Multidimensional Decision-Making II - A logical approach for simplifying decisions (Part 2/2)

What is this article about?

The first chapter of this article series discusses the importance of making effective decisions, especially when facing challenging decisions. The second (and last) chapter focus on demonstrating an approach to clarify challenging decisions.

This second chapter in divided in two parts:

Set-up of a decision framework: An introduction to the approach, and explanation of the logical decision framework. (Required)
Calculate and interpret choices outcome: An explanation of how to use the logical decision framework to extract choices’ outcomes. (You are here)

If you haven’t read the first part, I highly recommend you to read it first, as you are currently on the second part that completely depends on the first one: A logical approach for simplifying decisions - Part 1: Set-up of a decision framework

Chapter 2.2: Calculate and interpret choices outcomes

Let’s recap

With have a decision to make, but it’s being challenging, and obviously we’d like to get the best outcome from it. So we have been trying to identify what kind of outcome we would like, which we began to do by formulating a shape. Then we did some dimensions identification and weighting, on top of the decision and the shape.

The goal of identifying the dimensions and weighting them, is to build the framework to compare each possible choice against what we want, in order to be able afterwards to get representative scores for each possibility. Where higher is the score, higher is the correlation between the outcome of the choice and what we want.

Now that we have weighted dimensions for our example, we actually have the framework to extract choices’ score from it. Let’s play a little bit with number to do so.

1 - Evaluation process

The choices scoring must be done by evaluating each choice independently, and each choice scoring must follow some rules:

The impact of a choice should be evaluated on each dimension individually and independently.
All the impacts should be represented by a number included in a closed interval centered on zero: [-x ; x]. For example, from -100 to 100, or from -50 to 50. The actual border value doesn’t matter as long as it respects this format.
A negative number is a negative impact of the choice on a dimension, and a positive number is a positive impact of the choice on a dimension.
The closer is the number to 0, the more negligible is the impact. And the closer the number is to a border, the great is the impact of the choice on a dimension.
A choice’s score is calculated using a weighted sum (or average). Score of Choice X = Sum of ( the impact of the choice X on dimension D multiplied by the weight of dimension D), for each dimension.

Once again, let’s return to our concrete example to clarify this:

Concerning the interval, let’s use something simple, and large enough to give some granularity to our impacts: we will evaluate impacts from -100 to 100.

We have two choices: (1) Brownie, and (2) Fruit Salad. Let’s calculate their score independently, going on each dimension individually:

Choice 1 (Brownie)

Dimension 1 (Tasting experience):
- You love Brownie, especially chocolate. Going for this option will definitely have a positive impact on the taste experience.
- Even if it’s not your favorite thing on earth, you love it and would really appreciate it for this dessert. Going for it would have an impact of 70 (out of 100) on the tasting experience. (Where 0 would be no impact, and 100 greatest impact imaginable)
Dimension 2 (Digestive transit):
- It turns out you have lactose intolerance, and Brownie contains some. Choosing it would have a negative impact on your digestion.
- It’s not a severe intolerance, but it remains really unpleasant when it kicks in, so you’d like to score the impact of a brownie choice as -45 on your digestive transit.
Dimension 3 (Healthiness)
- You know Brownie is a lot of sugar, of fat, and is not an actual healthy dessert. Choosing Brownie is a negative impact over your healthiness goal.
- It contains a lot of bad things but it could be worse, you evaluate the Brownie impact as -60 on your healthiness dimension.

Choice 2 (Fruit Salad)

Dimension 1 (Tasting experience):
- A Fruit Salad is definitely not the tastier dessert at all. But you like so it will remain a positive impact on your tasting experience.
- As said before, you like it but it’s nothing incredible, you score the impact as 15.
Dimension 2 (Digestive transit):
- For sure those fruits will not have a bad impact on your digestive transit, but will it have a positive impact? Not really. Appart from this lactose intolerance, you usually don’t have trouble to digest, fruit or not, so the impact is pretty neutral here.
- Neutral impact here implies an impact value of 0.
Dimension 3 (Healthiness)
- Fruits are definitely healthy nutriments, that would be a positive impact on your healthy goal.
- Fruits also are for sure part of the healthiest things you could eat, this impact can be evaluated as 80.

2 - Score calculation

Now that we have evaluated each choice, we have all the number we need to apply the scoring formula and compare them. In order to do so, let’s recap our choices evaluations against the multi-dimensional decision framework:

Now let’s apply the last rule of the choices scoring, by calculating each choice’s score with a weighted sum:

Score1 (Brownie) = 20*70 - 25*45 - 10*60 = -325
Score2 (Fruit Salad) = 20*15 + 25*0 + 10*85 = 1150

3 - Score interpretation

Raw Interpretation

It looks like we found the best choice here! Brownie has the lowest score but also a negative score, meaning it’s not only the worst choice, but a bad choice. On the contrary, Fruits have a higher and positive score.

Bordered interpretation

A thing we can do is adding borders to our score, with the minimum and maximum possible scores.

Max/Min score absolute value is the weighted sum of a choice that would have all impacts at maximum, 100 in our case. Which can be can be simplified as the sum of dimensions weights multiplied by the max impact. Keeping in mind the negative aspect of bad choices, we have:

MaxScore = 100 * (20 + 25 + 10) = 5500.
MinScore = -100 * (20 + 25 + 10) = - 5500.

Which gives us:

Score1 (Brownie) = Negative choice of 325 / 5500
Score2 (Fruit Salad) = Positive choice 1150 / 5500

To get more friendly borders, you can also divide everything by the sum of the weights, which would get the score back between the borders you used while evaluating impacts. This can also be done during the score calculation, by applying a weighted average formula instead of a weighted sum.

In our example this would give:

Score1 (Brownie) = Negative choice of 6 / 100
Score2 (Fruit Salad) = Positive choice 21 / 100

With borders you might be able to interpret better. For example, we see here that brownie is a bad choice, but not that bad in the end, right?

Visual interpretation

We can also think some visual representations to interpret the result of the approach on a decision. A obvious representation is one comparing the scores of each choice against each other, for example:

We could also think of a more detailed representation to see the participations of each dimension in the final score values. This could look like this:

Many other ways are possible and imaginable to visually represent the outcomes or components of the approach. Beyond the interpretation purpose, it can also help gaining confidence about the used values and dimensions during the process.

That’s it!

Hopefully at this stage the approach is clear to you, because trusting the approach is trusting the result. Feel free the jump back on the beginning of the second chapter, and see how some concept could probably make more sense now.

In order to end this reading, I’ll leave you with 3 important things:

Tools to experience this decision making framework on real decisions
The limits to consider about this simplified approach.
The improvement opportunities that the approach brings.

Try out the Multidimensional decision making

If you’d like to play around easily with this approach or give it a try on real decisions, and relief yourself from doing all these steps manually, feel free to jump on the Decisioner tool, allowing you to use the multidimensional decision making model on your decisions:

Decisioner Web App: https://www.weetlab.com/decisioner
Google Sheet version: Google Sheets template

Feel free to visit the tool also for more concrete and complex examples of the approach.

Limits

Most importantly, we have to remember that this is a simplified approach based on a simplified model of the decision making process. Each decision and its context is different, and no model as of now can be able to accurately represent what happens in our brains.
As a consequence of the previous point, we have to bear in mind that this approach does not guarantee 100% reliable decisions outcomes or consequences. It actually also depends on the quality of the approach implementation.
Beyond the accuracy of the decisions outcomes, a limit of this approach is that it can’t adapt to absolutely all decisions, some are too complex to be addressed with this model.
Finally, using this approach can take a lot of time, especially in decisions involving many dimensions and choices.

The assimilation of these limits can be summarized simply: In the end, this approach should only be considered a as tool giving decision insights, it is as an additional opinion to consider in the decision process of challenging decisions.

Opportunities

Bearing in mind the limits, we can on also allow us to think bigger:

The model is really flexible. We were able to get a result and interpretations for our example, but any change in the process could have modified the result. A change in the choices, the involved dimensions, their weights, or the impact evaluation would have completely changed the outcome.
The given example is super specific and simple for the sake of the explanation. But think of it applied to a way more complex decision, involving more dimensions and/or choices. For example, when choosing a new job between a couple of offers, considering dimensions like the salary, the location, the company, the type work, and many more things. Applying this approach in such a decision can bring a lot of useful insights.
Despite the result reliability can not be 100% trusted, it can be improved. The logical complexity of the approach has also a lot of improvement possibilities. For example, the formulas used here to calculate the score are really simple, and could evolved using way more complicated mathematical and statistical concepts.

Thank you

I am no professional writer nor psychologist, only a thinker willing to share his thoughts and vision. In that regard, thank you very much for reading, and I hope these articles enriched your vision.

Through my writings, I express my quest about decoding our reality: Learn how on WeetLab.