Types of Biases

Reflection on types of bias:

what examples surprised you and why? Selection bias and publication bias seem very clear so there weren’t any examples that we discussed in class that were really surprising in these categories. However, survivorship bias, recall bias and Healthy user bias were a bit more difficult to envision so a few examples stood out among these that were interesting. In the Healthy user bias category we thought that the SAT test takers go to college example was interesting since we hadn’t really thought about it that way. Many of us take the SAT because it was the thing to do at the time and were surprised to think that we were among those included in the biased sample. In the Recall bias, we were also surprised by the 80% of drivers in the top 10 since we all know that there are bad drivers out there, but apparently most of them think that they are great drivers.

What can managers do to be alert to these biases and prevent them from undermining important decisions? The most important way to prevent managers from making decisions under the influence of a bias is to be aware of the different types of biases that can exist and be on the look out for instances when they are at play. Awareness is the best prevention. For instance, if your boss or a group of colleagues hold a specific opinion that you do not agree with, you may be tempted to agree with them because a large group of people or a person of authority holds this opinion. This commonly happens in the work place and could be a type of publication bias. Being aware that you might be tempted by this bias will help you recognize the influence you are under. Another related bias is a group mentality bias. An example in the workplace is if something that you know is doing something unethical, you may be less likely to blow-the-whistle on because no one else is doing it. That is why many times once someone finally choses to overcome the influence of this bias, often other people come forward with similar experiences. Generally speaking, it is a good practice, if someone in the work place is presenting you with statistics in a discussion, to inquire about how the samples were collected to determine if there is an inherit bias that was built into the sampling that would affect how the results are interpreted.

GMO debate:

Who won: Pro GMOs are safe

Most compelling evidence: From a statistics perspective, the use of meta-analysis studies for health and other impacts including yield, pesticide application and economic impact on farmers is very convincing. Also, very convincing is the epidemiological time-series study across 20 years comparing US/Canada vs UK/Western Europe for differences in health. From a logical perspective, every major scientific body and regulatory agency in the world has reviewed the safety of GMOs and concluded that those GMO currently on the market are safe which is also very compelling. Another very important factor in our decision is the credibility of the references used for this argument. All references used for the Pro GMO side were either peer reviewed scientific journals or from government agencies.

Potential biases: One bias could be publication bias. A meta-analysis of published articles would only include those results that made it through the publication process. Many times publication bias leads to publishing only positive or interesting results; therefore, a meta-analysis of published studies would have this bias.

Other side: Anti GMOs are safe

This team did a great job at making a logical argument about a potential bias that can exist which is that corporations often publish results that benefit their company. However, they overstated the number of studies that fall into this category which team 3 addressed directly in the rebuttal, less than half of 2000 published studies have ties to industry. Also, the references used by this side are not peer reviewed, and many of them personal blogs which are often more opinion focused. This team also did a good job of appealing to peoples fear by relating GMO foods with asbestos to win the debate, which is a very effective way, which team 3 pointed out in their intro, to scare people into believing that GMOs foods are not safe.

Driving age debate:

Who won: Pro raising the driving age

Most compelling evidence: Although none of the member on our team agree with this side of the argument, we believe that this team presented the most compelling argument using statistics and logic. They clearly structured their arguments into three categories and had statistics to support each (higher incidence, immaturity and alcohol use). They effectively used the audiences emotions tied to their children to gain favor. They also provided many references supporting their statistics and claims, many from government agencies and medical journals.

Potential biases: Publication bias can again impact these arguments just like any that are using publications as sources of information. Another bias could be selection bias. In those referenced studies without actually reading the original studies, we do not know how the samples were collected. For example, one of their stats:  “more than 5,000 teens die each year in car crashes”. We don’t know if all of these 5,000 teens died specifically because of a young driver. However, in many of their other statistics they specified that the population was specifically young drivers.

Other side: Although we favor this side personally, the way that the arguments were made were not as convincing as the Pro side. Using only one angle, which was kids need to pay for college and other expenses, wasn’t enough for a persuasive argument. It would have been nice if they were able to directly address some of the statistics on youth accidents and deaths. Later in the debate they mentioned that “Young drivers are less likely than adults to drive after drinking alcohol” and gave some statistics around that topic. This would have been more impactful placed in the opening argument.

College tuition, is it worth the cost?:

Who won: Anti going to college.

Most compelling evidence: In this debate the team arguing against going to college provided the most convincing case. They even did a good job of pointing out biases in the opposing teams arguments such as pointing out that the any comparison of people that go and don’t go to college has a form of healthy user bias in that the people that go to college are more likely to have several other advantages and privileges in life, like come from a more wealthy, more educated family compared to those that don’t go to college. Also from a logic and appeal standpoint they very effectively used their knowledge of the audience to build a persuasive argument about the net present value, something that we learned to calculate in the last module, of the cost of going to college.