Please use this identifier to cite or link to this item:
|Title:||Assessing Sapienza University alumni job careers : enhanced partial least squares latent variable path models for the analysis of the UNI.CO administrative archive||Authors:||Petrarca, Francesca||metadata.dc.contributor.advisor:||Terzi, Silvia||Keywords:||optimal scaling
|Issue Date:||3-Jun-2014||Publisher:||Università degli studi Roma Tre||Abstract:||For several years, the Italian universities have been urged to better understand the characteristics of labour demand for their alumni. For this purpose periodic reports of the results of sample surveys are produced by institutions or private entities. These studies are detected directly from the alumni and their employment status after graduation (the most common sample surveys coming from Inter-University Consortium AlmaLaurea). These results constitute important information for students, families and universities. The sample surveys are based on studies that take place at a distance of six months and of two years from the date of graduation. The alumni are interviewed regarding their satisfaction as well as to obtain information on their employment situation. Obviously, all ex post assessments provide us with a lot of information to define overall judgements regarding the educational career but also their employment status. Of course, the definition of a graduate's employment status is not unique. With regards to Italy, the ISTAT records the status of employment when the individual performs, in a reference week, at least one hour of paid work (regulated by a registered contract) regardless of the type of contract and the regularity of employment. Timing is certainly a critical element: studies too close to the end of the educational experience do not sufficiently reect the students' rethinking of their educational lives. However, if the surveys are too far from this experience, the rethinking can itself be strongly influenced by exogenous factors. The labour market demand for the employability skills and knowledge of Sapienza graduates can now be analysed through the new administrative archive called UNI.CO, which was built through a record-linkage between the Sapienza University of Rome archive with those of the Italian Ministry of Labour and Social Policy. The UNI.CO archive has allowed us to evaluate all the signed contracts obtained by Sapienza graduates in addition to full information on all the enterprises and institutions that offered these contracts. This archive allows us to perform a statistical analysis on more sophisticated and robust (and more reliable) details than the standard analysis based on sample surveys. In our study we have, therefore, used objective real data unaffected by any emotional influence. Another important feature is that the data from the UNI.CO archive allow us to study the evolution of alumni throughout the observation period and therefore to analyse changes in job qualifications (differently from the common sample survey supply information about the status of alumni at some fixed point in time (like a photograph)). In this thesis we adopted the administrative UNI.CO archive in order to be able to present the theme of the integration of Sapienza University of Rome alumni into the employee and para-subordinate labour market with more sophisticated and powerful details in relation to the standard analysis based on sample surveys. Moreover, the data of the UNI.CO archive allow us to also study the evolution of alumni throughout the observation period and therefore to analyse the changes in job qualifications. The study of the integration into the labour market of Sapienza alumni has been analysed by adopting the Non-Metric PLSPM method which allows us to handle a large dataset with numerical, nominal and ordinal variables. This technique recently proposed by Russolillo (2012) and now available in the plspm R-package, is characterized by the opportunity to analyse at the same time variables observed in different measurement scales, to investigate the non linearity and to work without distributional hypotheses. The Non-Metric PLSPM has demonstrated a remarkable adaptability to handle the UNI.CO large dataset and we have found that the quantifications of the original data obtained by the procedure are properly carried out using monotone transformations showing the remarkable fact that it is possible to discard the hard assumption of linearity in favor of the milder assumption of monotonicity. In this thesis, we have modified the inner structure of the Non-Metric PLSPM to implement the possibility of modeling binary endogenous latent variables through a logistic regression and then we have introduced the ROC curve for the model validation. This implementation is applicable in the case of blocks with only one MV and this is binary, therefore LVs have a binary structure and the weight and the loading are unitary. We have addressed the study of the integration into the labour market of Sapienza alumni by developing a few models. The first one concerns the quantitative study of the job success of the masters degree alumni of the Sapienza University who belong to the engineering disciplinary sector in terms of quality of work. In particular, we studied indicators of job success to estimate their relationships with educational and job curriculum. These indicators have been constructed defining the concept of optimal contract based on a contract with the characteristics of permanent position, highly qualified professional position identified using the ISCO classification and the actual duration more than or equal to 8 months. The concept of quasi-optimal contract is similar to the optimal definition without the condition of permanent position. The high values of the measurement model have confirmed that the optimal and quasi-optimal indicators are discriminant to the construction of the Job Success block. The second and the third models have been built in order to study the job career in terms of the contract type and quality evolution in the three years after graduation. For these studies we selected from the UNI.CO dataset, independently from the disciplinary group, two sets of data: alumni who at the time of the master's degree had not yet obtained an active contract (non-workers studied by the second model) and alumni who at the time of the master's degree had obtained an active contract (workers studied by the third model). The main purpose we addressed is to model, in the PLS-PM framework, the contract type and job professional quality evolution as latent variables to be studied in relation to educational curriculum, age, initial experience and tendency to evolve. The two models adopted consist essentially of four concepts: the university evaluation, the initial job experience, tendency to evolve and the career evolution. The career evolution is monitored by two constructs: one for the job professional evolution and the other for the job contractual evolution. We have found that in both the models there is a strong influence of the initial job experience and of the tendency to evolve over the job career, while the university evaluation at the level of the master's degree is less important. The age at the master's degree has a marginal effect in the model for non-workers while it disappears for workers. It is remarkable that the implementation of the logistic regression, in all the presented models, has shown a good classification measured in terms of high AUC values. The overall frame that arises from these studies is that when there is a natural individual aptitude of the alumni to change job, this tendency plays a positive important role with the aim of obtaining a satisfactory job; on the other hand, the university path of Sapienza alumni does not seem to have a large direct influence on their job career. It is clear that our analysis can be considered as a starting point for further studies aimed to investigate the relationships between the world of labour and that of the university education in order to improve the efforts that should be made to integrate these two worlds.||URI:||http://hdl.handle.net/2307/4167||Access Rights:||info:eu-repo/semantics/openAccess|
|Appears in Collections:||Dipartimento di Economia|
T - Tesi di dottorato
Show full item record Recommend this item
checked on Sep 23, 2020
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.