People Analytics and Psychometrics Laboratory (PAPsy Lab)

Estimating psychological networks for the measurement of organisational climate

This groundbreaking project aims to introduce how the psychological networks technique can be applied to the investigation of organisational climate through the mutual interactions among its dimensions. A dataset comprised of 1,053,322 workers from 160 companies is being used for this study. Even though the research on organisational psychology typically makes use of nonexperimental designs, psychological networks may suggest potential causal structures in a pathway. For example, workers might not rely on the organisational leadership, which in turn will impinge on the team morale and consequently increase turnover intentions. This causal structure indicates that we would be able to predict turnover intentions by knowing the attitudes towards the leadership that could lead a worker to leave his or her organisation. Nonetheless, we can also predict turnover intentions from team morale, making the knowledge on the attitudes towards leadership no longer necessary for the prediction of turnover intentions.

Untangling the performance of automatic video summarisation methods by video compression

To reduce the time users spend watching videos, automatic video summarisation methods seek to detect the most relevant video parts, which are condensed into a shorter video. Performance evaluation of video summarisation methods is still challenging since users are assigned the task of choosing which part of a video is more relevant so that this could ultimately support the development of automatic methods. These subjective choices, and potential cognitive biases may directly influence the actual performance of the automatic methods. Although the performance of video summarisation methods is typically assessed by well-known statistical metrics, the boundaries of these metrics have not yet been explored fully. This project analyses the limitations of the metrics used to evaluate video summarisation methods.

Should I stay or should I go? Predicting intentions to leave the organisation using machine learning algorithms

This study, in partnership with the University of Cambridge, aims to develop a machine learning based application to calculate the risk of leaving an organisation taking into account the interactions among different factors, such as personality traits, decision-making style, financial situation, health status, cognitive processes, among others. Data was collected online via Twitter and Multidimensional Item Response Theory will be combined with a machine learning algorithm (XGBoost algorithm) to predict the intentions to leave the organisation. Two papers will be produced, the first one, more methodological, will address the combination of techniques from psychometrics and machine learning and, the second, more applied, will report the findings of this investigation. This study can help companies to work beyond their turnover rates, mainly on the analyses of their talented employees with a stronger intention to leave the organisation, and then create new measures aimed at worker retention. Moreover, it may provide insights into the field of organisational behaviour and further contribute to the development of this topic and its measurement.