Take it as one data point in your meta analysis :-). With a minimum sample of 100 per model, the data is still interesting even if, as you pointed out, the level of confidence varies with the sample size. But we can reasonably assume that 1TB hard drives are selling more, in a generic consumer environment, than high end video cards or SSDs. And even if we had full data, there would still be room for argument. "Large sector" hard drives have suffered from a relatively high return rate that was caused more by incompatibilities than with real drive issues. A few years ago, dual GPU video cards had an awful return rate, not because they were defective, but because Vista didn't recognize them at first. Memories suffer as well because customers aren't always paying attention to requirements and chipset constraints.