In the previous article we introduced Randomised Control Trials (RCTs) as a methodology that was first deployed in medical industry. Now it has helped to improve decision-making for a broad range of topics including agriculture, microfinance, malaria and HIV prevention, to name just a few. RCTs can help with both governmental and corporate decisions. They allow you to study outcomes – employee retention, growth of small companies and behavioral changes. It is a tool enabling you to test a program before it's been fully rolled-out.
Elimination of selection bias
A good example is corporate training. Companies invest money in courses that are expected to have positive impact on performance. However, it is possible that only the most productive employees are selected to take part in the training, making the impact of the training overstated when compared to the performance of the rest of employees. On the other hand, when less productive employees finish the course, their productivity may have improved, but they still look less productive when compared to other employees, who were more productive before. Again, the true impact of training course remains hidden.
Small scale experimentation
To sum up, when working with observational data, you never know whether you really see a true causal effect. RCTs implement the program on a randomly selected set of employees and compare their results with another randomly selected set of employees, who didn’t take part. Any possible selection bias is eliminated. With the development of IT and increased usage of data, trials today are more accessible. Evidence-based recommendations are achievable.
-jk-