Randomization Distribution

The randomization distribution is a method of determining how often the observed difference between treatment groups would occur due to random chance. The observed difference is the difference between treatments actually seen in the study (difference in proportion or means). In order to construct a randomization distribution, all of the data points are put into a pool and shuffled. This shuffling is called a resampling (permuting). The data are reassigned a treatment label and the difference between treatment is calculated. The underlying assumption is there is no treatment effect (actual difference of outcomes). Therefore, shuffling the labels should not matter, and the person would have the same outcome regardless of treatment. We repeat the shuffling and calculating the difference process many times. We will compare the original observed difference with all of the permuted differences. Based on this comparison, we will make a decision to see if the treatment had an impact on the results. We will conclude the treatment does have an effect if the probability of seeing the observed difference or more extreme is 0.05 (p-value 0.05). Otherwise, we will conclude there is not enough evidence for treatment effect and the observed difference was due to random chance.

Study Design

This experiment was conducted to evaluate the effectiveness of animal facilitated therapy with dolphins, controlling for the influence of the natural setting, in the treatment of mild to moderate depression. The study was carried out in Honduras, and recruitment took place in the United States and Honduras.The participants were recruited through announcements on the internet, radio, newspapers, and hospitals. Of the 30 patients randomly assigned to the two groups of treatment, two dropped out of the treatment group after the first week and three withdrew their consent in the control group after they had been randomly allocated.The study was single-blind. We use the intent to treat approach towards analyzing the data. All 30 subjects who were recruited for the experiment will be included in the analysis, whether or not they dropped out.

Antonioli, C. & Reveley, M. A. (2005) Randomised controlled trial of animal facilitated therapy with dolphins in the treatment of depression. British Medical Journal, 331.

Therapy ImprovemuCodedImprovement CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL NO CONTROL YES CONTROL YES CONTROL YES DOLPHIN NO DOLPHIN NO DOLPHIN NO DOLPHIN NO DOLPHIN NO DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES DOLPHIN YES HHHHHHHHHHDDDDDHHHDDDDDDDDDDDD