PPPablo de Pedraza

Pablo de Pedraza

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Pablo de Pedraza has a total of 13 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Transparent policy evaluation through causal machine learning: An application to SME digital subsidies

    by Eugeni Gil-Ocana, Ana Garcia-Bernabeu, Pablo de Pedraza

    Causal machine learning lets policy evaluators estimate how the effect of an intervention varies across recipients, but the flexibility that makes this possible also turns the estimator into a black box, hard to reconcile with the transparency and accountability that public decision-making requires. This paper proposes an evaluation framework that pairs flexible estimation with explicit transparency devices and applies it to Kit Digital, a large Spanish programme of digitalisation subsidies for small and medium-sized enterprises. Using firm-level administrative data, we combine Double Machine Learning and a causal forest to estimate heterogeneous effects on revenues, value added, and employment one and two years after treatment, and we open the estimates to scrutiny through a post-hoc SHAP decomposition and an interpretable-by-design Best Linear Projection with firm-size group effects. The programme has a small positive effect on employment that grows over the two-year horizon. The effects on revenues and value added point the same way but do not survive plausible unobserved confounding. The transparency tools agree that the response concentrates among smaller firms, and re-estimation by alternative designs leaves the direction and broad magnitude unchanged. The contribution shows that a black-box causal estimator can be made auditable enough to inform the targeting of public support without sacrificing rigour.

  • Use of Large Language Models in Rubric-based evaluation: A Framework and Systematic Mapping Review

    by Pau Carles-Vega, Roberto Elias Cervelló-Royo, Pablo De Pedraza, Josep Domenech

    Large language models (LLMs) have shown significant potential in automating complex evaluation tasks. This paper introduces a comprehensive framework for rubric-based evaluation using LLMs, aimed at enhancing consistency, transparency, and alignment with human judgment. The framework is composed of stages that structure the evaluation process effectively: text selection, prompt engineering, LLM configuration, evaluation, and calibration. We confronted the framework with existing empirical studies. The review revealed predominant trends in educational applications, dominance of GPT models, underscored the importance of detailed rubric design and prompt engineering, and emphasized the use of rigorous metrics for ensuring system validity and reliability. Our findings raise important questions for future research on best practices for automated assessment methodologies.

  • Open Access

    Una encuesta voluntaria y continua en la red en España: sesgo, ponderación y eficiencia

    by Pablo De Pedraza García, Kea Tijdens, Rafael Muñoz de Bustillo Llorente, Stephanie Steinmetz

    Utilizando los datos obtenidos mediante una encuesta continua en Internet (CVWS), tusalario.es (WI), que tiene como población objetivo la población activa española, el trabajo analiza, en primer lugar, el sesgo que dicho método de encuesta produce. En segundo lugar, aplica dos metodologías de ponderación con la intención de corregir los errores de representatividad. En tercer lugar, para saber si la ponderación de los datos corrige el sesgo y hace posible la obtención de resultados y conclusiones aplicables al conjunto de la población activa, la eficiencia de ambas metodologías de ponderación es evaluada. La estrategia de evaluación consiste en comparar salarios medios, índices de desigualdad y regresiones salariales obtenidos antes y después de ponderar los datos del WI con aquellos que pueden obtenerse utilizando los datos de la Encuesta de Estructura Salarial (SES), que lleva a cabo el Instituto Nacional de Estadística. Los estadísticos y parámetros estimados se mueven en la dirección correcta después de ponderar.

  • Open Access

    General theory of data, artificial intelligence and governance

    by Pablo de Pedraza, Ian Vollbracht

    Big Data (BD) and Artificial Intelligence (AI) play a fundamental role in today’s economy that traditional economic models fail to capture. This paper presents a theoretical conceptualisation of the data economy and derives implications for digital governance and data policies. It defines a hypothetical data-intensive economy where data are the main input of AI and in which the amount of knowledge generated is below the socially desired amount. Intervention could consist of favouring the creation of additional knowledge via data sharing. We show that the framework suggested describes many features of today’s data-intensive economy and provides a tool to assist academic, policy and governance discussions. Our conclusions support data sharing as a way of increasing knowledge production on societal challenges and dilemmas of data capitalism and transparency in AI.


  • Open Access

    Are Spaniards Happier When the Bars Are Open? Using Life Satisfaction to Evaluate COVID-19 Non-Pharmaceutical Interventions (NPIs)

    by Pablo de Pedraza, María Rosalía Vicente

    The COVID-19 pandemic has challenged governments worldwide with the design of appropriate policies that maximize health outcomes while minimizing economic and mental health consequences. This paper explores sources of individuals’ life satisfaction during the COVID-19 pandemic, paying special attention to the effects of non-pharmaceutical interventions (NPIs). We studied the specific case of Spanish regions and focused on bar and restaurant closures using data from a continuous voluntary web survey that we merged with information about region-specific policies that identified when and where bars and restaurants were closed. We estimated an endogenous binary-treatment-regression model and found that closing bars and restaurants had a significant negative impact on happiness. The results were statistically significant after controlling for the pandemic context, health, income, work, and other personal characteristics and circumstances. We interpreted the results in terms of the positive effect of socialization, individuals’ feelings of freedom, and the comparative nature of life satisfaction.


  • Open Access

    The Roles of General Health and COVID-19 Proximity in Contact Tracing App Usage: Cross-sectional Survey Study

    by Dirk Witteveen, Pablo de Pedraza

    Background: Contact tracing apps are considered useful means to monitor SARS-CoV-2 infections during the off-peak stages of the COVID-19 pandemic. Their effectiveness is, however, dependent on the uptake of such COVID-19 apps. Objective: We examined the role of individuals’ general health status in their willingness to use a COVID-19 tracing app as well as the roles of socioeconomic characteristics and COVID-19 proximity. Methods: We drew data from the WageIndicator Foundation Living and Working in Coronavirus Times survey. The survey collected data on labor market status as well as the potential confounders of the relationship between general health and COVID-19 tracing app usage, such as sociodemographics and regular smartphone usage data. The survey also contained information that allowed us to examine the role of COVID-19 proximity, such as whether an individual has contracted SARS-CoV-2, whether an individual has family members and colleagues with COVID-19, and whether an individual exhibits COVID-19 pandemic–induced depressive and anxiety symptoms. We selected data that were collected in Spain, Italy, Germany, and the Netherlands from individuals aged between 18 and 70 years (N=4504). Logistic regressions were used to measure individuals’ willingness to use a COVID-19 tracing app. Results: We found that the influence that socioeconomic factors have on COVID-19 tracing app usage varied dramatically between the four countries, although individuals experiencing forms of not being employed (ie, recent job loss and inactivity) consistently had a lower willingness to use a contact tracing app (effect size: 24.6%) compared to that of employees (effect size: 33.4%; P<.001). Among the selected COVID-19 proximity indicators, having a close family member with SARS-CoV-2 infection was associated with higher contact tracing app usage (effect size: 36.3% vs 27.1%; P<.001). After accounting for these proximity factors and the country-based variations therein, we found that having a poorer general health status was significantly associated with a much higher likelihood of contact tracing app usage; compared to a self-reported “very good” health status (estimated probability of contact tracing app use: 29.6%), the “good” (estimated probability: +4.6%; 95% CI 1.2%-8.1%) and “fair or bad” (estimated probability: +6.3%; 95% CI 2.3%-10.3%) health statuses were associated with a markedly higher willingness to use a COVID-19 tracing app. Conclusions: Current public health policies aim to promote the use of smartphone-based contact tracing apps during the off-peak periods of the COVID-19 pandemic. Campaigns that emphasize the health benefits of COVID-19 tracing apps may contribute the most to the uptake of such apps. Public health campaigns that rely on digital platforms would also benefit from seriously considering the country-specific distribution of privacy concerns.


  • Open Access

    The Roles of General Health and COVID-19 Proximity in Contact Tracing App Usage: Cross-sectional Survey Study (Preprint)

    by Dirk Witteveen, Pablo de Pedraza

    BACKGROUND Contact tracing apps are considered useful means to monitor SARS-CoV-2 infections during the off-peak stages of the COVID-19 pandemic. Their effectiveness is, however, dependent on the uptake of such COVID-19 apps. OBJECTIVE We examined the role of individuals’ general health status in their willingness to use a COVID-19 tracing app as well as the roles of socioeconomic characteristics and COVID-19 proximity. METHODS We drew data from the WageIndicator Foundation Living and Working in Coronavirus Times survey. The survey collected data on labor market status as well as the potential confounders of the relationship between general health and COVID-19 tracing app usage, such as sociodemographics and regular smartphone usage data. The survey also contained information that allowed us to examine the role of COVID-19 proximity, such as whether an individual has contracted SARS-CoV-2, whether an individual has family members and colleagues with COVID-19, and whether an individual exhibits COVID-19 pandemic–induced depressive and anxiety symptoms. We selected data that were collected in Spain, Italy, Germany, and the Netherlands from individuals aged between 18 and 70 years (N=4504). Logistic regressions were used to measure individuals’ willingness to use a COVID-19 tracing app. RESULTS We found that the influence that socioeconomic factors have on COVID-19 tracing app usage varied dramatically between the four countries, although individuals experiencing forms of not being employed (ie, recent job loss and inactivity) consistently had a lower willingness to use a contact tracing app (effect size: 24.6%) compared to that of employees (effect size: 33.4%; P<.001). Among the selected COVID-19 proximity indicators, having a close family member with SARS-CoV-2 infection was associated with higher contact tracing app usage (effect size: 36.3% vs 27.1%; P<.001). After accounting for these proximity factors and the country-based variations therein, we found that having a poorer general health status was significantly associated with a much higher likelihood of contact tracing app usage; compared to a self-reported “very good” health status (estimated probability of contact tracing app use: 29.6%), the “good” (estimated probability: +4.6%; 95% CI 1.2%-8.1%) and “fair or bad” (estimated probability: +6.3%; 95% CI 2.3%-10.3%) health statuses were associated with a markedly higher willingness to use a COVID-19 tracing app. CONCLUSIONS Current public health policies aim to promote the use of smartphone-based contact tracing apps during the off-peak periods of the COVID-19 pandemic. Campaigns that emphasize the health benefits of COVID-19 tracing apps may contribute the most to the uptake of such apps. Public health campaigns that rely on digital platforms would also benefit from seriously considering the country-specific distribution of privacy concerns.

  • Open Access

    Survey vs Scraped Data: Comparing Time Series Properties of Web and Survey Vacancy Data

    by Pablo de Pedraza, Stefano Visintin, Kea Tijdens, Gábor Kismihók

    This paper studies the relationship between a vacancy population obtained from web crawling and vacancies in the economy inferred by a National Statistics Office (NSO) using a traditional method. We compare the time series properties of samples obtained between 2007 and 2014 by Statistics Netherlands and by a web scraping company. We find that the web and NSO vacancy data present similar time series properties, suggesting that both time series are generated by the same underlying phenomenon: the real number of new vacancies in the economy. We conclude that, in our case study, web-sourced data are able to capture aggregate economic activity in the labor market.