary surveys completed by people in the LiveJournal site ‘proanorexia’. The goal was to understand the age, BMI, gender, and other demographic characteristics of members of that group.
This was written nine years ago based on interaction with people on the LiveJournal group ‘proanorexia.’ It is shared for historic value and to give insight into the mindset of people with eating disorders, not to encourage disordered behaviors.
This is the term paper for a psychology class on experimental design. It describes the scope of a small experiment to investigate the connections between mothers and daughters with regard to eating habits and eating disorders.
Homework Assignment #1
Design an Experiment
Name: Glenn Booker
Instructions: For this assignment you will be designing a research EXPERIMENT. Come up with a question of interest and specify how you would design your experiment.
- Research Question: (i.e. does X cause Y?)
Does seeing ‘triggering’ images increase binge-eating behavior in adult females with Binge Eating/Purging Type (BP) anorexia nervosa?
Increased exposure to triggering images (e.g. images of very thin people) will make people with BP anorexia more likely to overeat (‘binge’) in the next 24 hours, as measured by a statistically significant increase in the Binge Percentage.
- Test Your Hypothesis
- a. Independent Variable:
- b. Operational Definition:
The duration of exposure in seconds to visual images of females with a body mass index (BMI) of 20 or less. The BMI is defined as:
BMI = 703*weight/(height)^2 where weight is in pounds and height is in inches.
- c. Dependent Variable:
- d. Operational Definition:
The percent of test subjects who binge within 24 hours of seeing the stimuli, as self-reported by survey. A binge is defined as consuming over 2000 calories of food at one sitting.
- Who is your population?
The population for this study is humans who have been diagnosed with the BP type of anorexia nervosa, but not the restricting type of behavior, as defined by DSM IV.
- Who is your sample?
The sample will be 100 (I’m guessing!) young adult females (age 18-24) who have been clinically diagnosed with BP anorexia nervosa. All test subjects will be part of both the experimental group and the control group, as described later.
No restrictions on race, income level, ethnicity, religion, or other characteristics of the test subjects.
While many females aged 14-17 also suffer from anorexia, it is much easier to get a study approved for adult test subjects, hence the age range chosen.
Males also suffer from eating disorders, however in much smaller numbers than females, hence obtaining an adequate sample size for males would be much more difficult, hence the limitation on gender.
- a. Experimental Group:
The experimental group will see a ten-minute video on three separate occasions, into which varying amounts of triggering images have been introduced; from clearly subconscious levels (a couple of frames of triggering video here and there) to more and/or longer duration images that might be consciously recognizable by the subjects.
- b. Control Group:
The control group will see the same video with no triggering images in it.
- Describe your methods and steps of conducting your experiment (i.e. describe the manipulation of variables and control of extraneous variables. Tell me how you would carry out this experiment and what it would look like).
The test subjects will be lied to. They will be told we’re conducting a study to test the effectiveness of watching soothing images (a babbling brook, kittens playing, etc.) on calming eating disordered behaviors. They will fill out an initial survey to provide the duration of their illness, typical caloric intake during the last four weeks, and how often they currently binge.
They will then be brought in four times, the same time and day of the week, to watch a video we’ve prepared. About 24 hours after they have seen the video, they will be prompted to fill out a brief survey about their eating habits in the last day, in addition to some extraneous questions whose answers we’ll ignore.
In reality there will be four slightly different videos prepared:
- The control video will be just the soothing images.
- The other three experimental videos will have varying amounts of triggering images hidden in them
The test subjects will see the videos in random order; this will allow the study to be double blind to prevent bias from the experimenters.
The key output from the survey will be their caloric intake during the 24 hours following seeing the images, and whether that intake included a binge.
Analysis of the data will determine the percent of subjects who binged after seeing each of the four videos. This will be compared to the percent of time they binge under normal circumstances, to assess whether there is a statistically significant increase in the Binge Percentage.
|Video number||Exposure Time||Binge Percentage|
where Z > Y > X
The expectation is that the Binge Percentage will increase as Exposure Time increases.
- Name 1 threat to your internal validity?
Many other factors could affect the caloric intake of the test subjects – stress from family or school or work, holidays, for examples.
If test subjects become consciously aware of the deception, then the test results for that subject are presumably invalid.
The test subjects know we are treating their eating disorder, so their eating habits might improve temporarily just for show.
- Name 1 threat to your external validity?
The scope of the test is limited to adult females, as noted earlier, hence the applicability of the results to males or those outside of the age range tested could be questioned.
The test subjects are also likely to be from the immediate geographic area near the test facility, so the applicability of results to other cultures, demographic levels, etc. could be questioned.
The test subjects have been clinically diagnosed; hence the results may not apply to those with eating disordered tendencies but do not meet the criteria of DSM IV.
Many people with eating disorders have multiple diagnoses, so finding test subjects could be difficult, and applicability of the results to people with multiple diagnoses could be limited.
- Identify possible ethical problems (i.e. will you use deception? What are the risks?):
We are deliberately triggering eating disordered behavior, so that’s a clear ethical issue with this experiment. A major resulting risk is that one or more test subjects could be so severely triggered that they binge uncontrollably. Follow-up with the test subjects is advised to look for such complications, as well as possible health complications from purging.
We are deceiving the test subjects, so that is an ethical issue as well. It is hoped that the benefits of the study outweigh the impact of the deception.
- What are you hoping will be implications of your experiment (why should others care about your study)?
It is widely reported that triggering images (e.g. “thinspiration”) can exacerbate eating disorders. The purpose of this study is to assess the accuracy of that impression, and quantify the extent to which it is true or not.
- Who would benefit from the results of your study?
The results of this study would benefit people with eating disorders, their therapists and counselors, and peer support (pro-recovery) websites. It might also help shut down pro-ana or pro-mia web sites, which encourage people to follow anorexic (ana) or bulimic (mia) eating practices.
 “Eating Disorder Diagnostic Criteria from DSM IV-TR,” http://casat.unr.edu/docs/eatingdisorders_criteria.pdf.
 Booker, Glenn. (2007) “Analysis of Surveys,” an unpublished study of self-reported age and weight characteristics of 83 people in an eating disorder support group on LiveJournal.
 Dias, Karen. (2003) ” The Ana Sanctuary: Women’s Pro-Anorexia Narratives in Cyberspace” Journal of International Women’s Studies Vol 4 #2.
Research Paper Writeup
Glenn Booker PSY101
The research paper selected is: Attia, Evelyn. (2010). Anorexia Nervosa: Current Status and Future Directions. Annu. Rev. Med. 2010. 61:425–35. As a review paper, it summarizes many other papers, and does not present an experiment in detail, so the specific questions of interest can’t be addressed.
Anorexia nervosa (AN) is described in terms of its demographics, mortality rate, and history. The two major types of AN are defined, restricting (R) and binge-purge (BP); the latter is distinguished from bulimia by low patient weight. The epidemiology of AN is discussed; the lifetime prevalence rate is between 0.3 and 1.0% for men and women, but differences in reporting may be due to better recognition of the disease by physicians.
The phenomenology and course of illness are discussed in detail. AN tends to be ego-syntonic (consistent with one’s sense of self), so patients don’t recognize symptoms as being anything wrong. Weight loss or lack of normal weight gain with age is typically accompanied by “food restriction, secretive eating, vomiting or other purging after eating, and excessive exercise.” (all quotes from (Attia, 2010)) They often believe that they are fat, even when normal or underweight. Females with AN sometimes experience amenorrhea (no period), but that is now less regarded as a meaningful symptom of AN.
Medically, AN slows the metabolism, causing bradycardia, hypotension, and hypothermia (slow pulse rate, low blood pressure, and low body temperature, respectively). Lower levels of white and red blood cells (leukopenia, anemia) and electrolytes (hypokalemia (potassium), hyponatremia (sodium)) are often present as a result of purging or drinking large amounts of water. Low levels of hormones (estrogen, testosterone) and low bone density (osteoporosis). Changes in heart rhythm (prolonged QT interval) increases the chance of sudden death. Hair loss and skin changes are common.
Psychological symptoms include “distractibility, depression, anxiety, agitation, sleep disturbance, obsessionality, and compulsivity,” but these are mainly a secondary result of severe caloric restriction. Recovery from AN is too rare – of those seeking clinical treatment, recovery rates from 30% to 67% were reported by various studies, over time periods from 5 years to a median 90 months (7.5 years). The mortality rate from AN was estimated at 5.6% per decade of illness, significantly higher than bulemia.
“Treatment for AN remains a challenge, as no treatment has clear empirical support.” Problems with good evidence-based treatment experiments include: rarity of patients with the illness, reluctance of patients to participate in a study to restore normal weight, and the use of complex and expensive treatment regimens. The effectiveness of medical treatments is hard to assess, since they are done in conjunction with psychiatric treatments which are already known to be effective. Antidepressants have not been shown to be effective in AN treatment compared to a placebo. “Clinical trials of psychotherapies have failed to identify effective interventions for underweight adults with AN.” For children and adolescents, family-based therapies have been shown effective, based on preliminary data. Behaviorally oriented treatment programs with strong “psychosocial support and reinforcement” have been offered in many different settings, and there is some evidence of success. [Editorial comment: they discuss using 3000-4000 cal/day diets in this context. I can’t imagine a person with AN who could begin to eat that much!] These programs can result in refeeding syndrome, an uncommon adverse effect that can cause severe fluid and electrolyte imbalances, and result in “congestive heart failure and acute mental status change.”
Relapse prevention is important because relapse rates are high (though not quantified). One small study (n=32) found that, for patients with the same BMI after weight restoration, those with higher percent body fat were more likely to remain a treatment success.
The biological causes of AN remain unknown. Twin studies have shown a genetic component to vulnerability to AN. Studies have shown that “anxiety, obsessionality, and perfectionism appear to be associated with AN.” Our only conclusion so far is that “the illness probably results from an interaction of genetic, developmental, and environmental factors.” Other biological factors have been investigated (monoamine neurotransmitters, serotonin (controls appetite, also seen in OCD), dopamine (is dysregulated), norepinephrine) as contributing causes of AN. The psychological traits of AN have led to study of “behaviors relevant to choosing, learning, and adapting” and have shown that “the ability to move back and forth between tasks or operations” (set shifting) is impaired in patients with AN.
These results have produced a model that integrates the biological and psychological aspects of AN (Figure 1), starting with predisposition to AN from high anxiety, obesessionality and perfectionism, amplified by our culture into rigid dieting practices, producing physical changes that strengthen the predisposition in a vicious cycle.
Figure 1. A Possible Model for Anorexia Nervosa