How relible are standard sources of descriptive statistics?

In other words, how reliable are the numbers in publications like the "U.S. Statistical Abstract"? Is it possible that some of them are carefully compiled and hence largely reliable while others are very carelessly collected to the point of being garbage? Is there really any way we can reliably tell the difference?





And do we use such traditionally "reliable" sources in the social sciences only because they are the only ones we have available? In other words, are we forced to use them because we are, quite simply, stuck with them?





And if we compile new data bases of statistical information, do these same possible problems of collection make them as unreliable as previously available collections?





To summarize all this in a single question, do we blindly use all these resources in the same traditional but unreliable ways largely because that is all we know how to do?





Harleigh Kyson Jr.

How relible are standard sources of descriptive statistics?
Is there are way we can reliably tell the difference between the carefully collected and garbage? Probably not, but we can look at who collected (and/or funded the collection of) the data. Government agencies are (hopefully) going to be very reliable, while those with a specific agenda may well be less reliable because of that agenda.





We use them because to a degree we are stuck with them yes, but that is because we can not necessarily change what we have. For example crime statistics, if you look at reported crimes, we can not really get more accurate results unless people actually report the crimes, so we are limited that way.





So do we blindly use these resources? Certainly not, as we are (or should try to be) aware of their limitations, but we can not change them.

human teeth