Skip to main content
eScholarship
Open Access Publications from the University of California

Published articles of faculty and affiliated researchers at the Department of Economics, University of California at Santa Barbara.

Cover page of Perceptions and detection of AI use in manuscript preparation for academic journals

Perceptions and detection of AI use in manuscript preparation for academic journals

(2024)

The rapid advances in Generative AI tools have produced both excitement and worry about how AI will impact academic writing. However, little is known about what norms are emerging around AI use in manuscript preparation or how these norms might be enforced. We address both gaps in the literature by conducting a survey of 271 academics about whether it is necessary to report ChatGPT use in manuscript preparation and by running GPT-modified abstracts from 2,716 published papers through a leading AI detection software to see if these detectors can detect different AI uses in manuscript preparation. We find that most academics do not think that using ChatGPT to fix grammar needs to be reported, but detection software did not always draw this distinction, as abstracts for which GPT was used to fix grammar were often flagged as having a high chance of being written by AI. We also find disagreements among academics on whether more substantial use of ChatGPT to rewrite text needs to be reported, and these differences were related to perceptions of ethics, academic role, and English language background. Finally, we found little difference in their perceptions about reporting ChatGPT and research assistant help, but significant differences in reporting perceptions between these sources of assistance and paid proofreading and other AI assistant tools (Grammarly and Word). Our results suggest that there might be challenges in getting authors to report AI use in manuscript preparation because (i) there is not uniform agreement about what uses of AI should be reported and (ii) journals might have trouble enforcing nuanced reporting requirements using AI detection tools.

Cover page of Father Absence and the Educational Gender Gap

Father Absence and the Educational Gender Gap

(2022)

The educational attainment of young women now exceeds that of young men in most of the developed world, and women account for about 60% of new four-year college graduates in the United States. Several studies have suggested that the increase in single-parent households may be contributing to the growing gender gap in education, as boys are more vulnerable to the negative effects of father absence and economic disadvantage than girls. Using data on recent cohorts of young men and women from the National Longitudinal Study of Adolescent to Adult Health (Add Health), I find evidence consistent with other studies that boys are relatively more likely to experience problems in school, including school suspensions, when their father is absent, but also that girls are relatively more likely to experience depression in adolescence, particularly in step-father families. By the time Add Health subjects are young adults, there is no evidence that father absence early in life is more strongly associated with lower rates of college graduation for men, compared to women, in either cross-sectional or family fixed-effect models.

Educational Inequality and the Returns to Skills

(2022)

Research and policy discussion about the diverging fortunes of children from advantaged and disadvantaged households have focused on the skill disparities between these children – how they might arise and how they might be remediated. Analysis of data from the National Longitudinal Study of Adolescent Health reveals another important mechanism in the determinants of educational attainment – differential returns to skills for children in different circumstances. Though the returns to cognitive ability are generally consistent across family background groups, personality traits have very different effects on educational attainment for young men and women with access to different levels of parental resources. These results are consistent with a model in which the provision of focused effort in school is complementary with parental inputs while openness, associated with imagination and exploration, is a substitute for information provision by educated parents and thus contributes to resilience in low-resource environments. In designing interventions to improve outcomes for disadvantaged children, we need to be cognizant of interactions between a child's skills and their circumstances.

Cohabitation and the Uneven Retreat from Marriage in the U.S., 1950-2010

(2022)

Since 1950 the sources of the gains from marriage have changed radically. As the educational attainment of women overtook and surpassed that of men and the ratio of men's to women's wage rates fell, traditional patterns of gender specialization in work weakened. The primary source of the gains to marriage shifted from the production of household services and commodities to investment in children. For some, these changes meant that marriage was no longer worth the costs of limited independence and potential mismatch. Cohabitation became an acceptable living arrangement for all groups, but cohabitation serves different functions among different groups. The poor and less educated are much more likely to rear children in cohabitating relationships. The college educated typically cohabit before marriage, but they marry before conceiving children and their marriages are relatively stable. We argue that different patterns of childrearing are the key to understanding class differences in marriage and parenthood, not an unintended by-product of it. Marriage is the commitment mechanism that supports high levels of investment in children and is hence more valuable for parents adopting a high-investment strategy for their children.

Cover page of Can targeted messages reduce COVID-19 vaccination hesitancy? A randomized trial

Can targeted messages reduce COVID-19 vaccination hesitancy? A randomized trial

(2022)

Background

Widespread vaccination is certainly a critical element in successfully fighting the COVID-19 pandemic. We apply theories of social identity to design targeted messaging to reduce vaccine hesitancy among groups with low vaccine uptake, such as African Americans and political conservatives.

Methods

Participants. We conducted an online experiment from April 7 to 27, 2021, that oversampled Black, Latinx, conservative, and religious U.S. residents. We first solicited the vaccination status of over 10,000 individuals. Of the 4,609 individuals who reported being unvaccinated, 4,190 enrolled in our covariate-adaptive randomized trial.Interventions. We provided participants messages that presented the health risks of COVID-19 to oneself and others; they also received messages about the benefits of a COVID-19 vaccine and an endorsement by a celebrity. Messages were randomly tailored to each participant's identities-Black, Latinx, conservative, religious, or being a parent.Outcomes. Respondents reported their intent to obtain the vaccine for oneself and, if a parent, for one's child.

Results

We report results for the 2,621 unvaccinated respondents who passed an incentivized manipulation check. We find no support for the hypothesis that customized messages or endorsers reduce vaccine hesitancy among our segments. A post hoc analysis finds evidence that a vaccine endorsement from Dr. Fauci reduces stated intent to vaccinate among conservatives.

Conclusions

We find no evidence that tailoring public-health communication regarding COVID-19 vaccination for broad demographic groups would increase its effectiveness. We recommend further research on communicators and endorsers, as well as incentives.