Nothing more than this. Information which is hold having more value is only regarded this way because of inference. Which is in itself a very delicate process which is dependent on several factors. This is because also quantitative research is also qualitative when statistics is used. When data is transformed into cumulative values we see that the relative frequencies are a measure of a value in comparison with other values when taken the effective into account. The only thing we do is thus an inference or a probabilistic estimation. As a matter of fact qualitative research doesn't uses a fixed interval but this fixed interval is only a projection when you solely look it from the statistical exact stance. Anyhow we use language to communicate results and even more to understand them. Some things are simple but others complex.
Features are qualitative because you have to count them. Data are quantitative because it are numbers themselves. The shift from an equivalence-class and a variable is exactly the shift from counting to inference. What is done into statistics is the featuring of equivalence-classes. We take a difference into account, let say we say that this is different from that. That is something different. Now can the difference be made between hard and soft science by the difference of quantitative or qualitative. There is although only one value when variability is active. You have to say if it is good or bad following a certain hypothesis.
When we take the effective we gonna use classes, when we don't use the effective we gonna look for differences. All differences together form the effective. The effective (the sum of the absolute frequencies) is scientific observative. Which means the making of classes. The relative frequency at the other hand is rather democratic and means the application of classes. Therefore we call it inference, everything becomes digitalized in use. It is a sort of percentage and there is a reduction to soft science. Cumulative functions are also never monotone increasing. Classes are presented by the middle of the class. Thereafter are parameters in use which are featuring values. The median takes the class-length into account the modus or the mean the class. This last section takes with many mathematics and not really statistics to the aspiration of hard science. This forms without more rather a problem with the median because of its dependence to the class itself and not reality as a context.