Multi-centre trial = better than single-centre?

That sounds like the kind of thing that is intuitively rather obvious, but you would likely run into a lot of problems if you were to attempt to quantify it.

From basic statistics though, one could argue that the quality of a study improves as N increases because errors in counting statistics are inversely proportional to the root of N. And a multi-institutional study is likely to have more numbers than a single institutional one.

Further, one might argue that biases are likely to be reduced across difference centres.

One issue that comes up is protocol control. In radiation oncology, for example, centres that participate in clinical trials are usually required to verify their dosimetry calibrations through a third party so as to control for inter-institutional differences.

 

I think Choppy is spot on here. I would be a bit surprised if you needed to cite research about designing a multi center study. It is more or less the statistics which favor such designs in order to recruit an appropriate amount of participants to increase statistical power (http://en.m.wikipedia.org/wiki/Statistical_power) and to avoid sampling bias (http://en.m.wikipedia.org/wiki/Sampling_bias). The former may not be the most important factor if the study will be performed in a large city where you can recruit enough subjects to detect some minimum change in the end points being studied. However even if you can recruit enough subjects to get the appropriate statistical power, the sample that you choose may not be representative of the entire population.