A concrete example of construct construction in natural language

Yeomans, Michael

doi:10.1016/j.obhdp.2020.10.008

Cited by 26 publications

(21 citation statements)

References 118 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To measure the content concreteness of cleaned tweets, we relied on the R package doc2concrete [ 63 ], which uses a dictionary of 40,000 common English words and expressions [ 64 ]. The concrete score has a range of 0-5, where 0 is abstract and 5 is concrete.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

The Association Between Dissemination and Characteristics of Pro-/Anti-COVID-19 Vaccine Messages on Twitter: Application of the Elaboration Likelihood Model

Saini¹,

Liang²,

Yang³

et al. 2022

JMIR Infodemiology

View full text Add to dashboard Cite

Background Messages on one’s stance toward vaccination on microblogging sites may affect the reader’s decision on whether to receive a vaccine. Understanding the dissemination of provaccine and antivaccine messages relating to COVID-19 on social media is crucial; however, studies on this topic have remained limited. Objective This study applies the elaboration likelihood model (ELM) to explore the characteristics of vaccine stance messages that may appeal to Twitter users. First, we examined the associations between the characteristics of vaccine stance tweets and the likelihood and number of retweets. Second, we identified the relative importance of the central and peripheral routes in decision-making on sharing a message. Methods English-language tweets from the United States that contained provaccine and antivaccine hashtags (N=150,338) were analyzed between April 26 and August 26, 2021. Logistic and generalized negative binomial regressions were conducted to predict retweet outcomes. The content-related central-route predictors were measured using the numbers of hashtags and mentions, emotional valence, emotional intensity, and concreteness. The content-unrelated peripheral-route predictors were measured using the numbers of likes and followers and whether the source was a verified user. Results Content-related characteristics played a prominent role in shaping decisions regarding whether to retweet antivaccine messages. Particularly, positive valence (incidence rate ratio [IRR]=1.32, P=.03) and concreteness (odds ratio [OR]=1.17, P=.01) were associated with higher numbers and likelihood of retweets of antivaccine messages, respectively; emotional intensity (subjectivity) was associated with fewer retweets of antivaccine messages (OR=0.78, P=.03; IRR=0.80, P=.04). However, these factors had either no or only small effects on the sharing of provaccine tweets. Retweets of provaccine messages were primarily determined by content-unrelated characteristics, such as the numbers of likes (OR=2.55, IRR=2.24, P<.001) and followers (OR=1.31, IRR=1.28, P<.001). Conclusions The dissemination of antivaccine messages is associated with both content-related and content-unrelated characteristics. By contrast, the dissemination of provaccine messages is primarily driven by content-unrelated characteristics. These findings signify the importance of leveraging the peripheral route to promote the dissemination of provaccine messages. Because antivaccine tweets with positive emotions, objective content, and concrete words are more likely to be disseminated, policymakers should pay attention to antivaccine messages with such characteristics.

show abstract

Section: Methodsmentioning

confidence: 99%

“…To measure the content concreteness of cleaned tweets, we relied on the R package doc2concrete [63], which uses a dictionary of 40,000 common English words and expressions…”

Section: Concretenessmentioning

confidence: 99%

The Association Between Dissemination and Characteristics of Pro-/Anti-COVID-19 Vaccine Messages on Twitter: Application of the Elaboration Likelihood Model

Saini¹,

Liang²,

Yang³

et al. 2022

JMIR Infodemiology

View full text Add to dashboard Cite

show abstract

“…Linguistic abstraction was also measured using the Brysbaert Concreteness Index (BCI; Brysbaert et al, 2014;Yeomans, 2021). The BCI index is the average abstractness and concreteness scores of words in an essay commonly used in research examining language abstractness (e.g., Bhatia & Walasek, 2016;Yin et al, 2022;Joshi et al, 2020).…”

Section: Linguistic Abstraction (Use Of Abstract Versus Concrete Lang...mentioning

confidence: 99%

Empirical Investigation of the Relationship between Construals of Structural Racism and Attitudes toward Critical Race Theory

Allen¹,

Brown

HIGGINBOTHAM³

et al. 2023

Preprint

View full text Add to dashboard Cite

An uptick in U.S. White Americans’ structural understanding of racism alongside opposition to critical race theory (CRT) and acknowledgement of the historical and pervasiveness of racism reflects a social paradox. How is it that the U.S. White Americans can seemingly acknowledge structural racism without supporting its logical corollary arguments? CRT itself might have the answer in the form of interest convergence. CRT posits that U.S. White Americans only support aspects of anti-racist ideology that benefit their group, suggesting that something about supporting structural racism is beneficial. In two studies (N = 968), we test whether differences in the construal of structural racism in concrete versus abstract ways can help explain this paradox, assuming that construing structural racism in more abstract ways absolves the self, and one’s group, from racial wrongdoing. Across multiple operationalizations of abstraction (i.e.., the use of “I”, “we”, Brysbaert Concreteness Index, and a concreteness manipulation) we find partial evidence that abstraction leads to a reduction in support for the historic and endemic nature of racism for U.S. White but not Black Americans. We discuss the implications of our findings for the future of CRT and the fight for justice.

show abstract

“…Understanding and mitigating bias in ML models is vital, since they are now a quintessential part of our everyday life. It is easy to understand why these models have become so valuable: they serve as helpful tools that allow us to identify and synthesize complex patterns in large datasets, and often outperform traditional statistical methods in prediction and classification across a variety of tasks (e.g., Goretzko & Bühner, 2020;Yeomans, 2021). To that end, they promise to simplify decision-making in a seemingly unbiased manner, making them instrumental in guiding the choices of powerholders and laypeople alike.…”

mentioning

confidence: 99%

“…These models have appeared in research examining how facial recognition software predicts users' sexuality (Wang & Kosinski, 2018), what jobs people are likely to opt into (Song et al, 2022), and how moral rhetoric online during protests predicts both future arrests (Mooijman et al, 2018) as well as future outrage (Brady et al, 2021). Scholars have also advocated for using these tools to advance our theory (e.g., Leavitt et al, 2021;Yarkoni & Westfall, 2017) and provide more robust ways of analyzing our data to reduce issues associated with traditional statistical models (e.g., Garten et al, 2019;Goretzko & Bühner, 2020;Yeomans, 2021). However, this path is rockier than it seems.…”

mentioning

confidence: 99%

The sins of the parents are to be laid upon the children: biased humans, biased data, biased models

Osborne¹,

Omrani²,

Dehghani³

2022

Preprint

View full text Add to dashboard Cite

Technological innovations have become a key driver of societal advancements. Nowhere is this more evident than in the field of machine learning (ML), which has developed algorithmic models that shape our decisions, behaviors, and outcomes. These tools have widespread use, in part, because they can synthesize massive amounts of data to make seemingly objective recommendations. Yet, in the past few years, the ML community has been raising the alarm on why we should be cautious in interpreting and using these models: they are created by humans, from data generated by humans, whose psychology allows for various biases that impact how the models are developed, trained, tested and interpreted. As psychologists, we thus face a fork in the road; Down the first path, we can continue to use these models without examining and addressing these critical flaws, and rely on computer scientists to try to mitigate them. Down the second path, we can turn our expertise in bias towards this growing field, collaborating with computer scientists to mitigate the deleterious outcomes associated with these models. This paper serves to light the way down the second path by identifying how extant psychological research can help examine and mitigate bias in ML models.

show abstract

A concrete example of construct construction in natural language

Cited by 26 publications

References 118 publications

The Association Between Dissemination and Characteristics of Pro-/Anti-COVID-19 Vaccine Messages on Twitter: Application of the Elaboration Likelihood Model

The Association Between Dissemination and Characteristics of Pro-/Anti-COVID-19 Vaccine Messages on Twitter: Application of the Elaboration Likelihood Model

Empirical Investigation of the Relationship between Construals of Structural Racism and Attitudes toward Critical Race Theory

The sins of the parents are to be laid upon the children: biased humans, biased data, biased models

Contact Info

Product

Resources

About