<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=536406409839660&amp;ev=PageView&amp;noscript=1">

Inbound Marketing Blog

 

Benchmarking Step 3 - Checking Validity

Posted by Andre' Savoie on Aug 25, 2010 9:18:27 AM

Making sure the results are accurate

After you determine what you are looking to evaluate in your web marketing campaign, how you will find it out, and how you can measure the results that you obtain, it is time to evaluate those results to ensure that they are valid and relevant. Any data that is going to be used for determining the course of your marketing strategy must be meaningful from a statistical perspective, which means that there must be substantial proof from an objective, numerical perspective that one of the options is superior to the others in regard to the measure that is being tested.

Margin for Error

In order to be statistically relevant, the results must have a very small margin of error. The margin of error refers to the likelihood that the information is correct, and generally statisticians will not accept any margin of error that is greater than five percent. If there is only a five percent margin of error then you could say that you are 95% certain that your conclusions are correct.

One way to keep the margin of error low is to make sure that the size of the sample is high. A bigger sample size reduces the margin of error because the more people that are included, the less likely it is that a single variable or person can throw off the whole data set.

What to watch for

There are several things that can negatively affect the validity of online testing. Some those include events that happen at that moment in time, referred to as history events, problems with the tools used in the test which are referred to as instrumentation effects, problems with the selection of the subjects so that they are not evenly distributed, which is called a selection effect, and sampling distortion effects which occur when there is not enough data collected during a given test.

Topics: benchmarking, testing and measuring

Subscribe to Our Blog!

Connect With Us