Using Machine Learning to Generate Novel Hypotheses: Increasing Optimism About COVID-19 Makes People Less Willing to Justify Unethical Behaviors

Sheetal, Abhishek; Feng, Zhiyu; Savani, Krishna

doi:10.1177/0956797620959594

Cited by 38 publications

(46 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, others N-cycle organisms, such as the free-living ammonia-oxidizing bacteria, were also associated with soil chemical fertility ( Ciccolini et al, 2016 ). Therefore, regularization is also a tool for suggesting hypothesis testing ( Sheetal et al, 2020 ), which future controlled studies should achieve.…”

Section: Discussionmentioning

confidence: 99%

Environmental Factors Associated With Nitrogen Fixation Prediction in Soybean

et al. 2021

View full text Add to dashboard Cite

Biological nitrogen (N)-fixation is the most important source of N for soybean [Glycine max (L.) Merr.], with considerable implications for sustainable intensification. Therefore, this study aimed to investigate the relevance of environmental factors driving N-fixation and to develop predictive models defining the role of N-fixation for improved productivity and increased seed protein concentration. Using the elastic net regularization of multiple linear regression, we analyzed 40 environmental factors related to weather, soil, and crop management. We selected the most important factors associated with the relative abundance of ureides (RAU) as an indicator of the fraction of N derived from N-fixation. The most relevant RAU predictors were N fertilization, atmospheric vapor pressure deficit (VPD) and precipitation during early reproductive growth (R1–R4 stages), sowing date, drought stress during seed filling (R5–R6), soil cation exchange capacity (CEC), and soil sulfate concentration before sowing. Soybean N-fixation ranged from 60 to 98% across locations and years (n = 95). The predictive model for RAU showed relative mean square error (RRMSE) of 4.5% and an R2 value of 0.69, estimated via cross-validation. In addition, we built similar predictive models of yield and seed protein to assess the association of RAU and these plant traits. The variable RAU was selected as a covariable for the models predicting yield and seed protein, but with a small magnitude relative to the sowing date for yield or soil sulfate for protein. The early-reproductive period VPD affected all independent variables, namely RAU, yield, and seed protein. The elastic net algorithm successfully depicted some otherwise challenging empirical relationships to assess with bivariate associations in observational data. This approach provides inference about environmental variables while predicting N-fixation. The outcomes of this study will provide a foundation for improving the understanding of N-fixation within the context of sustainable intensification of soybean production.

show abstract

Section: Discussionmentioning

confidence: 99%

Environmental Factors Associated With Nitrogen Fixation Prediction in Soybean

et al. 2021

View full text Add to dashboard Cite

show abstract

“…However, the machine learning approach also identified multiple other aspects of cultural change that have not received much attention, including Protestant work ethic, political orientation, political action, and prosociality. Thus, the 30 current research indicates that theory-blind machine learning approaches can complement traditional theory-driven approaches to generate novel insights that researchers might miss otherwise (Bleidorn & Hopwood, 2019;Sheetal et al, 2020).…”

Section: Discussionmentioning

confidence: 98%

A machine learning model of cultural change: Role of prosociality, political attitudes, and Protestant work ethic.

Sheetal¹,

Savani²

2021

American Psychologist

Self Cite

View full text Add to dashboard Cite

What attitudes, values, and beliefs serve as key markers of cultural change? To answer this question, we examined 221,485 respondents from the World Values Survey, a multi-wave cross-country survey of people's attitudes, values and beliefs. We trained a machine learning model to classify respondents into seven waves (i.e., periods). Once trained, the machine learning model identified a separate group of 24,611 respondents' wave with a balanced accuracy of 77%. We then queried the model to identify the attitudes, values, and beliefs that contributed the most to its classification decisions, and therefore, served as markers of cultural change. These included religiosity, social attitudes, political attitudes, independence, life satisfaction, Protestant work ethic, and prosociality. Although past research in cultural change has discussed decreasing religiosity and increasing liberalism and independence, it has not yet identified Protestant work ethic, political orientation, and prosociality as values relevant to cultural change. Thus, the current research points to new directions for future research on cultural change that might not be evident from either a deductive or an inductive approach. This research illustrates that the abductive approach of machine learning, which focuses on the most likely explanations for an outcome, can provide novel insights. Public SignificanceStatement. This research found that in recent years, people around the world have been becoming less religious and more liberal in their social attitudes and political orientation. People have been valuing independence more, although there appears to be a decline in independence in the last few years. The extent to which people emphasize hard work, thrift, and acting in a prosocial manner has also declined in recent years.

show abstract

“…In other words, the number of parameters (p) needs to be smaller than our sample size (n) (Faraway, 2014). In ML applications in psychology, there are often many more parameters to be estimated than the sample size (i.e., p>>n) (e.g., Joel et al, 2020;Sheetal, Feng, & Savani, 2020). Machine learning algorithms can handle p>>n and address overfitting when there are many predictors (Putka, Beaty, & Reeder, 2018).…”

Section: Inputsmentioning

confidence: 99%

A Conceptual Framework for Investigating and Mitigating Machine Learning Measurement Bias (MLMB) in Psychological Assessment

Tay¹,

Woo²,

Hickman³

et al. 2021

Preprint

View full text Add to dashboard Cite

Given significant concerns about fairness and bias in the use of artificial intelligence (AI) and machine learning (ML) for assessing psychological constructs, we provide a conceptual framework for investigating and mitigating machine learning measurement bias (MLMB) from a psychometric perspective. MLMB is defined as differential functioning of the trained ML model between subgroups. MLMB can empirically manifest when a trained ML model produces different predicted score levels for individuals belonging to different subgroups (e.g., race, gender) despite them having the same ground truth level for the underlying construct of interest (e.g., personality), and/or when the model yields differential predictive accuracies across the subgroups. Because the development of ML models involves both data and algorithms, both biased data and algorithm training bias are potential sources of MLMB. Data bias can occur in the form of nonequivalence between subgroups in the ground truth, platform-based construct, behavioral expression, and/or feature computing. Algorithm training bias can occur when algorithms are developed with nonequivalence in the relation between extracted features and ground truth (i.e., algorithm features are differentially used, weighted, or transformed between subgroups). We explain how these potential sources of bias may manifest during ML model development and share initial ideas on how to mitigate them, recognizing that the development of new statistical and algorithmic procedures will need to follow. We also discuss how this framework brings clarity to MLMB but does not reduce the complexity of the issue.

show abstract

Using Machine Learning to Generate Novel Hypotheses: Increasing Optimism About COVID-19 Makes People Less Willing to Justify Unethical Behaviors

Cited by 38 publications

References 32 publications

Environmental Factors Associated With Nitrogen Fixation Prediction in Soybean

Environmental Factors Associated With Nitrogen Fixation Prediction in Soybean

A machine learning model of cultural change: Role of prosociality, political attitudes, and Protestant work ethic.

A Conceptual Framework for Investigating and Mitigating Machine Learning Measurement Bias (MLMB) in Psychological Assessment

Contact Info

Product

Resources

About