Note: The following is excerpted from an article written in 2005. For lay (Non trained) people, there are more updated recommendations. The following is an example of the process, not the current recommendations.
To be effective, CPR must restore adequate coronary and cerebral blood flow. Interruptions in chest compressions lower coronary perfusion pressure and decrease rates of survival from cardiac arrest. In the first minutes of VF SCA, ventilation does not appear to be as important as chest compressions, but it does appear to contribute to survival from prolonged and asphyxial arrest. Certainly the ventilation rate needed to maintain a normal ventilation-perfusion ratio during CPR is much smaller than normal because pulmonary blood flow is low.
That is one of the opening paragraphs from this article published on the American Heart Association (AHA) website from 2005, regarding CPR evaluation and recommended changes. (VF SCA = Ventricular Fibrillation Sudden Cardiac Arrest)
As a summation, a group of scientists and heart experts (281 experts over 36 months) gathered to review all the study and epidemiological data regarding survival rate in witnessed SCA. They reviewed the (then) current survival rates, sequence and priorities in CPR to see how differences affected the survival rate. There are 57 articles cited throughout the article that are all listed with links.
One of the biggest factors was that few people received early CPR, and those that did, it wasn't always effective CPR. Some of the factors that they found were that chest compressions were inadequately performed, slow, and often interrupted for too long for rescue breaths, especially among lay CPR performers (Non EMS trained). This resulted in inadequate cerebral blood flow and cardiac output.
Once they determined that, the following excerpt explains how they evaluated the recommended changes (at that time, this is a 2005 publication.)
Mathematical and animal models showed that matching of pulmonary blood flow and ventilation might be more appropriate at compression-ventilation ratios higher than 15:2. There was concern, however, particularly among pediatric experts, that inadequate ventilation rates could reduce survival from pediatric and asphyxial (eg, drowning) arrest. To achieve optimal compression rates and reduce the frequency of interruptions in compressions, a universal compression-ventilation ratio of 30:2 for all lone rescuers of victims from infancy (excluding newborns) through adulthood is recommended by consensus, based on integration of the best human, animal, manikin, and theoretical data available. The 30:2 ratio is recommended to simplify training in 1-rescuer or 2-rescuer CPR for adults and all lay rescuer resuscitation. A compression-ventilation ratio of 15:2 is recommended for 2-rescuer CPR (a skill taught chiefly to healthcare providers and lifeguards) for infants and children (to the onset of puberty). This recommendation will result in the delivery of more rescue breaths per minute of CPR to victims with a high prevalence of asphyxial arrest.
So basically, a panel of experts gets together and reviews current literature and reports on survival from both field and hospital based sources. Along with that they review changes and practices that have been implemented (There is reference to early Automatic External Defibrillator {AED} devices being a key factor in survival rates) since the last recommendation. They pair this with animal and computer modeling to determine what should used for compression rates and compression/breath ratios.
As far as the differences, I do not know how to account for those other than people using outdated information, or not following guidelines. Here in the United States, (as far as I know), all CPR certification falls under the guidelines of the AHA.