Andy Updegrove made some interesting comments some time ago comparing Wikipedia to archeology. The idea is that by being able to recover the multiple layers behind the construction of a Wikipedia entry, one is capable of reconstructing the collective thought process behind it. In a previous post I have argued something similar in relation to scientific writing, and now I will try to argue that by doing so there will be direct implications for the improvement in the degree of objectivity we will be able to gain in science.
But does objectivity have to do with the ability of reconstruct layers behind a scientific text? According to the feminist philosopher Sandra Harding (Harding, 1995), the objectivity aim of science is simply unattainable. Although researchers frequently claim that their experiments should be as objective as possible, subjectivity is inherent to every research experiment. Let's take as an example a study comparing two medications to treat pain. The central pillar in current clinical research is to use the so called randomized controlled trials (RCTs) to test and potentially approve medication for entry in the US market. The choice of RCTs was made since it theoretically capable of reducing the internal bias within a study. In other words, by randomly assigning the pain medication to two groups of patients, the groups become balanced or comparable in terms of the severity of pain, thus decreasing the internal bias or likelihood of one group responding better to the medication compared to the other group because they have patients with different pain levels. But does this mean that RCTs are really objective? The answer is absolutely no. While designing a RCT, I might know based on pilot studies that the pain medication being tested is better because it is capable reducing effects on patients stomach such as heartburn, but that that it might be inferior in its safety profile and cause, say a heart attack. In other words, my study can legitimately focus on that single aspect that I expect to be superior on the medical, leading the RCT to show that this new medication is better than its predecessor. In the meantime, the safety issue is hidden, not because a company is intentionally trying to hide it, but simply because that was not the chosen focus of the study. Unlikely? Well, that was the case in the recent case of Vioxx.
Multiple other examples could be raised where the choice of focus of a study leads results in a certain way, thus introducing a subjectivity into the study. This perspective makes it clear that the pure data are not, as many would like to believe, judgment-free. By deciding to focus on a certain aspect of a phenomenon, and by consciously or unconsciously deciding not to approach a different phenomenon, researchers are biasing their results toward whatever they might be trying to prove.
But is this a catch 22? If by the mere choice of design I am introducing subjectivity in my design, how can science achieve its goal of objectivity? Well, that's a much longer discussion, to be continued ...
But does objectivity have to do with the ability of reconstruct layers behind a scientific text? According to the feminist philosopher Sandra Harding (Harding, 1995), the objectivity aim of science is simply unattainable. Although researchers frequently claim that their experiments should be as objective as possible, subjectivity is inherent to every research experiment. Let's take as an example a study comparing two medications to treat pain. The central pillar in current clinical research is to use the so called randomized controlled trials (RCTs) to test and potentially approve medication for entry in the US market. The choice of RCTs was made since it theoretically capable of reducing the internal bias within a study. In other words, by randomly assigning the pain medication to two groups of patients, the groups become balanced or comparable in terms of the severity of pain, thus decreasing the internal bias or likelihood of one group responding better to the medication compared to the other group because they have patients with different pain levels. But does this mean that RCTs are really objective? The answer is absolutely no. While designing a RCT, I might know based on pilot studies that the pain medication being tested is better because it is capable reducing effects on patients stomach such as heartburn, but that that it might be inferior in its safety profile and cause, say a heart attack. In other words, my study can legitimately focus on that single aspect that I expect to be superior on the medical, leading the RCT to show that this new medication is better than its predecessor. In the meantime, the safety issue is hidden, not because a company is intentionally trying to hide it, but simply because that was not the chosen focus of the study. Unlikely? Well, that was the case in the recent case of Vioxx.
Multiple other examples could be raised where the choice of focus of a study leads results in a certain way, thus introducing a subjectivity into the study. This perspective makes it clear that the pure data are not, as many would like to believe, judgment-free. By deciding to focus on a certain aspect of a phenomenon, and by consciously or unconsciously deciding not to approach a different phenomenon, researchers are biasing their results toward whatever they might be trying to prove.
But is this a catch 22? If by the mere choice of design I am introducing subjectivity in my design, how can science achieve its goal of objectivity? Well, that's a much longer discussion, to be continued ...
No comments:
Post a Comment