Quantcast
Channel: Wesley Tansey » Wesley
Viewing all articles
Browse latest Browse all 13

How to Make Tough Decisions Objectively

0
0

Posted in Uncategorized

This week I had a really hard choice between two summer internship offers. One was from Google and the other was from a mid-sized company (~100 employees) that I’ve used a lot in the last couple years. It’s one of those situations where you drift back and forth in your mind, coming up with a reason to choose A then thinking “well, but B could be better because…” and vice versa. Fortunately, someone once taught me a really valuable approach to making tough decisions: Decision Sheets.

Pros and Cons

Everyone who gets in these situations and is vocal about the agony they’re going through is inevitably told to make a pros and cons list. We’ll start there and then turn that into a more principled approach later.

Here’s the criteria that I would write down on a pros and cons list for a PhD student choosing a summer internship:

  • Location/commute
  • Team size
  • Amenities
  • Compensation
  • Company size
  • Research relevance
  • Networking
  • Prestige
  • Publishing potential
  • Impact on company
  • Knowledge gain
  • Work content
  • Impact on society
  • Manager compatibility
  • Project flexibility
  • Team

Now, the traditional pros and cons list would simply have checks for the good things and minuses for the bad, then you add up and get the result. Of course this makes no sense, since it’s weighting things like knowledge gain and amenities equally, when they’re certainly not.

Decision Sheets

A decision sheet enables you to hierarchically weight different criteria in an organized fashion. The result is a single, objective score in the range of 0-10.

Hierarchical organization

The first step is to group all of your criteria into some high-level categories. Note that you don’t need to do this if you only have a handful of criteria. However, when you have a long list it’s helpful to simplify things as much as possible to avoid overweighting certain criteria that may have more fine-grained criteria.

For my internship, I organized the criteria into three categories:

Quality of Life

  • Location/commute
  • Team size
  • Amenities
  • Compensation
  • Company size

Professional Advancement

  • Research relevance
  • Networking
  • Prestige
  • Publishing potential
  • Impact on company
  • Safety

Personal Satisfaction

  • Knowledge gain
  • Work content
  • Impact on society
  • Manager compatibility
  • Project flexibility
  • Team
  • Safety

Note that I also added the Safety criteria to the latter two categories. This is meant to account for the fact that I’m unsure of the accuracy of my estimations. What is the risk that I’m going to have a horribly unsatisfying project or that I will not really get much out of this internship in terms of profession advancement? The more certain I am, the higher the score for Safety.

Once you have the categories established, you can assign weights to both the categories and the criteria.

Weighting categories and criteria

Each category and criteria gets a weight of 1-10. These weightings are relative to other values at the same level. For instance, quality of life is not nearly as important to me as professional advancement, and both are much lower than personal satisfaction.

The criteria will affect the overall score of their category, which will then influence the final overall score.

Scoring each criteria

For each low-level criteria, you score the different choices on a score of 0-10. For instance, Google has terrific amenities, so they got a 10, where as the smaller company has to be more economical and only gets a 5. Conversely, the ads group is the goose that lays the golden egg for Google, so they only get a 3 for publication potential– I may come up with an idea that I can later publish, but I’ll have to reimplement everything and use non-proprietary code and data; the smaller company has people actively publishing research based on their users, so they get a 9.

The final weights and scores I assigned are as follows:

Category Weight Other Company Google
Quality of life 7
Location/commute 8 3 9
Team size 5 10 7
Amenities 5 5 10
Compensation 9 8 8
Company size 4 7 3
Professional advancement 8
Research relevance 9 6 8
Networking 8 8 10
Prestige 7 6 10
Publish potential 7 9 3
Impact on company 8 9 7
Safety 8 6 9
Personal satisfaction 10
Knowledge gain 10 8 9
Work content 10 9 10
Impact on society 10 6 5
Manager compatibility 5 9 7
Project Flexibility 7 10 8
Team 7 10 9
Safety 8 8 10

Calculating the final decision score

Once you have all the rankings done, you can calculate the final score by doing a weighted average of the criteria to get the category scores, and then doing a weighted average of the categories to get an overall score. By that, I mean you multiply the weight of a criteria times the score you gave an option for that criteria. You sum all of the weighted criteria scores in a given category and divide by the sum of all the criteria weights to get a normalized (0-10) score for the category. Then finally you multiply the category score times the category weight, add up all the weighted category scores, and divide by the sum of the category weights to get a normalized overall decision score in the range of 0-10 for each option.

My personal overall result: 7.49 vs. 8.02– I am going to Google this summer! Feel free to look at the spreadsheet that calculates everything.

Note: do not put the results on the same page as the scores and do not look at the results until you decide you are happy with the scores. The idea is that, no matter what the result, you are going to stick to it. If you can see the results as you score the options, you will bias yourself unconsciously. Instead, put the results on a separate page, like I did in the above spreadsheet.

Conclusion

Some decisions are going to be hard to swallow no matter which way you go. The beauty of this approach is that you decouple your value system from your decision by first figuring out what’s important to you via the categories and weights. It also prevents you from biasing the scores one way or another because there are typically so many criteria that you can’t tell just by looking at the scores which answer is currently ahead. In the end, the decision will still be hard to execute, but at least you can feel confident that you thought it through objectively and in a principled manner.

If you enjoyed this article, please upvote it on Hacker News


Viewing all articles
Browse latest Browse all 13

Latest Images

Trending Articles





Latest Images