|
on Experimental Economics |
| By: | Cruces, Guillermo (CEDLAS-UNLP); Tortarolo, Dario (World Bank); Vazquez-Bare, Gonzalo (UC Santa Barbara) |
| Abstract: | We develop a framework to analyze partial population experiments, a generalization of the cluster experimental design where clusters are assigned to different treatment intensities. Our framework allows for heterogeneity in cluster sizes and outcome distributions. We study the large-sample behavior of OLS estimators and cluster-robust variance estimators and show that (i) ignoring cluster heterogeneity may result in severely underpowered experiments and (ii) the cluster-robust variance estimator may be upward-biased when clusters are heterogeneous. We derive formulas for power, minimum detectable effects, and optimal cluster assignment probabilities. All our results apply to cluster experiments, a particular case of our framework. We set up a potential outcomes framework to interpret the OLS estimands as causal effects. We implement our methods in a large-scale experiment to estimate the direct and spillover effects of a communication campaign on property tax compliance. We find an increase in tax compliance among individuals directly targeted with our mailing, as well as compliance spillovers on untreated individuals in clusters with a high proportion of treated taxpayers. |
| Keywords: | partial population experiments, spillovers, randomized controlled trials, cluster experiments, two-stage designs, property tax, tax compliance |
| JEL: | C01 C93 H71 H71 H26 H26 H21 H21 O23 |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17256 |
| By: | Ro’i Zultan (BGU); Yamit Asulin (BGU); Yuval Heller (Bar Ilan University); Nira Munichor (Bar Ilan University) |
| Keywords: | Social image, social distance, field experiment, crowding up, prosocial behavior |
| JEL: | C93 D64 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:bgu:wpaper:2410 |
| By: | Leonardo Gambacorta; Han Qiu; Shuo Shan; Daniel M Rees |
| Abstract: | In this paper we examine the effects of generative artificial intelligence (gen AI) on labour productivity. In September 2023, Ant Group introduced CodeFuse, a large language model (LLM) designed to assist programmer teams with coding. While one group of programmers used it, other programmer teams were not informed about this LLM. Leveraging this event, we conducted a field experiment on these two groups of programmers. We identified employees who used CodeFuse as the treatment group and paired them with comparable employees in the control group, to assess the impact of AI on their productivity. Our findings indicate that the use of gen AI increased code output by more than 50%. However, productivity gains are statistically significant only among entry-level or junior staff, while the impact on more senior employees is less pronounced. |
| Keywords: | artificial intelligence, productivity, field experiment, big tech |
| JEL: | D22 G31 R30 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:bis:biswps:1208 |
| By: | Philip Brookins; Jennifer Brown; Dmitry Ryvkin |
| Abstract: | Reward schemes may affect not only agents' effort, but also their incentives to gather information to reduce the riskiness of the productive activity. In a laboratory experiment using a novel task, we find that the relationship between incentives and evidence gathering depends critically on the availability of information about peers' strategies and outcomes. When no peer information is available, competitive rewards can be associated with more evidence gathering than noncompetitive rewards. In contrast, when decision-makers know what or how their peers are doing, competitive rewards schemes are associated with less active evidence gathering than noncompetitive schemes. The nature of the feedback -- whether subjects receive information about peers' strategies, outcomes, or both -- also affects subjects' incentives to engage in evidence gathering. Specifically, only combined feedback about peers' strategies and performance -- from which subjects may assess the overall relationship between evidence gathering, riskiness, and success -- is associated with less evidence gathering when rewards are based on relative performance; we find no similar effect for noncompetitive rewards. |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2409.06248 |
| By: | Bašić, Zvonimir (University of Glasgow); Bortolotti, Stefania (University of Bologna); Salicath, Daniel (NAV Norwegian Labour and Welfare Administration); Schmidt, Stefan (Max Planck Institute for Research on Collective Goods); Schneider, Sebastian O. (Max Planck Institute for Research on Collective Goods); Sutter, Matthias (Max Planck Institute for Research on Collective Goods) |
| Abstract: | Incentives are supposed to increase effort, yet individuals react differently to incentives. We examine this heterogeneity by investigating how personal characteristics, preferences, and socio-economic background relate to incentives and performance in a real effort task. We analyze the performance of 1, 933 high-school students under a Fixed, Variable, or Tournament payment. Productivity and beliefs about relative performance, but hardly any personal characteristics, play a decisive role for performance when payment schemes are exogenously imposed. Only when given the choice to select the payment scheme, personality traits, economic preferences and socioeconomic background matter. Algorithmic assignment of payment schemes could improve performance, earnings, and utility, as we show. |
| Keywords: | effort, productivity, incentives, personality traits, preferences, socio-economic background, ability, heterogeneity, sorting, algorithm, lab-in-the-field experiment |
| JEL: | C93 D91 J24 J41 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17287 |
| By: | Ziyan Cui; Ning Li; Huaikang Zhou |
| Abstract: | Artificial Intelligence (AI) is increasingly being integrated into scientific research, particularly in the social sciences, where understanding human behavior is critical. Large Language Models (LLMs) like GPT-4 have shown promise in replicating human-like responses in various psychological experiments. However, the extent to which LLMs can effectively replace human subjects across diverse experimental contexts remains unclear. Here, we conduct a large-scale study replicating 154 psychological experiments from top social science journals with 618 main effects and 138 interaction effects using GPT-4 as a simulated participant. We find that GPT-4 successfully replicates 76.0 percent of main effects and 47.0 percent of interaction effects observed in the original studies, closely mirroring human responses in both direction and significance. However, only 19.44 percent of GPT-4's replicated confidence intervals contain the original effect sizes, with the majority of replicated effect sizes exceeding the 95 percent confidence interval of the original studies. Additionally, there is a 71.6 percent rate of unexpected significant results where the original studies reported null findings, suggesting potential overestimation or false positives. Our results demonstrate the potential of LLMs as powerful tools in psychological research but also emphasize the need for caution in interpreting AI-driven findings. While LLMs can complement human studies, they cannot yet fully replace the nuanced insights provided by human subjects. |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2409.00128 |
| By: | Bohren, Noah (University of Lausanne); Hakimov, Rustamdjan (University of Lausanne); Lalive, Rafael (University of Lausanne) |
| Abstract: | Generative artificial intelligence (AI) has made substantial progress, but some capabilities of AI are not well understood. This study compares the ability of AI to a representative population of US adults in creative and strategic tasks. The creative ideas produced by AI chatbots are rated more creative than those created by humans. Moreover, ChatGPT is substantially more creative than humans, while Bard lags behind. Augmenting humans with AI improves human creativity, albeit not as much as ideas created by ChatGPT alone. Competition from AI does not significantly reduce the creativity of men, but it decreases the creativity of women. Humans who rate the text cannot discriminate well between ideas created by AI or other humans but assign lower scores to the responses they believe to be AI-generated. As for strategic capabilities, while ChatGPT shows a clear ability to adjust its moves in a strategic game to the play of the opponent, humans are, on average, more successful in this adaptation. |
| Keywords: | artificial intelligence, ChatGPT, Bard, creativity, experiment |
| JEL: | I24 J24 D91 C90 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17302 |
| By: | Antinyan, Armenak (Thames Water); Burn, Ian (University of Liverpool); Jones, Melanie K. (Cardiff University) |
| Abstract: | While hiring discrimination against disabled candidates is widely documented, the reasons for such discrimination and the mechanisms designed to reduce it are not well understood. This study aims to tackle these questions through a large-scale correspondence study. Fictitious job applications were sent to about 4, 000 job vacancies for accountants and financial accounts assistants in the UK. Consistent with discrimination, we find a 5.6 percentage point (15%) gap in the employer callback rate associated with mobility impairment indicated by the use of a wheelchair, but substantial occupational heterogeneity. Productivity signals designed to reduce statistical discrimination, including the offer of a positive reference from a previous employer and, enhanced education and technical skills, do not reduce, and actually widen, the disability gap in callbacks. Our findings are suggestive of taste-based discrimination being a significant barrier to employment for disabled people that requires policy attention. |
| Keywords: | disability, discrimination, correspondence studies, productivity signals |
| JEL: | J14 J71 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17290 |
| By: | Jingru Jia; Zehua Yuan |
| Abstract: | This study explores the potential of large language models (LLMs) to conduct market experiments, aiming to understand their capability to comprehend competitive market dynamics. We model the behavior of market agents in a controlled experimental setting, assessing their ability to converge toward competitive equilibria. The results reveal the challenges current LLMs face in replicating the dynamic decision-making processes characteristic of human trading behavior. Unlike humans, LLMs lacked the capacity to achieve market equilibrium. The research demonstrates that while LLMs provide a valuable tool for scalable and reproducible market simulations, their current limitations necessitate further advancements to fully capture the complexities of market behavior. Future work that enhances dynamic learning capabilities and incorporates elements of behavioral economics could improve the effectiveness of LLMs in the economic domain, providing new insights into market dynamics and aiding in the refinement of economic policies. |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2409.08357 |
| By: | Keller, Bryan; Wong, Vivian C (University of Virginia); Park, Sangbaek; Zhang, Jingru; Sheehan, Patrick; Steiner, Peter M. |
| Abstract: | Within-study comparisons (WSCs) use real, rather than simulated, data to compare estimates from observational studies against a benchmark randomized controlled trial (RCT). A primary goal of WSCs is to assess whether well-designed quasi-experimental designs (QEDs) can produce internally valid causal effect estimates comparable to those from RCTs. In this paper, we describe the design and implementation of a new type of WSC. Motivated by Shadish et al. (2008), we examine the impact of a mathematics training intervention and a vocabulary study session on posttest scores for mathematics and vocabulary, respectively. We extend the original design in three ways. First, before random assignment, we ask participants to express a preference for either the mathematics or vocabulary training session, after which they are randomly assigned regardless of preferences. This allows us to experimentally identify and estimate the overall average treatment effect (ATE) and two conditional ATEs: the average treatment effect on the treated (ATT) and the average treatment effect on the untreated (ATU). Second, participant recruitment and sample size (N = 2200) were determined through power analyses for comparing RCT and QED estimates, ensuring sufficient power for methodological comparisons. Finally, the study’s eligibility criteria, recruitment, treatment allocation, and analysis plan were preregistered on the Open Science Foundation platform, and the data are publicly accessible. We believe that this WSC design and the resulting data set will be valuable for researchers seeking to evaluate causal inference methods and test identification assumptions using real-world data. |
| Date: | 2024–09–10 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:2gur9 |
| By: | Jin, Shuxian; Spadaro, Giuliana (Vrije Universiteit Amsterdam); Balliet, Daniel |
| Abstract: | Cooperation underlies the ability of groups to realize collective benefits (e.g., creation of public goods). Yet, cooperation can be difficult to achieve when people face situations with conflicting interests between what is best for individuals versus the collective (i.e., social dilemmas). To address this challenge, groups can implement rules about structural changes in a situation. But what institutional rules can best facilitate cooperation? Theoretically, rules can be made to affect structural features of a social dilemma, such as the possible actions, outcomes, and people involved. We derived 13 pre-registered hypotheses from existing work and collected six decades of empirical research to test how nine structural features influence cooperation within prisoner’s dilemmas and public goods dilemmas. We do this by meta-analyzing mean levels of cooperation across studies (Study 1, k = 2, 340, N = 229, 528), and also examining how manipulations of these structural features in social dilemmas affect cooperation within studies (Study 2, k = 909). Results indicated that lower conflict of interests was associated with higher cooperation, and that (1) the implementation of sanctions (i.e., reward and punishment of behaviors) and (2) allowing for communication most strongly enhanced cooperation. However, we found inconsistent support for the hypotheses that group size and matching design affect cooperation. Other structural features (e.g., symmetry of dilemmas, sequential decision making, payment) were not associated with cooperation. Overall, these findings inform institutions that can (or not) facilitate cooperation. |
| Date: | 2024–09–02 |
| URL: | https://d.repec.org/n?u=RePEc:osf:osfxxx:9r2qb |
| By: | Javier Espinosa; William N. Evans; David C. Phillips; Tim Spilde |
| Abstract: | A new wave of social service programs aims to build a pathway out of poverty by helping clients define their own goals and then supporting them flexibly and intensively over multiple years to meet those goals. We conduct a randomized controlled trial of one such program. Participants randomly assigned to intensive, holistic, wrap-around services have 10 percentage points higher employment rates after one year compared with a control group offered only help with an immediate need. Most of this effect appears to persist after programming ends. However, we find limited evidence that intensive, holistic services affect areas beyond employment, even when other areas of life are participants’ primary goals. We find some evidence that the program works by increasing hopefulness and agency among participants, which may be more useful in supporting labor force participation than in meeting other goals. |
| JEL: | D91 I38 J22 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:32911 |
| By: | Ro’i Zultan (BGU); Aniol Llorente-Saguer (Department of Economics, Queen Mary University of London); Santiago Oliveros (Department of Economics, University of Bristol) |
| Keywords: | information acquisition, representative heuristic, base-rate neglect, laboratory experiment |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:bgu:wpaper:2411 |
| By: | Grund, Christian (RWTH Aachen University); Harbring, Christine (RWTH Aachen University); Klinkenberg, Lisa (RWTH Aachen University) |
| Abstract: | The organization of work and the characteristics of tasks have undergone considerable changes in recent years. The developments include (i) an increased relevance of virtual teams and (ii) a higher demand for non-routine tasks in organizations, including creativity. Existing research on creative teams focuses on one-shot or existing teams, overlooking the importance of the formation phase of teams. This formation phase is particularly relevant for teams working in a virtual workplace setting, where communication and coordination may be constrained by the environment. Next to virtual work, hybrid working models ascend, also for teams. Therefore, we examine the influence of workplace settings and changes in these settings on creative performance of teams. We also investigate whether the individuals' ability to choose their workplace affects creative performance. We answer those questions by conducting a 2-phase experiment with dyadic teams in the lab and online to model a presence and a virtual workplace setting and account for the formation phase of teams. We implemented the "Unusual-Uses Task" as non-routine creative task. Our results showed that teams working in presence outperform those working online. Interestingly, working at least one phase in presence induces higher creative performance than entirely working online, underscoring the relevance of hybrid workplace settings. Moreover, no significant effects of self-selection on performance were found. |
| Keywords: | teams, creativity, work from home, hybrid working models, self-selection |
| JEL: | C92 M5 |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17258 |
| By: | Ronak Jain |
| Abstract: | Street vending is an important source of self-employment for the urban poor. I combine observational, survey, and experimental data from Delhi to study this market. Partnering with vendors to randomize prices and passersby they solicit, I find that even with identical goods, child vendors are 97% more likely to make a sale and earn 2x more than adult vendors. Despite no differences in valuations, couples and women are 90% and 28% more likely to buy than men and they are targeted more often and quoted higher prices. I show that sellers strategically leverage insights about social preferences to influence buyer decision-making. |
| Keywords: | Social preferences, child labor, price discrimination, consumer behavior |
| JEL: | C93 D91 J46 L10 O17 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:zur:econwp:451 |
| By: | Jonathan Lafky; Robin Ng |
| Abstract: | We examine how product ratings are interpreted in the presence of heterogeneous preferences among both raters and consumers. Raters with altruistic motives should rate for the benefit of future consumers, however an ambiguity arises when preferences are heterogeneous. Multiple equilibria exist in which ratings may reflect the preferences of raters or the preferences of future consumers. In an online experiment, we examine how ratings are selected by raters and interpreted by consumers, and how information about rater preferences or product attributes can influence equilibrium selection. We show how both raters and consumers update their evaluation of what a rating represents in each environment, doing so in similar ways. |
| Keywords: | Ratings and Reviews, Altruism |
| JEL: | C91 D64 D83 L86 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:bon:boncrc:crctr224_2024_594 |
| By: | John List; Haruka Uchida |
| Abstract: | An unsettling stylized fact is that decorated early childhood education programs improve cognitive skills in the short-term, but lose their efficacy after a few years. We implement a field experiment with two stages of randomization to explore the underpinnings of the fade-out effect. We first randomly assign preschool access to children, and then partner with the local school district to randomly assign the same children to classmates throughout elementary school. We find that the fade-out effect is critically linked to the share of classroom peers assigned to preschool access-with enough treated peers the classic fade-out effect is muted. Our results highlight a paradoxical insight: while the fade-out effect has been viewed as a devastating critique of early childhood programs, our results highlight that fade-out is a key rational for providing early education to all children. This is because human capital accumulation is inherently a social activity, leading early education programs to deliver their largest benefits at scale when everyone receives such programs. |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:feb:framed:00797 |
| By: | Goulas, Sofoklis (Brookings Institution); Gunawardena, Bhagya N. (RMIT University); Megalokonomou, Rigissa (Monash University); Zenou, Yves (Monash University) |
| Abstract: | Using Greek administrative data, we examine the impact of being randomly assigned to a classroom with a same-gender top-performing student on both short- and long-term educational outcomes. These top performers are tasked with keeping classroom attendance records, which positions them as role models. Both male and female students are influenced by the performance of a same-gender top performer and experience both spillover and conformist effects. However, only female students show significant positive effects from the presence of a same-gender role model. Specifically, female students improved their science test scores by 4 percent of a standard deviation, were 2.5 percentage points more likely to choose a STEM track, and were more likely to apply for and enroll in a STEM university degree 3 years later. These effects were most pronounced in lower-income neighborhoods. Our findings suggest that same-gender peer role models could reduce the underrepresentation of qualified females in STEM fields by approximately 3 percent. We further validate our findings through a lab-in-the-field experiment, in which students rated the perceived influence of randomized hypothetical top-performer profiles. The results suggest that the influence of same-gender top performers is primarily driven by exposure-related factors (increased perception of distinction feasibility and self-confidence) rather than direct interactions. |
| Keywords: | role models, random peer group formation, natural experiment, lab-in-the-field experiment, gender gap, self-confidence, STEM |
| JEL: | J24 J16 I24 I26 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17271 |
| By: | Arthur B. Nelson; Dmitry Ryvkin |
| Abstract: | We study experimentally contests in which players make investment decisions sequentially, and information on prior investments is revealed between stages. Using a between-subject design, we consider all possible sequences in contests of three players and test two major comparative statics of the subgame-perfect Nash equilibrium: The positive effect of the number of stages on aggregate investment and earlier mover advantage. The former prediction is decidedly rejected, as we observe a reduction in aggregate investment when more sequential information disclosure stages are added to the contest. The evidence on earlier mover advantage is mixed but mostly does not support theory as well. Both predictions rely critically on large preemptive investment by first movers and accommodation by later movers, which does not materialize. Instead, later movers respond aggressively, and reciprocally, to first movers' investments, while first movers learn to accommodate those responses. |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2409.06230 |
| By: | Kipchumba, Elijah (Trinity College Dublin); Porter, Catherine (Lancaster University); Serra, Danila (Texas A&M University); Sulaiman, Munshi (BRAC University) |
| Abstract: | We evaluate the impact of a role model intervention on the gender attitudes, college aspirations and education outcomes of youths in Somalia. In 2018, we randomly selected elementary schools to receive a visit from a college student. Within each treatment school, we selected four grades, two to receive a visit from a female college student and two from a male college student. The "role models" gave unscripted talks about their personal study journeys, including challenges and strategies to overcome setbacks. Six months after the intervention we found a significant and large impact of (only) female role models on boys' and girls' attitudes toward gender equality but no impact on college aspirations. Data collected two and four years later from the cohorts graduating primary school produce smaller and non-significant treatment effects on the survey outcomes, but positive impacts on enrollment in high school and a lower probability of early marriage as reported by teachers. |
| Keywords: | role models, education, gender, aspirations, field experiment, Somalia |
| JEL: | J16 O12 I25 C93 |
| Date: | 2024–08 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17261 |
| By: | Ma, Mingye (University of Southampton); Riener, Gerhard (University of Southampton); Xu, Youzong (University of Nottingham Ningbo China) |
| Abstract: | We explore the role of self- and peer evaluations in education, with a particular emphasis on gender differences. We construct a model of (self-)deception to predict outcomes for scenarios with and without self-evaluation. By using unique data from a first-year economics class at a Sino-UK university, we examine how students assess their own and their peers' contributions to group projects under varying self-assessment conditions. Our findings reveal a significant self-serving bias across both genders, though with subtle distinctions. Women, despite greater societal recognition, exhibit smaller self-social evaluation gaps (SSEG). The variation in abstention rates between treatments is mainly attributed to lowerperforming males. These findings indicate that the possibility of self-assessment influences rating behavior, potentially exacerbating gender disparities and affecting gender equity. |
| Keywords: | higher education, incentives, field experiment, peer evaluation, gender |
| JEL: | D01 D91 I23 C93 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:iza:izadps:dp17267 |
| By: | Ana Costa-Ramón; Ursina Schaede; Michaela Slotwinski; Anne Ardila Brenøe |
| Abstract: | The “child penalty” significantly reduces women’s lifetime earnings and pension savings, but it remains unclear whether these gaps are the deliberate result of forward-looking decisions. This paper provides novel evidence on the role of information constraints in mothers’ labor supply decisions. We first document descriptively that mothers are largely inattentive to the long-term financial consequences of reduced hours. In a large-scale field experiment that combines rich survey and administrative data, we then provide mothers with objective, individualized information about the long-run costs of reduced labor supply. The treatment increases demand for financial information and future labor supply plans, in particular among women who underestimate the long-term costs. Leveraging linked employer administrative data one year post-intervention, we observe that mothers who underestimate the long-term costs increase their labor supply by 6 percent over the mean. |
| JEL: | J16 J22 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:zur:econwp:452 |
| By: | Inbal Dekel; Rachel Cummings; Ori Heffetz; Katrina Ligett |
| Abstract: | Privacy considerations and their effects on behavior are becoming increasingly important. Yet the extremes of full and no privacy are rarely an option. How much does behavior change with small changes in privacy? Dekel et al. (2023) introduce the concept of privacy elasticity, the responsiveness of economic variables to small changes in privacy protections. This concept combines elasticity—a key economic measure of responsiveness of one variable to changes in another—and differential privacy—a computer science theory emerging as the standard tool for protecting and quantifying privacy. Together, they create a measure of privacy elasticity that is portable and comparable across contexts. The applicability of this concept is demonstrated by reviewing how privacy elasticity can be estimated in a public-good lab experiment. |
| JEL: | C91 D82 Z00 |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:32903 |
| By: | Ali Merali |
| Abstract: | This paper derives 'scaling laws' -- empirical relationships between the amount of training compute used for a Large Language Model (LLM) and its performance -- for economic outcomes. In a preregistered experiment, 300 professional translators completed 1800 tasks with access to one of thirteen LLMs with differing model training compute sizes (or a control). Our results show that model scaling substantially raises productivity: for every 10x increase in model compute, translators completed tasks 12.3% quicker, received 0.18 s.d. higher grades, and earned 16.1% more per minute (including bonus payments). Further, the gains from model scaling are much higher for lower-skilled workers who gain a 4x larger improvement in task completion speed. These results imply further frontier model scaling -- which is currently estimated at 4x increase per year -- may have significant economic implications. |
| Date: | 2024–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2409.02391 |
| By: | De La Cruz, Marjorie; Abi-Hassan, Sahar; Denly, Michael |
| Abstract: | We provide a reproduction and replication of Brutger (2024), which examines the effects of the University of California, Berkeley's Pipeline Initiative in Political Science (PIPS) program on five self-reported outcomes related to interest and preparation towards pursuing graduate school. We are able to reproduce the author's results but do note some minor coding challenges. Our additional replication analysis confirms that the study's original results are robust to different model specifications. In future analysis of PIPS, we suggest that the author address our suggestions regarding the wording of the survey questions, sample selection, and statistical power. Overall, we commend the author on a good study of an important topic. |
| Keywords: | Diversity, Equity, Inclusion, DEI, Political Science, Graduate School, Program Evaluation, Randomized Controlled Trial |
| JEL: | D63 M14 I24 I23 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:i4rdps:152 |
| By: | Samuel Chang; Andrew Kennedy; Aaron Leonard; John List |
| Abstract: | We provide twelve best practices and discuss how each practice can help researchers accurately, credibly, and ethically use Generative AI (GenAI) to enhance experimental research. We split the twelve practices into four areas. First, in the pre-treatment stage, we discuss how GenAI can aid in pre-registration procedures, data privacy concerns, and ethical considerations specific to GenAI usage. Second, in the design and implementation stage, we focus on GenAI's role in identifying new channels of variation, piloting and documentation, and upholding the four exclusion restrictions. Third, in the analysis stage, we explore how prompting and training set bias can impact results as well as necessary steps to ensure replicability. Finally, we discuss forward-looking best practices that are likely to gain importance as GenAI evolves. |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:feb:artefa:00796 |
| By: | Margherita Comola (Université Paris-Saclay (RITM), PSE - Paris School of Economics - UP1 - Université Paris 1 Panthéon-Sorbonne - ENS-PSL - École normale supérieure - Paris - PSL - Université Paris Sciences et Lettres - EHESS - École des hautes études en sciences sociales - ENPC - École des Ponts ParisTech - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement); Agnieszka Rusinowska (CES - Centre d'économie de la Sorbonne - UP1 - Université Paris 1 Panthéon-Sorbonne - CNRS - Centre National de la Recherche Scientifique, PSE - Paris School of Economics - UP1 - Université Paris 1 Panthéon-Sorbonne - ENS-PSL - École normale supérieure - Paris - PSL - Université Paris Sciences et Lettres - EHESS - École des hautes études en sciences sociales - ENPC - École des Ponts ParisTech - CNRS - Centre National de la Recherche Scientifique - INRAE - Institut National de Recherche pour l’Agriculture, l’Alimentation et l’Environnement); Marie Claire Villeval (GATE Lyon Saint-Étienne - Groupe d'Analyse et de Théorie Economique Lyon - Saint-Etienne - UL2 - Université Lumière - Lyon 2 - UJM - Université Jean Monnet - Saint-Étienne - EM - EMLyon Business School - CNRS - Centre National de la Recherche Scientifique, IZA - Forschungsinstitut zur Zukunft der Arbeit - Institute of Labor Economics) |
| Abstract: | We experimentally investigate how players with opposing views compete for influence through strategic targeting in networks. We varied the network structure, the relative influence of the opponent, and the heterogeneity of the nodes' initial opinions. Although most players adopted a best-response strategy based on their relative influence, we also observed behaviors deviating from this strategy, such as the tendency to target central nodes and avoid nodes targeted by the opponent. Targeting is also affected by affinity and opposition biases, the strength of which depends on the distribution of initial opinions. |
| Keywords: | Network, Influence, Targeting, Competition, Experiment |
| Date: | 2024–09–23 |
| URL: | https://d.repec.org/n?u=RePEc:hal:cesptp:hal-04706311 |
| By: | Gautam, Sanghmitra; Gechter, Michael; Guiteras, Raymond P.; Mobarak, Ahmed Mushfiq |
| Abstract: | We conduct an organized review of intervention-based studies that aim to promote improved sanitation adoption and use RCTs for evaluation. We impose systematic inclusion criteria to identify such studies, and compile their microdata to harmonize outcome and covariate measures as well as estimands across studies. We then re-analyze their data to report metrics that are consistently defined and measured across studies. We compare the relative effectiveness of different classes of interventions implemented in overlapping ways across four countries: community-level demand encouragement, sanitation subsidies, product information campaigns, and offering microcredit to finance product purchases. Interventions with financial benefits generally outperform information and education campaigns. Effects are typically larger for households with higher shares of women and differ little by poverty status, but more research is needed to confirm our conclusions on effect heterogeneity by household characteristics. |
| Keywords: | Consumer/Household Economics |
| Date: | 2024–01 |
| URL: | https://d.repec.org/n?u=RePEc:ags:nccewp:340057 |