In article <R91n9.13274$[EMAIL PROTECTED]>, <[EMAIL PROTECTED]> wrote:
Something went wrong with my first attempt to reply to this, which seems to have disappeared (no, I DON'T suspect a conspiracy). Since then the tone of the debate has gone down further, degenerating to really disgusting sexist putdowns. I'll make a few comments below but I don't expect I'll make any more. >With large surveys it is often possible to develope subsamples that fill out >crosstabulations. If not, then the extremes of the causes need to be trimmed >until we get a cross tabulation with the corners containing representative >combinations of the causes. To trim just remove the data above and below >some designated level of cause z-score, sayz=+/- 1.7, for example. Surveys >in general could be designed to collect such data, though the scientist >might need to seek out those individuals who can fill out the extremes. One might suspect, however, that such unusual individuals are not representative, so basing conclusions on them seems dangerous. >This point really strikes at a deeper issue. If the measures are so designed >that they are necessarily correlated, then they are bad measures. A repeat >an example I described earlier of a survey we designed to measure the abuse >that occurs in cults. We factor analyzed a large survey of former cult >members descriptions of their groups. Four meaningful and reliable factors >showed up. A number of items loaded on more than one factor. I removed all >the items that loaded on more than one factor, in order to unconfound the >items. It seems like in the end you're attributing causation to latent factors, which may sometimes make sense, but is certainly not a simple exercise. I wonder whether the method is applicable to what I would take to be typical problems. For instance, an observational study in which obesity, dietary cholesterol, blood cholesterol, and amount of arterial plaque are measured for each subject, with the aim of seeing whether any or all of the first three cause the third. By "cause", I mean that an intervention changing one of them would affect arterial plaque. Note that the potential interventions are for the actual variables, so we're not really interesting in whether some latent variable might be a "cause". >If the correlation is less that .95 then there is at least error variance >included with the measure. Perfect correlation is beyond the scope of CR >because it works by looking at the changes in the relationships of the >causes across the levels of their mutual effect. If two variables are >perfectly correlated, then numerically they are confounded. No statistic >will unconfound them. Most of the phenomena of interest to correlational >researchers, however, are not perfectly correlated with anything. So the >dialectical (two causes or more approach is promising). We are looking at >the causes of measured effects. Some of the variance in the measures will >likely be measurement error. This measurement error counts as a causal >variable! It seems from this that you are claiming that from a scatterplot (mentally trimmed if you like) of two variables, x and y, that aren't perfectly correlated you can determine whether x causes y, y causes x, or both are caused by z. (Since there are two potential causes - eg, x and the measurement error.) Presumably I've misunderstood, since this is clearly impossible. Radford Neal ---------------------------------------------------------------------------- Radford M. Neal [EMAIL PROTECTED] Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] University of Toronto http://www.cs.utoronto.ca/~radford ---------------------------------------------------------------------------- . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
