search

Neo-Cat / NBCU-Loss-Percentage-Research

0 stars 0 forks source link

Evans' Proposal comments #1

Open rickecon opened 7 years ago

rickecon commented 7 years ago

@WanlinJi ,

Your research question seems too broad. Can you narrow/focus the scope? What is the importance of answering your question? What does an estimate of the bias tell you?

1) You need something more than quantifying the bias. Most of your proposal described how you will quantify the bias. 2) How can you distinguish which of the possible explanations is driving the bias? That would be a good research design and question.

Neo-Cat commented 7 years ago

Hi Professor,

Here is what I think about this topic. I intend to make it an empirical study in the setting of a lab experiment that examining the model of Bayesian updating in people estimating their abilities and adjusting their belief. As what I found out in the past literature on behavioral economics, I think this study can have ad hoc applications (like those field experiments that can perform similar tests in a real setting like an investment company) as tracking people's decision associated with their beliefs/demand (a metal process that affect decision before customers have a chance to consider budget and service), like how to make a quantitative marketing strategy for businesses like a gym.

  1. Eliciting the belief is part of my experimental design. Based on my hypothesis of Bayesian updating, I intend to let the experimental data suggest a disciplined way for theorists to relax Bayes’ rule, preserving the core properties of invariance, sufficiency and stability while allowing for biased interpretations. By doing this, I hope to derive the updating behavior of a decision-maker who optimally interprets the informativeness of positive and negative signals to balance the demands of ego and decision-making.

  2. Past literature proposed various interesting explanations of "Motivated/Instrumental beliefs", that kind of belief having internal utility from biased beliefs. Knowing what the bias pattern is like could help marketing strategies to some extent, but the next step towards a sound theoretical study would be looking for the systematic explanations. Currently, I have not found much literature that could use computation to support a more intuitive theory(cause it is still difficult to come up with an intuitive hypothesis about internal utility), but I look forward to adding them into my thesis as extending explanations.