When possible, we prefer to use BLS employment and salary calculations over similar statistics that we create from Census data. Here are what we see as general advantages of BLS data over Census data:
BLS estimates are created by statisticians and economists with tremendous experience and focus on accurate reporting for narrow and well-defined quantities. We aggregate Census data from rich household-level surveys to totals that compare several of many measurements in the surveys, and our final estimates lack that statistical precision that a narrow study on the topic at hand would provide.
The primary source for BLS data is employer surveys, and generally employers are familiar with the breadth job classifications as well as other survey questions regarding salary or benefits. We also suspect that the responses are informed by careful data records when numerical data is requested from employers. As such, we believe employer-based survey responses will generally be more reliable that individual-based responses.
BLS data is generally available at the SOC codes job description level, which agree with our specialty level for Ididio career pages that have multiple job specialties, and otherwise are the same as a our main career pages. Census data is available at the Ididio career page level only. It is the discrepancy in reporting between these information sources that motivated our page structures.
For these reasons, we would recommend BLS data as your primary data source for information that both sources provide. However, there are compelling reasons to look at Census data in addition:
The household surveys from which we draw to create aggregations for employment topics from Census surveys are incredibly rich, and they allow us to examine connections that are not present in the BLS data. Although the distributions we derive from the surveys don't stand up to the BLS employment reporting methodology, they still provide compelling insights that should encourage further investigations.
BLS surveys have source limitations that we describe more fully in their primary source descriptions. As an example, the BLS OES data that forms the basis of salary reporting on Ididio omits agricultural and home (self-employed) workers.
Thus, while we respect the accuracy of the BLS reporting on employers, we find that the Census data can add to our understanding of careers. Whenever possible we share data from both BLS and Census sources, with the belief that all problems are best considered from many angles.