Summary on pitfalls of data that have driven decisions written by Megan MacGarvie and Kristina McElheran

Summary on pitfalls of data that have driven decisions written by Megan MacGarvie and Kristina McElheran

Chapter 15

This chapter is about the pitfalls of data that have driven decisions written by Megan MacGarvie and Kristina McElheran. He states that managers are still likely to be vulnerable to an array of pitfalls when applying data to back up their hard decisions even if they apply impressive large data sets, careful statistic methods, and best analytics tools. There are different traps in the decision-making process coming from the fact that individuals do not carefully process all pieces of data in all decisions. They instead depend on simplified procedures. Megan and Kristina have provided three primary cognitive traps that normally skew decision making, even when people use the best data.

The first cognitive trap is the confirmation trap, according to Megan and Kristina. One is likely to fall into this trap when he pays more attention to the findings aligning with the belief he had before and ignore other facts and patterns in the data. This can happen even with small data due to focusing n expectations on how the world has to operate. Peers and bosses can make it hard to prevent confirmation bias by providing pressure. It is better to prevent avoiding information that may counter one’s beliefs to obtain a fair assessment of what the data communicates. It is therefore advisable to specify analytical and data in advance, look for findings actively that disapprove beliefs and avoid dismissing findings automatically

The other trap is the overconfidence trap. This where people seem to assume that the accuracy of their judgments is more favorable than the data would suggest. It may also encourage other pitfalls, even if psychological. More information can increase overconfidence and fail to enhance accuracy. To avoid it, it is necessary to describe the perfect experiment, making it a formal part of the process, and finally, before deciding, perform a pre-mortem and keep track of the predictions.  The other trap is over-fitting, which happens when a statistical model describes random noise instead of the underlying connection one needs to capture. Although they typically do suspiciously good work, they have a significant difficulty forecasting the future. To guard this trap, one can randomly divide the data into two groups, specifying the relationship being tested, keeping the analysis simple, construct alternative narratives, and being aware of all-too-human tendency. Dealing with these biases with the right mindset will lead to better analysis of data and better decisions.

Chapter 22

The chapter was written by Nick Morgan, who states that decisions don’t start with data. He begins with a description of how an executive involved his colleagues in making a decision. He states that to affect human decision making, one has to be at the place where decisions are really made, and that is in the unconscious mind, which is controlled by emotions, but there is no data. Although data is useful as a supporting material, it should be used with care as it spurs thinking in the unconscious mind. Members of the decision-making team have to be persuaded using stories. Data offers new insight and proof to inform in decision making.

Critical Issues

Chapter 15

The chapter contains three critical issues involving the traps that an analyst can face. The first critical issue concerning the confirmation trap is that one can fall into this trap by paying attention to findings that align with the prior beliefs and also ignoring facts. This becomes critical since it can be counterproductive and costly. The other critical issue is where the statistical model describes random noise instead of the underlying connection one may need to capture, leading to the over-fitting trap. The issue can make it difficult to predict the future. The third critical issue is the overconfidence trap. It is critical since people tend to assume that the accuracy of their judgment is more favorable than what the data would indicate. This is critical since it can lead to poor decision making.

Chapter 22

There are several critical issues, although the chapter is short. The first critical issue is that a company had to do away with a longtime vendor to favor a new one. It is critical because members have to agree with the executive and decide to favor a new vendor. The other critical issue is that the members did not support the executive’s idea. They failed to support it since they had an established connection with the longtime vendor. It is critical since they were opposing, and he had to persuade them. The third critical issue is that the executive wanted to persuade them using data only. It is critical since data alone would not work.

Lessons learned

Chapter 15

I have learned several lessons in this chapter. However, there are three most essential lessons. I have learned that data analytics can be an effective tool to enhance consistency and shared understanding, but it can also make us complacent. I have also learned that as a manager, one needs to be aware of the pitfalls, including over-fitting, overconfidence, and the confirmation trap.

Chapter 22

I also learned several lessons from this chapter. I learned that to make a decision, especially like this one, where members have to separate from a supplier who they have a long relationship is difficult. I also learned that it is not all about data to convince people to support a certain decision. It is about persuading their emotions.  I have also realized that managers face tough times in the process of decision making. For example, the executive has to persuade the members in order to support him.

Best Practices

Chapter 15

The best practices I have learned in this chapter is those of overcoming the pitfalls. The first practice is specifying in advance and analytic approaches, actively looking for findings that disapprove of one’s beliefs and avoiding automatically dismissing findings that fall below the threshold for statistical significance. The other practice is describing a perfect experiment and making it a formal part of the process. The third best practice is avoiding over-fitting pitfall by randomly dividing data into two sets since it can be significant at predicting within the training set.

Chapter 22

The first practice I noted in this chapter is the use of data to show that he was making the right decision, although it could not work alone. The other best practice is making a decision in the right place of the mind to influence human decision making. The final best practice is using effective persuasion. This is because, with data, it can be very useful.

Relation to Topics Covered

The content in the chapter relates to the topics covered in class. The idea that people face traps in data analysis since they aren’t careful is common in the topics covered. Chapter 22 also relates to topics covered in the class as it is about making the right decision and including members in decision making.

Alignment to Class Concepts

Chapter (15) aligns with the class concept since it is all about accurately analyzing data for accuracy in decision making. Chapter 22 also aligns with the class concept since it talks about using data to make informed decisions. In class, the concept is that one has to collect data, analyze, and then get the alternatives to make an informed decision. However, the chapter has something that the class concept lacks; the chapter talks about effective persuasion. Using both can be effective in making business decisions.