State Library Agencies Survey/State Library Administrative Agency Survey (STLA/SLAA)
Originally started by NCES with the 1994 data, this survey was published annually through 2010 and biennially since then. The data are now available through 2022 and are currently collected and published by IMLS. These are the data from the state libraries. All states have a state library and the files include data from them and also from the District of Columbia library.. These libraries are also known as the State Library Administrative Agency(ies) (SLAA.) I am not yet sure when the new title was created but I will refer to the series here as the STLA/SLAA data except in those cases of the early data which were the STLA data. These libraries are members of the Chief Officers of State Library Agencies (COSLA.) More information on these libraries and their programs can be found at the COSLA Website.
In my journey through library data, I also tried to create a longitudinal file of these data and I gave up. It has been some time since I looked at my notes but my recollection is that the STLA/SLAA data, like the Academic Library Statistics (ALS) data, have an enormous number of variables. This result comes about because they (that is, NCES when I was working on these data) both seemed to have collect a number of variables but for a few years. I remember thinking that the STLA data, particularly, were chasing what I thought was changing politically important variables which seemed reasonable given the circumstances. But while perhaps politically important but not necessary for the long run. The public library data were more stable than the STLA data while I was looking at them. The ALS data surprised me because I had a great deal of experience with academic data and assembled some long-term longitudinal files but the ALS data were a puzzle. The academic data series I worked on were largely based on the ARL infrastructure which had relatively few variables and libraries and were consistent over the years. If I had more time I might have cracked the ALS data.,Of course, the ALS data are no longer collected as such but librarians are still collecting the academic data without the government agencies. This is what we should be doing, IMHO.
The STLA/SLAA Data are, like all such library data, issued annually, and as I indicated they change a good bit over the years. Among the things that change are the naming conventions. This characteristic of government data publication is consistent in its inconsistency and I have discussed elsewhere my original approach was to "correct" the file names to make them into a more orderly arrangement. That proved to be a bad idea and I repented quickly when I saw the results. These data are named as I got them. Now, I did not collect the STLA data much after I tried to create a longitudinal file so what is here is what I collected recently and they are in order here by year and organized by HTML. They will not sort by year using the filenames.
Note the “Revised” data and also a characteristic change one sees in variable names. For instance, note that in 2001, the zipped SAS file is named istla01a_sas.zip and the zipped SAS file for 2003 is named istla03b_sas.zip. It has been my experience with NCES/IMLS data that the addition of a letter indicates that these are revised data. However, in looking through the documentation of these data, I found that my notion about these data and their revisions are off the mark. In fact, these data that I just collected are not those as originally published and documented. That observation comes from the filenames available now (and here) and those referenced in the documentation.
So, changes are occasionally made.There have been a few times where I discovered a problem and got in touch with someone at NCES and then there was a new file with a slight name change. Now, how do you know if you have the latest? Well, check the IMLS Website. But what if it changes?
One thing that changes is filetypes which, practically, is necessary over time. What program created the file you are using? Note in the files there that there are ASCII files, csv files, SAS, and SPSS files, and even Access files. I know I have seen DBF files in the early years of other series. So, someone went back and moved the ASCII files to SAS and then other formats. The change to SAS was usually the first one. NCES used SAS when I was at NCLIS, as did I, but SPSS is now commonly here. There is even one SPSS file (I think) that has a .txt in its filename. Reading the documentation indicated to me that perhaps another level of changes were made to the data. I never had any reason to distrust the work from NCES and I did attend a meeting with folks from Census, I believe, who had a contract to work on the STLA data. I thought they were serious analysts. I mentioned the large number of variables in the series and I can share one piece of information. I asked one of the analysts if they had made sure that no variable names were duplicated? Yes, they checked that. The short variable names used in programming these data sometimes require disciplined attention to avoid error.
In creating longitudinal files, one will find all manner of filetypes as mentioned above so part of creating these kinds of files is converting the various years of data into a common format. When one is converting data in paper formats, one has to key the data into a computer-friendly format so that they can be manipulated. Keying data is an arcane skill now and I expect our library colleagues would find it daunting. Commonly, NCES and IMLS will convert previously issued data. For instance, if in 1990 the data were published in the then state of the art text files, that is ASCII, that is the record. Later, as statistical utilities like SAS became common, NCES issued SAS files of these data. And later still, SPSS files became common so one would find both for later data. In the PLDF3 raw data files, I have early versions of the files that are no longer available at IMLS. That fact is probably just as well for most users because not many people will have the software or skill to deal with .dbf or ACCESS formats, let alone keying data so conversion is a useful and practical—and tricky—procedure. I am not sure exactly what was done with these STLA/SLAA data files you will find here so if you want to work with these data, please read the documentation carefully. I suspect that it is all OK but I do think you need to approach the data with that fact in mind. The reader will note that there are years of the documentation that explicitly says "Revised."
What we have here are data, documentation of the data series, and reports. Here's a link to the capture of the IMLS STLA site when I downloaded the files. And a live link to the IMLS SLAA site.