To clarify, the -most- accurate IQ score is obtained in the first year of a norm's life. All scores after that are slightly, and increasingly, inflated. It's not that there's a sweet spot a few years down the line when the scores are accurate. The -least-accurate scores are obtained on a test that is being replaced (or afterward, obviously). The reason some people want their children to be tested on nearly-obsolete norms is that they suspect that their children are borderline for qualifying for a higher classification, and are taking advantage of score inflation to artificially boost their apparent IQ scores.
A more innocent reason to use the old WISC-IV when the WISC-V is already out would be that the examiner is experienced with it, and might be more skilled at administering it, interpreting it, or both. A reason for giving the WISC-IV that has nothing to do with the client is that the new WISC-V test kit costs around $1000 to purchase, and maybe you as an evaluator aren't in a business position to incur that expense immediately, or you still have a pile of WISC-IV record forms lying around, and tossing them unused is a bit painful, at about $10 a pop. It is considered ethical, though not optimal, to continue administering an old test for up to a year after the new version has been released. And for kids near the population mean, the Flynn effect (norm obsolescence that results in score inflation) is not marked enough to affect high-stakes decisions (estimated to be 0.3 points per year, or 3 points per decade). There is some research that indicates that the Flynn effect is more significant further away from the mean, such as for the gifted population.
If you want more detailed info on score inflation over time, google "IQ scores Flynn effect", and you'll find several articles on it. But, in short, no one really knows why (although you will find a number of ideas floating around in the field), but IQ scores have been steadily rising over the past 100 years or so (which is how long the history of IQ testing is), at a rate of roughly 3 points per decade. As modern IQ scores are deviation IQs, meaning they represent your relative standing in a population of your age and cultural peers, this score inflation forces publishers to re-standardize tests periodically, so that you don't have more than 50% of people performing above the 50th %ile, which would make no sense, statistically. Various theories have been suggested, ranging from the idea that people in the industrialized nations are actually getting more intelligent on the average (generally dismissed by the most reputable researchers), to changes in what is considered common knowledge, as cutting edge knowledge or technology becomes everyday. An example of this that gives face validity to this reasoning would be the difference between a 70 year-old struggling with using a new smartphone, and a 2 year-old's seemingly natural facility with an iPhone. Is it true that the 2 year-old is more intelligent than the 70 year-old? Not necessarily, but their prior experiences and cognitive expectations/schema are quite different, which affects how they approach the same problem-solving situation. This is part of what is captured in using the most recent norms possible. When you use old norms (WISC-IV), you are comparing your six year old to the population of six year olds who were born in 1996 (currently 18 year-olds). You may contemplate whether you think their experiences and cultural context are comparable.
According to the publisher, the new WISC-V has a more robust floor and ceiling, so it should be better at distinguishing both the top and the bottom ends of the population. The factor structure of the WISC-V is also more in line with current intelligence theory. All other things being equal, I would go with the WISC-V (but then, I am an evaluator myself, and I like new tests!).
Depending on your state's regulations about giftedness, you may be able to ask your local public school district to test at no or minimal cost. In some states, giftedness is considered a special population, with educational protections. Check your state DOE's website for info on gifted ed in your state. If you go the private route, you should be able to get an IQ only for under $500, but not much under, depending on where you are. Thousands of dollars usually comes in if you want other things, like achievement or neuropsych testing. If your local university has any research projects about giftedness, the graduate students administering the tests will have had some training on evaluating the gifted, and will be supervised by faculty members. But you will probably be better off if pre-doctoral students who are already master's level-qualified, or post-doctoral fellows working on licensure, are giving the tests, than if you have second or third year grad students.
