If it is possible to affect IQ by discrimination,
If by IQ you mean performance on IQ tests, then of course. IQ itself is not a trait, nor is it scientifically valid in the first place.
it doesn't necessarily mean it happened in the case of the racial difference.
If you are denying the fact of massive discrimination against nonwhites -- past AND present -- then your position is delusional.
If you are instead arguing that it may not necessarily have been the case that the discrimination waged against nonwhites (in your mentioned case, "blacks") is the sole specific factor in accounting for differences in scores on IQ tests, well of course that's possible. It's rare for results to come from any one cause. Simply eating a good breakfast or not can shift a score on a test.
Are black Americans really "transformed – into timid and subservient children,
Literally speaking, of course not. In terms of specific contexts, however, YES. There is plenty of evidence, not the least of which is hundreds (if not thousands) of formal research interviews in which "black" people confirm the "double life" phenomenon describe by W.E.B. DuBois and other social critics with regards to nonwhites playing what amount to character roles in a white-dominated racial mythology. This takes many forms, including, but not limited to, walking a tightrope of expectations (i.e. don't act too smart or you'll be seen as being "uppity" or having an agenda, don't act too dumb or you'll be interpreted not as individually unqualified, but as a racial exemplar of why nonwhites generally are not to be considered -- by racists -- to be qualified, etc.). Naturally, when dealing with adults, the specific form this kind of character-acting takes will be quite different from the impulses followed by the children in Elliott's original exercise.
including those who had previously been dominant in the class"? Personally I'm not sure telling a bunch of intelligent kids they are stupid would make them stupid,
Certainly not permanently, but playing DOWN to underwhelming low expectations is -- once a gain -- a well-documented finding not confined to grade school children. You could practically grab any management professional with more than a year's experience and they can tell you the same thing from firsthand observation.
and I also doubt that black American kids face this level of discrimination today.
The severity and consistency has changed, but not the basic character of the discrimination. For my own part, every last nonwhite person I know or have worked with for an extended period of time has either had several such occasions, or I have seen it happen to them right in front of me. Keep in mind that you must have developed some level of trust before people will just spill out their painful experiences and allow themselves to have that kind of vulnerability around you, but even in my own relatively small circles, I still regularly encounter nonwhite professionals who can easily recall numerous occasions where their basic competence was challenged or presumed lacking in ways where their "white" colleagues were not.
How does this discrimination theory explain the low IQ of black run nations?
Go look up "isolation of the variable under study" and get back to us in a few years. This has already been dealt with. While you're at it, IQ is itself also not an actual trait, but a deeply contingent and narrow metric.
And anyway, I have conceded that it may be possible for discrimination to affect IQ, but the consistent pattern across times and places indicates it doesn't very much. The only way to know for sure is to do the experiment.
Any experiment with the purported goal of isolating factors leading to disparities in scores on IQ tests (or any other metric) vis-a-vis "race" would -- in order to isolate the variable under study -- have to involve centuries of placing "white" people under the heel of racist oppression and maintaining an imposed life of degradation and artificially constrained opportunities in order to see if "white" people would fare any better if/when the tables were turned on them (on the axis of "race"). Because we still live under white supremacy, such arrangements aren't likely to be done. As such an experiment (oppressing people just for the sake of trying to settle an argument) is profoundly unethical, it shouldn't be done. In any case, the level of dominance and institutional power needed to implement such an experiment in the first place would already be more than sufficient to vanquish racist oppression generally -- ending the negative effects of racist oppression, period -- rather than preserving an unnatural and unnecessary set of harmful institutions (thus providing another compelling reason to not even attempt such experimentation).