On Error

Today I want to write a bit about human error and it's role in our research. Although science attempts to interpret the world in an objective manner, errors in both the recording and subsequent interpretation of data can call our scientific method into question.

We use a variety of different types of data in this research. Some of this data is self-reported by the participant. I had already written a post about the problems that stem from self-reported data, since participants can often lie about themselves (how much they smoke, their medical history, etc). Although we are able to figure out some of these lies through a series of interviews and medical tests, they often hurt the process of our research by wasting time and resources on participants who are not actually eligible.

However, another sort of error comes from us, the researchers. I will talk about how these errors occur, and what we do to prevent them.

The first consideration to make about data lies in the instruments that we use to record our measurements. There are a few instruments - rulers, weight scales, blood pressure machines, CO breathalyzers, etc. We use these machines to help make accurate measurements that could not be made by humans themselves. However, there can be mistakes made through these instruments (whether it is caused by a faulty mechanism or by an error in the machine's usage). This means that the data we use is wrong, and would thus skew the results and conclusions we draw from the data. Errors in measurement can be counteracted by doing frequent measurements. Although this is done to measure the change in values over time, by comparing any value to the values next to it, we are able to make a judgement about how accurate our measurements are.

If we do indeed collect the data correctly, an error could be made in the database maintenance aspect of the research. For example, while entering our data into Excel spreadsheets, mistakes can easily be made by entering the wrong information into the spreadsheet. The best way that we can combat this is to triple check the data once it is entered. Thankfully, since we have a large number of bodies in the room, there are always multiple people who are willing to look at the same data to make sure that it is accurate.