Regarding the data quality, several remarks are necessary.
Firstly, what we count in this paper are job adverts, not the actual number of jobs advertised. This is because for job seekers, it does not matter much (other than chances of success possibly being affected) whether an employer advertises just one or several posts at the same time: each job seeker after all only needs one job (assuming that it is not possible to hold two or more of the jobs advertised at once at the same time). This is even more so if the job seeker does not have the required qualifications for the kind of job for which several openings are advertised. What matters more to job seekers is the number of different opportunities with different qualifications as requirements: the more such different opportunities are offered, the greater the chance that all job seekers will find at least some post(s) they are qualified for, and for which they can apply. And perhaps even more importantly, many job adverts, while clearly offering multiple posts of the same kind, do not mention the specific number of posts offered (e.g. speaking of ‘several openings’, or naming jobs in the plural, e.g. ‘research assistants’ rather than the singular ‘research assistant’). It is thus impossible for us to say how many jobs were advertised in the period covered with any reasonable level of accuracy. The number of adverts, on the other hand, is accurately known, and that is why it is what we count. This, however, leads to a degree of uncertainty in the conclusions we draw from the interpretation of our data: if for instance, in one year, many adverts were published in a country which were aimed at filling just a single position each, while in another year, only few adverts were published, but each for a large number of positions to be filled, the total number of jobs advertised in those two years may actually be the inverse of what our data seems to show. While such a situation is not particularly likely to have occurred, this caveat has to be kept in mind when reading our interpretation below.
Secondly, we do not invest an overly great effort into searching for jobs, but rather rely on the supply of adverts from a number of easily accessible or searchable sources, or adverts that we receive through our normal professional links with colleagues and/or institutions. Thus, the jobs database is certainly not a complete record of all archaeology posts advertised in Austria, Germany and the UK in the period since November 2003. However, while certainly not complete, we are reasonably sure that for much of the period from mid-2006 onwards, for most sectors of archaeological work, we have a quite good record. As such, we believe that for the period covered in this article, 1/1/2007-15/9/2012, our dataset is at least representative for the three countries examined in this article. We are even reasonably convinced that we have registered a majority of all job adverts in all sectors except fieldwork (on which a caveat will follow below) in the three countries examined here. Also, the data collected over the period was collected consistently, in that we did not change our data mining methods or even data sources. Thus, trends shown in our data are very likely to be reflecting real changes in the supply of new job adverts becoming available.
Thirdly, job adverts are searched for and entered manually; data collection is not automated. In addition, everyone with a valid IÖAF website account, which is available for free, can enter job adverts themselves, and several volunteers thankfully have done so over the past 9 years. Thus, the quality of adverts, e.g. how much detail about posts is supplied, as well as categorisations of posts and other details, can vary considerably. However, some key information has been supplied for virtually all posts since the job entry form requires compulsory completion of some fields. By and large, we do not check the accuracy of such information upon data entry. In the pre-analysis phase of the research for this article we have, however, corrected obvious mistakes in the data where possible, or excluded individual datasets from the analysis if the information supplied was obviously faulty and could not be reconstructed with reasonable accuracy. A total of 2.1% of adverts was identified in this pre-analysis phase as having at least one or, in some cases, several fields with faulty entries. Since the total number of adverts in the database at the time of analysis was 9,381, this fault rate is negligible; particularly since most such affected entries had obvious faults only in single fields of a multiple field database, and the fault rate for each individual data field used in the analysis thus is considerably lower than 2.1%. The standard deviation for all numbers given in the various analyses in this paper is thus likely to be no more than 5%.
And fourth and finally, as already noted above, fieldwork jobs are considerably underrepresented in our database (see fig. 2). This is for several reasons, which differ at least to some extent between the UK and German (including Austrian) archaeology labour market. In the UK, most fieldwork jobs are now available online via the BAJR website, which is free for job seekers, so we saw no need to make these posts available (by entering them manually) onto yet another jobs resource. Where we do get fieldwork job adverts for the UK through other channels, or where these are entered directly by volunteers or employers, we do of course publish them regardless, since we do not check BAJR before we add adverts (as that would further complicate already tedious and labour intensive work). Nonetheless, the result is that for the UK, fieldwork jobs are grossly underrepresented in our jobs database. For German and Austrian archaeology, the fieldwork jobs are equally underrepresented, but for a very different reason. It is not as if there was a German and/or Austrian fieldwork jobs online resource that already does a perfectly good job that we thus need not duplicate. Rather, fieldwork jobs in Germany and Austria rarely seem to be advertised in a form that is accessible via the Internet; most of them seem to either be advertised only by word of mouth or by pinning them to notice boards at local universities. Despite the very different reasons, the result nonetheless is the same: fieldwork jobs are underrepresented.
Fig. 2: Distribution of adverts according to sector.