Faro Card Game

Removing Bias From Big Decisions

Researchers tell us that many decisions are emotional, then rationalized later. This leads to many “bad” decisions. Here is a process that removes much of the bias, while taking into consideration multiple viewpoints and “gut feel”.

[Note: for discussion purposes, I am assuming that the board has set up a committee to address an issue. It could be a manager asking anyone for a recommendation.]

The order of the steps is very important.

Step 1: State the problem.

I am continually amazed at the number of times a board or committee is facing a decision, but cannot actually make a simple, clear statement of the problem to be resolved. If a committee is formed to face a decision, the board should give clear direction, stating the problem and the deliverable. For example: “Determine the feasibility of [whatever] to resolve the issue of [whatever]. Submit a written recommendation and plan to the board for action.”

Of course, if you chair a committee and have not been given a problem statement, you must create your own. Bias can will enter here. He who frames the problem usually gets what he wants.

Step 2: Set the Criteria.

This is a critical step and it must be completed here, before going forward.

From the point of view of the board, or other authority, you want the committee to come to you with the “best” recommendation. But, how do you know the recommendation is the best if you cannot state the approval criteria? You don’t. You don’t even know if the solution is acceptable. For example, I continually see boards approve a budget without ever having an idea of what a “good” budget looks like. They just approve it, never having provided guidelines.

The goal of the committee is to get approval from the board, without objection. This is called “completed staff work.”  This means that the committee says to the board, “We recommend [this course of action], having considered [list of alternate courses of action], for the following reasons: [list of criteria].”

If the entire process is followed, there can be no objection to the process, only objections to the criteria weighting.

Every potential objection is a criteria. If someone raises a valid objection, that means that a criteria was missed.

It helps to run a brainstorming / affinity exercise at this level to flush out the potential objections / criteria.

Then it is very important that you define each criteria in detail. Create a written Criteria Dictionary. Don’t assume that you know what you meant at the last meeting. You won’t.

Step 3: Develop Options.

This is where you brainstorm potential recommendations, then use an affinity exercise to narrow the options. Make sure each option is well defined. Do not pass judgement on options at this point.

Step 4: Determine the Direction of Impact of Each Criteria on Each Option.

Ultimately, in Step 5, we are going to calculate a figure of merit, but the criteria are not of equal importance and do not have equal impact.

There are two parts to this step. The first is to determine the direction of the impact of each criteria on each option. The second is to determine the relative weight of each criteria relative to the other criteria.

Step 4A: At the simplest level, each criteria is positive, neutral, or  negative with respect to each option. For example, suppose the decision is to choose what truck to buy. One option is to buy a Ford F-150 pickup and another option is to buy a Mack truck and trailer. The cost criteria is relatively favorable in the first option and unfavorable in the second, while the capacity criteria is the other way around. The direction of each option is usually obvious and can be decided by group consensus, as long as everyone clearly understands the definition of each option and each criteria. Option Criteria MatrixThis is recorded in a spreadsheet with criteria along one axis and options along the other. Each cell has a 1, 0, -1 to indicate a positive, neutral, or negative impact. You could use a wider scale with more gradations, but I have never found this to be needed.

Step 4B: There are several ways to determine the importance (weight) of each criteria with respect to the other criteria. This, too, is critical. This is where the real decision is made. This is where everyone’s opinion counts.

In the truck example, there are several criteria: capacity, acquisition cost,  range, fuel economy, operating cost, ease of loading and unloading, color, driver licensing criteria, reliability, ease of repair, etc. The logistics department would like more capacity, the financial department would like lower acquisition cost. Opinions differ on what is important, but each opinion must be considered.

My preferred method of setting the weights is to give each participant 100 votes and let each participant, individually and privately, distribute their votes among the criteria. They can distribute them any way they liked: all on one criteria, spread equally among all criteria, or, more likely, unequally.

These votes are then totaled for each criteria. Each criteria total is divided by the grand total of all votes cast. The result, when multiplied by 100,  is the percentage weight of each criteria.

The weakness in this approach appears when one department  has a disproportionate number of participants. In this case, you can group participants by department and generate department-level weights which are, in turn, used to calculate the overall weights. The decision to group participants should be made prior to the vote to prevent the possibility of bias.

Another popular weighting technique to use a series of pair-wise comparisons of each criteria against each of the other criteria. However, pair-wise comparisons can lead to an indeterminate circular answer such as a>b>c>a. I prefer the multiple vote method, which always yields a determinate result.

Step 5: Calculate the Figure of Merit.

Calculating the Figure of Merit is now routine. Multiple each of the cells in the Step 4a matrix by the the weight of its criteria. Then add the result across to get the Figure of Merit for each option. [If you have a spreadsheet whiz available, they can do this quickly using a SUMPRODUCT function.]

Step 6: Rank the options and test for reasonableness.

The “best” solution has the highest Figure of Merit. It is a good idea to do a some sensitivity analysis: What if we scored the Step 4a impacts differently? What if Joe from Finance had been able to vote and what if he put all his votes on the cost option?

I have never seen these type of changes move the score enough to change the ranking, but the analysis will end any speculation.

Step 7: Prepare and present the Recommendation

The final step is to prepare the deliverable. You want to say “We recommend [this course of action], having considered [list of alternate courses of action], for the following reasons: [list of criteria].” 

I like to start with the “Here it is. Ain’t it great. However” approach. Remind them of the problem you were asked to solve. Make the recommendation statement above. Describe why this is the right recommendation (the highest weighted criteria). Tell them of any risks to this recommendation. [Usually risks come from any assumptions you made.] Tell them what they must do (e.g. fund the project, assign an implementation team, etc.) Give them what they need to do it, such as the exact wording of the motion to approve, or the capital request, or the draft charter for the implementation team, etc.

There may be questions. There may even be disappointment. But, as long as you follow this process, you will have made the “best” decision.