In my opinion, intelligence is not a 1-dimensional thing. [An analogy would be a 3-dimensional object which we try to describe by its width in a particular direction]. So any IQ test can only give some sort of combined statistic and can be very misleading with respect to more specialised types of intelligence (such as the ability to play chess well, for which you could have an analogy of IQ just by transforming a rating scale), and particularly abilities which involve much more complex combinations of abilities (such as doing groundbreaking scientific research, for example).
A more suitable analogy would be using the volume of the object to describe its dimensions. If you take all objects, man-made or not, there will be a pretty decent correlation between volume and width, length and height, respectively, and also between the different dimension parameters.
You seem to imply that humans have different cognitive abilities that would all fall under the general category intelligence, and that these abilities are independent of each other. This does not seem to be the case, as all these abilities (such as memory, verbal ability, spatial ability etc) are correlated on a statistical level. The basis for this correlation is the general intelligence, according to intelligence theory, and this is what IQ tests that only give a single IQ score intends to measure.
If we return to your objects, volume would correspond to this general intelligence and width, length and height would correspond to e.g. working memory, verbal ability and spatial ability. Sure, volume does not perfectly predict length, width or heigth, but objects of greater volume will in most cases be longer, wider and higher than objects of lesser volume. And all people with high IQ will not have a better spatial ability than people with low IQ, but in the majority of cases, they will have a better spatial ability.
It is no more accurate to think of intelligence as 1-dimensional as it is to think of fitness as 1-dimensional. The reason that the analogy of IQ to one linear dimension of multidimensional intelligence is better than the analogy to volume is mathematical. IQ tests use a linear sum of the scores in the test to determine a single number. Each question can clearly only measure one dimension of the multidimensional space of intelligence (using a large number of ways to measure it, say), so any linear sum of them can also only determine a single dimension in the space. If IQs were defined a product of the results of several tests, the analogy of a volume might be more appropriate, but it would still only give a small part of the information, like describing a 3-dimensional space by its volume rather than its shape.
It is frustrating when people use the sort of sloppy thinking that says "A and B are correlated, therefore they can be considered to be the same thing". Correlation does not mean this. For example, age and height are strongly correlated for young people (much more so than some different sorts of intelligence), so presumably you would say that it is best to think of height and age as being a single thing? I certainly would not. Another quite strong correlation is between height and weight. See the point? Correlations are of interest, but they do not mean that you can think of things as 1-dimensional.
+ 
= 
Moram pronaći nit je danas, koja govori o otmicama izvanzemaljaca...
unfair again!