Not always. There are many population-based human, animal, and plant-based research studies, also referred to as cross-sectional studies. Using large sample sizes to compare two or more groups that differ in some variable, influences of that variable may be determined. The large sample sizes eliminate any potential influences of non-controlled factors, which in the absence of any selection bias wind up being equally distributed across the groups being compared.
With a large enough sample size, a population of comics of one page quality can be compared with validity to a population of another page quality. The Hulk 181 datamining is an example. If, however, the sample size is too small, as is the case with the paucity of sales data for BB28, then no valid comparison can be made. In this latter case, a well-controlled comparison is necessary, such as provided by back-to-back sales on Comiclink, E-Bay, or through a dealer such as yourself.
Great post (thumbs u Pretty much a first week topic in stats 101. Not enough data for a statistical inference is the same(or potentially worse) as no data for a statistical inference.