I still haven't been able to study the methodology of the EU survey on globalisation, but one point I wanted to add, is that we can of course add up quantified observations which really refer to different things.
For example, among other things, on my writing desk I have at the moment: 1. a computer monitor, 2. a keyboard, 3. a PC computing device, 4. a printer, 5. a cup, 6. a packet of drum tobacco, 7. a calculator, 8. a Velvet Underground CD, 9. a book by Thomas Friedman, 10. a red notebook, 11. a book by Studs Terkel, 12. a book by Isaac Deutscher, 13. some writing paper, 14. two visiting cards. 15. a book by George Soros, 16. a lighter, 17. a book by Ernest Mandel, 18. a postcard advertising the movie "Divine Intervention" If I said, after recording and counting up these observations, "I have at least 18 objects on my desk", I would state the truth, because the term "objects" as aggregation principle could be validly applied to all those things on my desk. But if I said "I have at least 18 mechanical devices on my desk" then I would not be telling the truth, because although some are, many of the objects on my desk are not mechanical devices, at least not in ordinary language. It could of course be argued that, if I had a certain sort of lifestyle, then all the objects on my writing desk represent mechanical devices in some sense - i.e. there is some relationship between the objects on my desk, and mechanical devices which they symbolise. Nevertheless, if I were to ring up my sister and report "I have 18 mechanical devices on my desk" she would probably not understand what I mean immediately, or indeed think that I had gone nuts, or at least was repairing my computer. In which case my report is a miscommunication. Suppose that I now do a telephone survey, and I ask respondents how many objects are placed on their kitchen table. I would evoke considerable response burden, because the aggregation principle is not clear - how do we define discrete objects ? If some respondents are a bit blurry or blind, or if they have two kitchen desks, this might add to the response burden. One respondent might ask, do I count both the individual fruits in the fruitbowl, or, for example, do I count the fruitbowl as one object - do I count the pen and the cap of the pen as one object or two objects ? But on the other hand, they might not ask this at all, either because they assume that they know the meaning of the question, or, if they do not know it, they don't want to ask, lest they be considered stupid by the interviewer. In aggregating my telephonically gathered observations, I might therefore get a considerable distortion between the data distribution and the actual situation, exclusively because I asked respondents about "objects" without specifying distinctions and rules necessary to count them as discrete objects. The same trick of course could be applied when we ask about globalisation. There are many theories and ideas about globalisation, and therefore if I ask people "are you in favour of globalisation, or not" (which contains the assumption that globalisation is something you can be for or against) I may get a "yes" or a "no" answers, but I still do not know what the "yes" or "no" refers to, and it might well refer to a variety of things in the minds of respondents. The result might be the finding that the majority of respondents are in favour of globalisation, but I do not know exactly what they are in favour of. This is somewhat analogous to the survey question "do you like dogs, or not" ? Some respondents may like all dogs, some respondents may not like any dogs, okay, but some respondents might like only some dogs, or only one dog, or only fox terriers, or only pictures or sculptures of dogs but not real dogs, or only the dogs which belong to someone else at a considerable distance, and so on. And so they will understand the question differently, based on their likes and dislikes and their interpretation of the survey question. I would obtain a result that X number of people like dogs, and X number of people do not like dogs, but I do not really know what the result really means, what reality it refers to. Knowing that this is the case, we could design a survey question about globalisation which is guaranteed always to obtain over 90% of respondents in favour of globalisation, and in this way we could manufacture a public opinion in favour of globalisation. The problem however is, that if the discrepancy between survey results and real opinion is too great, people will not longer trust surveys, and refuse to co-operate with them. In that case, the survey research culture would be destroyed by the linguistic communication used. Hence, only a limited amount of conceptual distortion is compatible with survey research, such that the discrepancy between the survey result and real opinion is not too great. You can fool some people all of the time, you can fool all people some of the time, but you cannot fool all people all of the time. Jurriaan
