One of the nicer things about political risk analysis is that you’re usually protected from accusations of partisanship.

We’re not in the business of saying what is happening (that’s journalists) or what should happen (activists, pundits, and politicians). We are asked only what is likely to happen or what might happen.

Usually, that’s a dry enough (or vague enough) task that we avoid the debates that surround policy proposals or the New York Times’ latest botched headline.

We can usually get away with writing with a “view from nowhere” and stick to the basic facts in our analysis. But there is a flaw in the system. Even if we can write the platonic ideal of political risk analysis, it doesn’t ascend to some analytical Valhalla. It’s handed to someone to read.

What should we do if that person is overwhelmingly partisan - to the point of rejecting facts? What if our efforts to be objective run into someone whose view of the world is so different that they refuse to accept our analysis?

And what if that person is the client or our boss?

Forecasting impeachment

Back in 2017, I remember talking about the importance of tracking impeachment as a political risk.

I was not saying that impeachment was inevitable, but that it was a realistic possibility. This came from a simple calculation.

I put the chance that President Trump would do something impeachable in his term at 80%. He refused to divest from his companies and the Emoluments Clause was already triggering lawsuits. Combine that with his past disregard for following rules and unawareness of governance norms, and it seemed highly likely that he would do something that crossed a line.

The chance that Republicans in Congress would impeach Trump was low, but not zero. There were scenarios in which he would do something so egregious that, simply to jump off a sinking ship, Representatives would impeach him. Let’s put that at 10%.

We also knew that Democrats were favored to win control of the House in the midterms, when the president’s party usually loses seats. It wasn’t certain at the start of 2017, but one could estimate the chance around 60%.

If Democrats were in control of the House, I estimated an 80% chance that they would launch an impeachment inquiry if evidence of an impeachable act emerged. This was based solely on the logic that a party would have both moral and political reasons to impeach, but that the evidence might emerge too close to the 2020 election.

Let’s further assume that, if an impeachable act were committed, the evidence had a 50% chance of emerging during Trump’s first two years and a 75% chance of coming out in his second two years.

That gives us the following calculation:

Impeachment in first two years: 80% chance of impeachable act * 50% chance of evidence emerging * 10% chance of Republican House impeaching = 4%.
Impeachment in second two years: (80% impeachable * 75% evidence * 10% Republican impeachment * 40% chance of Republican control of House) + (80% impeachable * 75% evidence * 80% Democratic impeachment * 60% Democratic control) = 31.2%
Impeachment in Trump’s first term = 35%.

If I were to present this analysis to a client, I would expect tough questions on a number of fronts.

What are the real odds of the midterms? Why is the Democratic inclination to impeach only 80% and not 100%? Should the chance of evidence emerging be different?

All such questions are normal and valid. They make the analysis better and push me towards checking my work for any mistakes.

But what if the client were an extremely pro-Trump conservative, and they flatly rejected that he would do anything impeachable. I present them the above and they say that the most important factor - the 80% chance that he does something impeachable - is fake news?

How can I honestly analyze a situation when the client denies a crucial element and, in this case, would have rejected an analysis that turned out to be broadly correct?

Confronting partisanship

This example might be extreme, but you’ve probably witnessed some version of it.

It could be about an estimation of effects, like a client arguing that the 2017 tax reform package would lead to 4% growth rates when you argue the impact will be far less.

It could be office politics, like a boss saying that your operations in Brazil are safe because he once went on vacation there and knows the country.

It could even be a simple non-partisan observation, like forecasting that Jared Kushner would be a disaster in government but the client went to Harvard with him.

How do you convince the audience that your analysis is correct without backing down on the integrity of the analysis?

There is no one right way. But here are four approaches that have proved themselves useful.

1. Opinion launder

If it becomes a standoff between the analyst’s opinion and the client’s opinion, the analyst will lose. If the client doesn’t respect the analyst’s position, it’s no longer about the content but about the power dynamic between who is paying and who is being paid.

In that case, you need some reinforcements.

Rather than saying “I think the Ukraine saga was impeachable,” write that “As Mitt Romney, Republican candidate from 2012, said when he voted for impeachment…”

Maybe the client still doesn’t respect the argument, but it’s clear that it’s now not just what you think. It’s you plus someone that he may have voted for 8 years ago. The facts are the same but the impression is different.

2. Show your work

It’s very easy to dismiss a claim wholesale. If I had said in 2017 that there was a 35% chance of Trump being impeached, a client might well have thought that was ridiculous and told me I was wrong.

But if I broke that 35% down as I did above, it would lead to a much different conversation. I could ask where the client disagreed.

If it was the 80% chance of an impeachable offense, I could talk about what counts as impeachable, the track record of what he’s done, the likelihood that an official would internally push back on something, etc.

If it was the chance that the evidence would come to light, we could talk about whistleblower statutes or the pace of investigative journalism.

Either way, it is much easier to argue those points, where data is readily available and the logical connections shorter, than the end product. It might not be enough to diffuse the most stubborn partisan, but it helps with those who have kneejerk reactions but little behind that.

3. Crowdsource the analysis

Opening up analysis to multiple perspectives has a lot of inherent advantages. It can filter out biases, bring in information quickly, and weight different sources appropriately. A market for gold in the 1860s has even been used to identify how likely the Union was to win the Civil War on a given day because it aggregated information from across battlefields and Washington.

But crowdsourcing also has the advantage of protecting your analysis from accusations of partisanship.

While you might be accused of overinterpreting one piece of information, it’s harder for a client to accuse you, your colleagues, external experts, some interns, and your boss from all conspiring to overinterpret the same information in the exact same way.

The downside is that groupthink can reinforce bias, so crowdsourcing should only be used if it can be designed to help the quality of the product. But if it can, it can have benefits beyond its immediate analytical purpose.

4. Build different scenarios

Scenario planning is a useful technique in its own right, and should be used more than it is already, especially as COVID-19 makes everything about the future highly uncertain.

An added benefit is that it can deflect client partisanship by presenting multiple alternative futures, so that they don’t get fixated on the flaws they perceive in one.

If your client thinks that there’s no way that Trump commits an impeachable act, build out scenarios that include whether he does or doesn’t. They may argue that he won’t, but now, rather than disputing the exact odds of such an act, you have to only say that it’s possible.

People have a much easier time acknowledging that something is possible rather than that something is probable. The debate about what happens after impeachment can take place without immediately triggering partisan defenses.

Conclusion

Ultimately, the most important part of your analysis is the analysis. There’s no substituting for good work.

But as most people know, sometimes good work isn’t enough to convince everyone. Being strategic in defusing instinctive partisanship can save a report from being tossed aside and instead be put to good use.

Of course, if your boss or client is relentless in denying well-founded analysis and insists on seeing it their own way, the best course of action is probably to walk away and avoid doing business with them. There’s only so much you can do to help someone who refuses to acknowledge what’s in front of them.

For more advice about how to design an analytical process protected from partisanship, get in touch.

How to stay non-partisan when your clients (or bosses) aren't