Degrees of agreement

On Fri, Dec 17, 2010 at 21:26:43
Len asked this question:

My question has to do with language and in that sense it could be a linguistics or a philosophy of language question. Of the two, I'm not really sure into which category it falls.

If you agree or disagree with a statement, it seems to me this is an absolute. However, on many psych tests employers use these days for candidates seeking to fill the open position, they give choices of 'agree,' 'strongly agree,' 'disagree' or 'strongly disagree. For example; if the the statement is 'The sky is blue,' I can either agree or disagree with the statement. How could I further agree or disagree about the state of the color of the sky or any other statement for that matter. If you and I both disagree, how could either of us disagree 'more' than the other? Herein lies my question:

How can you assign an adverbial quantifier to something that I believe is an absolute? I'm pretty sure I'm not the only person who thinks this way so could tell me the difference between agree and strongly agree?

This is a fascinating question in the philosophy of language. Somewhere (I can't remember where) Michael Dummett raises the possibility of a speech act similar to assertion, where the speaker is less than fully confident about what they are saying. I think the term he used was 'probabilistic assertion', an idea he associated with Michael Polanyi. I remember long ago discussing this with my thesis supervisor John McDowell, who was roundly dismissive of Dummett's proposal.

Consider weather forecasts. People complain when the weather girl says, 'It will be fine tomorrow,' when she knows damn well that there is only a 70-80 per cent probability that it will be fine tomorrow. (I'm talking about BBC weather girls who've studied meteorology and actually know what they're talking about, on other TV stations they just read a script.) In the discussion I made the point that the context (a TV weather report) makes it clear that when the weather girl makes an assertion about tomorrow's weather, she isn't doing what we normally do when we make assertions. It isn't necessary for her to quote the probability figure, or express some degree of doubt about what she is telling us. It's understood.

But that's just the thin end of the wedge. (I think that this was McDowell's objection.) We would have to admit a whole family of speech acts, speculative assertion, tentative assertion, cautious assertion, confident assertion, emphatic assertion. And that just seems wrong. To make an assertion is to aim at truth. There are only two possibilities, you aim at truth or you aim to miss (i.e. you tell your audience a deliberate lie). It's understood that failure is a possibility. But you can't include a rider to that effect without destroying the whole point of this language game. Or, if not, then the rider adds nothing to what you've already said, the force and semantic content of your speech act.

However, my intuitions tell me that there is a point in the way these questionnaires are constructed, and the options they give. To extract this point, we need to do quite a bit of of work in a number of related areas: game theory, probability theory, the analysis of knowledge from testimony, as well as philosophy of language. Just to give a sense of the complexity involved, here's a short parable:

I am having a pleasant stroll in the hills around Athens with my three companions, Parmenides, Zeno and the young Socrates. Somehow, we've managed to get lost. I'm sure we passed that broken tree half an hour ago. We reach a point where the path forks three ways. 'Which we should we go?' I ask. Zeno scratches his chin. After what seems like an eternity he says, 'It's not right and it's not straight ahead, so I think it must be left.' 'No, no!' shouts the young Socrates waving his pointing finger enthusiastically, 'We have to go right, I'm sure of it!' Parmenides scowls. He stares straight ahead and nods. 'That is the way,' he says in a quiet tone.

Which way do you go?

I don't think that there's any doubt. I would follow Parmenides, I'd go straight ahead. Zeno isn't completely sure, so we can discount him. Socrates' wild gesticulations aren't convincing. Whereas Parmenides impresses us with his authority. He doesn't need to make a fuss about it. He knows.

There's a discussion of the connection between knowledge and authority in my Answer to Demetreus. If you think about it, there could not be a linguistic device which qualified a statement in a way which reliably gave the hearer information about the speaker's authority to make that statement, the credence one should place on it. And yet, we make these kinds of judgements all the time. The reason why we couldn't have such a device is that people aren't always the best authority on how credible an authority they are.

However, there is no objection in principle to introducing new devices into the language game, provided they have a use. Indeed, arguably, we already have such a device in the various ways and means available for conveying the strength with which you hold a belief or opinion. The finesse here is that the 'measure of strength' isn't like assertion, it doesn't function in the same way as a speech act, nor does it function as a qualifier of the speech act. It's information which you give out, more or less voluntary, of the same order (or at least closer to) the information you give out when your face blushes, or you tremble, or your features contort in anger. It is almost impossible to imagine what human life would be like if these features were absent.

When you tick the boxes (and I fully accept, sometimes it doesn't seem to make a lot of sense when you are asked whether you 'agree' or 'strongly agree' to a particular statement which is just plain true so far as you are concerned) you are giving out information which will be processed to yield a result. A numerical scheme is applied, somewhat like the various proposed preferential voting schemes for proportional representation. In a similar way to preferential voting, knowing this gives you some additional measure of control over how your application will be assessed. And the people who designed the form, know that you know this. In other words, you are being invited to participate in a game.

Here's just one example: Good psych tests (I mean, ones that are actually researched empirically, and constructed so you can't just 'cheat' your way to a better result) give you plenty of opportunity to contradict yourself. If you strongly agree to X and also strongly agree to Y, and the implicit assumptions behind X are inconsistent with the implicit assumptions behind Y, you earn a higher demerit than if one or other or both of your statements was less emphatic.

Your doubts justifiably reflect uncertainty about exactly what game you are being invited to play. Who designed the test and what is its real purpose? You are at a disadvantage because you don't know the rules. You don't know what numerical scheme will be applied. Or maybe — and this is potential source of criticism of this kind of exercise — you don't agree to this game at all. (That's what I feel about the new '0-5 star' system of appraisal introduced by eBay. If you're happy with the transaction, there ought to be only one choice, so far as I can see.)

However, if you are applying for a job, you don't really have the option. Honesty is, or ought to be, the best policy. But if it seems to you as if you are being required to be dishonest, give a false account of yourself, then maybe you should consider how badly you want the job.