Sunday, November 30, 2008

Inference #2 – Cause

Note – this series started 9/26/2008 (or 350 B.C.E., in as much as it is a continuation of work started then).
The first of the six inferences was covered last week, when we started with the strongest inference – Inference of Example. It is the strongest because examples are in the real world. The next strongest inference is this week’s topic – Inference of Cause (or causal inference).

Causal inference indicates that one event has influence over another. It also indicated that the influence cannot be directly observed.
Causal inferences are used to:
- Predict events (Reducing the number of nuclear weapons causes a reduction in the likelihood that they will be used)
- Relate means to ends (Studying causes good grades)
- Explain paradoxes (What causes a candidate to get twice as many electoral votes as another if they only get 2% more of the popular vote?)
- Assign responsibility (What causes the proposition we thought would win to actually lose?)

For the sake of this article, I am not going to include the statistical methods that are available to explain causation, except in simple terms. They are powerful, and you should know them and use them if they are appropriate for the case with which you are concerned. There are many sources for that information and not many for the following, so I will devote this space to the lesser-known ways.

Let’s start with a work example: You claim “Well, I can expect to start having trouble with Sharon. “ Your co-worker asks “Why do you say that?” You answer “I just criticized her work.” And the co-worker asks “So why is that a problem?” and you say “Criticizing people’s work causes them to start trouble for you.”

In the case I created above, my claim (I expect trouble with Sharon) is supported by evidence (my criticism of her work) and connected via a causal inference (criticism of co-workers causes trouble). I am using it to predict future events. If my audience believes that criticism of co-workers usually causes trouble, they will accept my claim; if they don’t, they won’t accept the claim and I will have to convince them. Let’s explore how causal inferences should be evaluated, both before it is made and after. What are the reasonable evaluations for causal inferences like the following:
- Stock prices dropped because America is no longer an economic power.
- The river overran its banks due to the rain.
- Low birth weight reduces the likelihood of college admission.

Statistical approaches would have us experiment. We would hold all factors in two samples the same, and then vary one of them. If the output varies, then the factor we varied could be said to cause the variance. This is a method popularized by John Stuart Mills and is a great approach in a laboratory, but becomes impractical as we start to look at many business issues.

Rhetorical approaches would have us (in a first step) identify how some factor COULD be the cause and, in a second step, have us explain why it SHOULD BE CONSIDERED the cause. In the law, the first step is used to show means and opportunity, and the second step to show motive.

In testing casual inferences, it is critical to determine that one thing indeed influences another, and isn’t merely being attributed. For instance, the stock price question that I raised above is too complex to be so simply explained away. Stock prices rise and fall based on many things, and there most certainly are many factors that can cause price fluctuations. In fact, a strong argument can be made that stock prices frequently drop after a period when there has been too much speculation and the market needs to “normalize” in order to get healthy. So if you plan on attributing lowered stock prices to a single cause, then you need to prepare more than a single reason and bring lots of evidence.

One must evaluate whether there are multiple causes and not just one. For instance, while it is true that rain can cause a river to overrun its banks, so can a dam. If a car drove off the road and into the river on a rainy night, it may not be the rain that caused the river to overrun its banks. One must also remember that one cause can have multiple effects, for instance turning off the water main in order to stop a pipe from leaking and spilling water on the floor. It will work because there is certainly a causal relationship between the presence of water in the leaky pipe and water on the floor, but there are other effects that shutting the water main has that are perhaps unwanted.

One must evaluate whether there are common causes underlying the supposed cause and effect relationship. For example, I recently read that American babies with low birth weight tend not to grow up to go to college. Reading this, I immediately thought that a common cause underlying both conditions (low birth weight and failure to go to college) is poverty. In checking, I found that the author had misinterpreted some statistics and “bent them” to try to show why it is important for mother’s to try to achieve higher birth weights!

One must evaluate whether one is confusing temporality with causality. In other words, just because something happens BEFORE something else, doesn’t mean it is the CAUSE of it. Just because I correct someone’s work and later I start having trouble with them on the job doesn’t mean that the two events are causally related. It could be that, once I correct them, my attitude changes about them and they pick up my attitude. Or that the reason I am correcting them is that they are overwhelmed by the work and they would rather work somewhere else and eventually this starts to manifest as “trouble” to me.

There are many fallacies associated with causal inference. Here are two:
- The Post Hoc fallacy – (if something occurs BEFORE an event, then it means it CAUSED the event.) This fallacious thinking is used all the time. When you hear it, question it.
- Appeal from Ignorance – (Well, if the rain DIDN’T cause the river to overrun its banks, then what did?) This appeal is meant to put the listener in a position of either offering an alternate explanation or accepting the one offered. Remember there is a third choice – that something else caused the event and you don’t know what it is. You needn’t ever buckle to the appeal from ignorance.

Insist on great business results! Go to Pathfinder Communication

Sunday, November 23, 2008

Inference #1 - Example

Since we started this series on October 26, we have:
- Identified the underlying resolution in a controversy
- Categorized the resolution into one of four categories of claims
- Learned four ways to identify the issues that need to be addressed to support the resolution
- Learned what kind of evidence we can accept, and how to examine it

Today, we will begin to cover Inference – the connection between the evidence and the claim.

At first, it is not obvious that there is an inference at all. I mean, the evidence is either supportive of the claim or not, right? And if it supports the claim, it is good evidence and all good evidence is equally good, right? Well, no – not right. Allow me to explain.

The way that the evidence supports the claim is a very important factor in rating the strength of a position. There are six different inference types; from strongest to weakest they are: Example, Cause, Sign, Analogy, Narrative, and Form. Today, I will go over Inference of (or Inference from) Example.

Let’s say that I made a statement that “members of the National Association of Communicators (the NAC) generally ask intelligent questions”. This is a claim of fact and you all know what questions to ask (the last two newsletters were all about finding the right questions), so let’s talk about some different evidence I might offer and the difference in strength. Consider the following five evidence statements:

1-When I spoke to the NAC, I was asked many questions by one of the people that attended the meeting and they were all intelligent questions.

2-When I spoke to the NAC, I was asked questions by many of the people that attended the meeting and they were all intelligent questions.

3-After speaking to the NAC, I received email from many of the members that didn’t attend the meeting and they all asked intelligent questions.

4-After speaking to the NAC, I received email from each of the members that didn’t attend the meeting and they all asked intelligent questions.

5-After speaking to the NAC, I received email from each of the members of the NAC and they all asked intelligent questions.

These are all inferences of example meaning I have examples of actual questions being asked from actual members and (presuming that we agree on what an intelligent question is) we can review them. The inference of example is the strongest type of inference, but are they all of equal strength?

Evidence statement #1 shows that I was asked many intelligent questions when I spoke to the NAC, but is it enough evidence to show that “members of the National Association of Communicators (the NAC) generally ask intelligent questions” as I have claimed? Well, not on its own. It simply shows that ONE member asks intelligent questions. Can we agree to extend that generalization to all members? I think it would take more evidence to convince me.

Evidence statement #2 shows that I was asked many intelligent questions when I spoke to the NAC, and further that they came from many of the attending members. Is it enough evidence to show that “members of the National Association of Communicators (the NAC) generally ask intelligent questions”? Still, no. The statement only talks about members that attended the talk. Even if EVERY member that attended asked intelligent questions, it may be an insignificant percentage of all of the members. We would have to assure ourselves that the number and variety of attendees was representative of all the members – either by further examination of this evidence OR introduction of OTHER evidence. Maybe a few other speakers have found the same kind of intelligence when they spoke to the NAC. I think it would take more evidence to convince me.

Evidence statement #3 shows that I was asked many intelligent questions AFTER I spoke to the NAC, this time by some members that didn’t attend. Is it enough evidence to show that “members of the NAC generally ask intelligent questions”? I don’t know. The statement only talks about members that DID NOT attend the talk. It fails to be compelling evidence for the same reason #2 failed - it may be an insignificant percentage of all of the members. If coupled with #2, however, we have some compelling evidence – a sample of members that attended and a sample of those that didn’t attend asked exclusively intelligent questions. But if not coupled with #2, it would take more evidence to convince me. NOTE that the emails serve as DOCUMENTED evidence, which means that we no longer are relying on my credibility as evidence as we did in scenarios 1 and 2.

Evidence statement #4 shows that I was asked many intelligent questions AFTER I spoke to the NAC, this time by ALL of the members that didn’t attend. Is it enough evidence to show that “members of the NAC generally ask intelligent questions”? I don’t know. If it is a significant percentage of the membership (that is, if most members didn’t attend and are now emailing me intelligent questions), it is pretty sure that these NAC guys are a sharp bunch.

Evidence statement #5 shows that I was asked many intelligent questions AFTER I spoke to the NAC, this time by ALL of the members whether they attended or not. And the evidence is documented. This is locked up tight.

This is all about the inference of Example, in which I had examples of questions asked by NAC members. Early scenarios were less substantial and later ones were more so. In the first four scenarios, I was describing specific examples and asking whether or not we could reasonably draw a conclusion about the population that the examples came from. This is called generalization and there are two types - statistical generalization and anecdotal generalization. In both cases, the keys are representativeness, sample size, range of samples, and absence or present of counterexamples.

In the fifth scenario, where we had documented evidence of intelligent questions from ALL if the members, we needed no other evidence. In most cases, this is impossible. Seldom do we have this kind of complete enumeration because, long before we had received the email from each and every member, we would have inferred that they were all intelligent. Certainly in business, it is rare to know why every single customer buys your product either because it is not possible, or it is prohibitively expensive.

Another way to use examples is classification – that is making a statement about an individual based on their relationship to an accepted generalization. For instance, let’s say that after you have accepted that the members of the NAC ask intelligent questions, you run into a friend that mentions that he is a member of the NAC. Would it be appropriate to assume that he asks intelligent questions? Of course.

Two fallacies associated with inference by example the “Hasty Generalization” (Generalization or classification is made without enough data to support conclusion) and “Guilt by Association” (you associate with thieves, therefore you must be a thief). We will explore many logical fallacies in later articles.

Insist on great business results! Go to Pathfinder Communication

Monday, November 17, 2008

What Makes You Say That?

Some of you have been practicing the last few techniques I’ve posted – GOOD FOR YOU. Communications is something that you practice frequently so that when you need it, you are confident in your abilities. One friend likened it to CPR in this way – if, when you need the skills, you have only an academic understanding and are under-practiced, you may choose not to use them for fear of doing it “wrong”. I suppose that is true, and I salute you that are practicing and sending me email.

A common question asked has been “What if one’s emotions boil over during a talk?” Some conversations can have a high emotional content, and by using the right skills the CONVERSATION needn’t become emotional – needn’t cause defensiveness. Right now, we are exploring the model for a straightforward business discussion and examining its parts and protocols as if neither side becomes emotional while the topic at hand is being discussed. In later lessons, once we have some mastery of these skills, we will examine how to conduct a more “charged” discussion.

So far, we have:
- Identified the underlying resolution in a controversy
- Categorized the resolution into one of four categories of claims
- Learned four ways to identify the issues that need to be addressed to support the resolution

Today we will learn what to do with those issues, and what work they are expected to do. We will be building on the examples we have been using all along (“Far more kids are engaged in cheating in school than the teachers think”). For the moment assume that the key questions (issues) can be asked straight out, without any “shaping” as we might if there were some emotional charge associated with them.

Let’s agree to a limited set of issues that would be required to support this claim.
- How many kids are engaged in cheating?
- How many kids do the teachers think are cheating?
- How many more would be “Far More” than teachers think?
- Are that many more kids cheating?

We would ask the claimant (that is the person making the claim) one of these questions and expect some kind of answer. The answer needs to be based on something and we call that basis “evidence”. There are three kinds of evidence that we will listen for; Objective Evidence, Social Consensus, and Credibility of the Speaker.

- Objective evidence is anything that can be examined further. Statistics, documents, testimony, opinions, and examples are all types of objective evidence. Evidence of this type can be examined and questioned until it is accepted.
- Social consensus is a belief that can act as a fact. Common knowledge, shared values, previously established facts, and stipulations are all types of social consensus. Evidence of this type must be agreed to by both sides before it has any real power.
- Credibility can substitute for evidence and we allow it to do so frequently. If the person has a good track record for being honest, is an authority on the subject, can point to a clear basis for their conclusion, does not have a bias or vested interest, and if other credible sources agree with them, it is very likely that they are credible.

Using our example, let’s imagine that we’ve asked the claimant “How many kids are engaged in cheating?” Let’s imagine that they respond with “Practically every single one of them”. We ask “how do you know that?” and their response is, “I just know it”. They are asking you to accept their credibility as evidence. So look back at the criteria for credibility; track record, expertise, clarity of basis, bias, etcetera. You would have to ask questions about those items before you could accept their credibility as evidence.

What if they would have responded differently? Let’s imagine that they respond with “Practically every single one of them”. We ask “how do you know that?” and their response is, “Everyone knows that”. They are now asking you to accept Social consensus as evidence. So look back at the criteria for social consensus; common knowledge, shared values, previously established facts, stipulations, etcetera. Did you stipulate (or are you willing to stipulate) that ‘practically every single student’ is cheating? If not, could you find it to be a widely held belief? Is it a previously established fact? If not, it isn’t suitable evidence and you would reject the claim pending stronger evidence.

Finally, let’s imagine that they respond with “Practically every single one of them”. We ask “how do you know that?” and their response is, “I have a survey here that shows it”. They are now asking you to accept statistics as evidence. This is a form of Objective evidence. So what does this mean? It means you can examine the data further if you wish. If you don’t accept it at face value, you should ask a few questions about the survey to see if it is relevant. Is it from the same population of students that we were originally talking about or a different population? Is it from a prison school located in Madagascar in 1760?? Is there a sufficient sample size and a sufficient variety and range of respondents? Can it be fact checked? Can the source data be reviewed?

These rules can be applied as stringently as is needed to guide the course of the discussion. For instance, if this was a discussion about something very risky (“should we launch the shuttle in cold weather?”), you would have a higher standard of evidence than for something that was low risk. If you have done some independent checking and you feel you know about how many students are cheating and the claimant has a similar number, you may accept it as Common Knowledge and move on. If your numbers are significantly different, you may accept their number based on their credibility or survey. Or you may reject it until a number is found that is mutually agreeable based in the evidence.

You should understand now that in order to engage in a critical discussion, the tools used are questions and answers. The answers are called ‘reasons’ and the process is called ‘reasoning’ – exchanging reasons back and forth until we can agree on one. You should also recognize that if YOU are the one making the claims, you would be well-advised to use the Topoi to analyze the issues surrounding your claims and be prepared with some compelling evidence. This is how you develop that good track record required for using your own credibility as evidence.

Insist on great business results! Go to Pathfinder Communication

Monday, November 10, 2008

Aristotle’s Issues

In the last few weeks, we have gone over the following:
· Finding the resolution in a controversy
· Finding the issues inherent in the controversy (3 out of the 4 ways)
· The four types of claims that are made to answer the issues raised
The resolution and issues both ‘flow’ from the controversy, the claims flow from the issues and are analyzed to see if they support the resolution. Graphically, it looks like this:


I mentioned a fourth way to formulate the issues for a given controversy. This method was described by Aristotle. As I mentioned before, I always try to develop issues using two or more methods and this method (the Topoi) is the one I rely on most.

Topos is a Greek word for ‘place’. The plural is topoi and refers to ‘places in the mind’ where questions are kept. There is a story that goes with this, but for now just recognize that this is a method to formulate issues based on the type of claim being made. Again, the four types of claims are:

1 - Claim of Fact – presented as a plain factual statement like “Far more kids are engaged in cheating in school than the teachers think”
2 - Claim of Definition – presented to offer a definition or characterization like “The kid wasn’t cheating; he was just harmlessly cutting a corner”
3 - Claim of Quality – presented to make a value judgment about something like “Those teachers are incompetent”
4 - Claim of Policy – presented as a suggested course of action like “They should allow the kids to work together and share ideas like they’ll have to in the workplace”

For a claim of fact, there are two topoi; 1) ‘How would we know if the claim is true?’ and 2) ‘Have those conditions been met?’ So going to our example (“Far more kids are engaged in cheating in school than the teachers think”), some issues suggested by those topoi are: How many kids are engaged in cheating? How many kids do the teachers think are cheating? How many more would be “Far More” than teachers think, and are that many more cheating?

The degree to which you can answer those questions indicates the strength of your position. NOTE – (this is important) - It is usually NOT PRACTICAL (or even possible) to answer the issues with absolute proof. If it were possible, then there would NOT be a controversy! There are three kinds (and a few subtypes) and we will talk about how you weigh supporting evidence in the coming weeks. For now, just know that there is seldom “incontrovertible evidence” for anything that is controversial in business.

For a claim of definition, there are three topoi; 1) ‘Is the definition relevant?’, 2) ‘Is the definition fair?’, and 3) ‘How do we choose between competing definitions?’ So going to our example (“The kid wasn’t cheating; he was just harmlessly cutting a corner”), some issues suggested by those topoi are: Does it matter if we share a common understanding of what ‘cheating’ is? Is it important not to cheat in school? Is there a difference between ‘cheating’ and ‘cutting corners’? Would the activity that the kid was engaged in constitute what we consider cheating? How are ‘cheating’ and ‘cutting corners’ different? Can we agree on enough differences between ‘cheating’ and ‘cutting corners’ so that we can differentiate between them?

You may see why the topoi are my favorite method – it offers guidance in creating issues. This was always a problem for me when I started out in business. People would say something that didn’t seem quite right, but I never knew what to say to challenge them. Learning the topoi helped me a great deal.

A claim of quality is similar to a claim of definition. It is presumed, however, that we agree on the definition of the quality. If not, then we would handle it like a claim of definition until we agreed on the definition of the quality and then proceed. For a claim of quality, there are three topoi; 1) ‘Is the value true?’, 2) ‘Is the value generally relevant or specifically relevant?’, and 3) ‘How do we choose between competing values?’ So going to our example (“Those teachers are incompetent”), some issues suggested by those topoi are: How do we judge if the teachers meet our definition of incompetent? Are they incompetent in all things, some things, or just in some specific thing that may not be relevant to the resolution? If we characterized their incompetence as ‘stupidity’ or ‘laziness’, would we feel that those (competing) values better describe our perception of them?

Claims of policy are common in business, because we are constantly trying to correct problems or create opportunities. For a claim of policy, there are four topoi; 1) ‘Is there a problem (reason to change from the current state)?’, 2) ‘What is the problem?’, 3) ‘Does the proposal solve the problem?’, and 4) ‘Is the proposal better than the problem?’ So going to our example (“They should allow the kids to work together and share ideas like they’ll have to in the workplace”), some issues suggested by those topoi are: Is there a problem with kids competing? Is there a problem with kids competing in school differently than they will at work? Is competition the problem? Is there enough collaboration going on so that kids learn both working styles? Does allowing kids to collaborate keep them from cheating? Are there other impacts (expense, curriculum, facilities, teacher training, etc.) that would have to change in order to permit collaboration? Will it be worth it?

Now, you know a lot more about questioning claims made by people; how to categorize those claims, and how to define the key issues needed to resolve them. As always, you need to practice these techniques as often as possible until you are doing so automatically. It is critical that you practice when the stakes are low so that you are comfortable if the stakes are raised.
Next we will learn how to answer issues and by extension, learn what to expect for answers when someone else is responding to your questions.

Insist on great business results! Go to Pathfinder Communication

Monday, November 3, 2008

We Have Issues

Last week, we started to explore some details in having a critical discussion. We talked about defining the resolution of the discussion (that is, the main claim) and expressing it. I asked you all to practice doing so. While I am certain you ALL practiced, I will cover what you may have encountered in a brief dialog before moving on to identifying issues (the questions that flow naturally and must be answered in fully exploring the resolution). Here goes:
Tom – “I am so upset with my kid. He got in trouble at school!”
Me – “Sorry to hear that. What was it about, if you don’t mind my asking?”
Tom – “He has a terrible history teacher. He tries hard, but can’t really follow all of the details for this class.”
Me – “So how did that get him in trouble?”
Tom – “He borrowed some ideas from another kid’s homework, just to get some ideas. They’re making a mountain out of a molehill.”
Me – “How did they know he did that?”
Tom – “Well, he copied the other kid’s words exactly! All the kid’s copy from each other – but he got caught”
Me – “So, you’re saying that what he did was alright?”
Tom – “No…of course not. I just don’t think he should be singled out”
Me – “So, you’re saying that he shouldn’t be punished?”
Tom – “No – I am just saying that cheating in school is common and punishing my kid won’t help”
Me – “So, you’re saying far more kids are engaged in cheating in school than the teachers think”
Tom – “That’s exactly right”

In this conversation I had to try three times to actually gain agreement on what the resolution is, but I finally made it. The resolution is “Far more kids are engaged in cheating in school than the teachers think” and that gives us a starting place for a critical discussion.
The first thing we may ask is: do we ourselves know this to be true? Do we KNOW with certainty that “Far more kids are engaged in cheating in school than the teachers think”? I DON’T know that so I may want to test the truth of that claim a bit. In order to test a claim, one must identify the issues underlying it.

The resolution is a special claim. I called it the “main claim” earlier. Even though it is the main claim in a controversy, it is still just a claim. There are four kinds of claims. That’s right – In all the controversies you have ever been involved in, there have only ever been four kinds of claims in play. Our resolution (“Far more kids are engaged in cheating in school than the teachers think”) is called a “claim of fact” – a statement one makes that is presented as a plain fact. We will examine all four kinds as we progress, but for now we will just define them:
1 - Claim of Fact – presented as a plain factual statement like “Far more kids are engaged in cheating in school than the teachers think”
2 - Claim of Definition – presented to offer a definition or characterization like “The kid wasn’t cheating; he was just harmlessly cutting a corner”
3 - Claim of Quality – presented to make a value judgment about something like “Those teachers are incompetent”
4 - Claim of Policy – presented as a suggested course of action like “They should allow the kids to work together and share ideas like they’ll have to in the workplace”

The underlying issues for each kind of claim can be found in four ways:
1 – Breaking down the key words and phrases in the claim, and asking questions about them
2 – Reviewing the context of the claim, rather than simply breaking down the phrases
3 – Reviewing all that we both agree on about the subject, and questioning that which we do NOT agree on
4 – Using an ancient model (called the Topoi) to determine the underlying questions based on the kind of claim being made.

We will illustrate method number one, two, and three in this article, with special attention to the fourth next time.

In breaking down key words or phrases, we would view the resolution in parts and ask questions about those parts. See the table below.















Key word or phrase


Issues


Far more kids

How many kids? How many more is “far more”?

are engaged in cheating in school

What do we mean by engaged in cheating? What does one
have to do to be “engaged in cheating”? What does it look like? What exactly
is “cheating” in school? Does this exclude cheating in other places? Is it
only in academic subjects, or can one cheat in art or physical education?

than the teachers think

Do we know how many kids the teachers think are
cheating? Are the teachers in a good position to know? What would make the teachers
think that kids are cheating?




If we were to examine the context, we might ask questions about the pressure put on kids to do well in school; the erosion of ethics in society in general; the temptation of a “quick fix” approach over longer term and deeper education strategy; or the incongruent emphasis between individual achievement in school and team achievement at work.

In reviewing all that we agree on within the topic, we might stipulate that kids do cheat and teachers can’t be certain as to how many do it, and that cheating means turning in work for credit for which you were not the original creator. We would still have discuss the teachers’ estimate of how many cheaters there are and how may cheaters there actually are, and would the difference in the two numbers qualify as “far more” or not.

When I review a claim I like to use at least two of the four methods, which ever seems more appropriate at the time, in order to make sure I have a good perspective of the resolution. Practice it this week – first, question until you have the resolution defined to the other party’s satisfaction and then develop all the salient issues you can. You don’t have to ask them all (I never do) but identify as many as you can and see if you can ask the most important ones. You will never be at a loss for something to ask about a claim again!

Next Week – the topoi!

Insist on great business results! Go to Pathfinder Communication