1 Automated Syllabus of Empirical Papers
Built by Rex W. Douglass @RexDouglass ; Github ; LinkedIn
Papers curated by hand, summaries and taxonomy written by LLMs.
2 COVID-19
2.1 Covid Pandemic
Be aware of the limitations of serological tests for SARS-CoV-2 infection, including variations in sensitivity and specificity depending on the timing of testing and the type of test used, and consider combining multiple sources of information (such as PCR testing, serological test results, and self-reported symptoms) to improve the accuracy of prevalence estimates. (Cordioli et al. 2022)
Consider using Stochastic Frontier Analysis (SFA) to estimate models suffering from one-sided or skewed measurement error, specifically for assessing undercounting in COVID-19 cases and deaths. (Millimet and Parmeter 2022)
Carefully consider the implications of combining different types of tests (e.g., LFD and PCR) when calculating positivity rates, as they may have different sensitivity and specificity, leading to biased estimates. (Ghebremichael‐Weldeselassie et al. 2022)
Consider using Bayesian dynamic multilevel regression and poststratification to analyze complex datasets, particularly when dealing with small sample sizes or non-representative data, as it provides stable estimates and reduces bias compared to traditional survey-weighting methods. (Pouwels et al. 2021)
Carefully consider and document the sources of your data, as significant discrepancies can exist between different data sources, even when they purportedly measure the same construct. (Smith 2021)
Consider using Autoregressive Integrated Moving Average (ARIMA) models for time series analysis when dealing with non-stationary data, as demonstrated by the successful use of ARIMA (0, 1, 4) in accurately predicting the trend of COVID-19 infections among unvaccinated individuals before and after the start of a vaccination program. (Daniel et al. 2021)
Use nationally representative surveys to assess public understanding and compliance with public health messaging, and consider using open-text reflections to gain deeper insight into participants perspectives. (Barari et al. 2020)
Consider incorporating multiple sources of information, especially household and community exposure, beyond symptoms alone when developing models to identify infected individuals from self-reported data. (Allen et al. 2020)
Consider using statistical models to analyze patterns in disease importation based on travel data, specifically by employing generalized linear regression models to identify locations with potential underreporting of cases. (Salazar et al. 2020)
Consider using a Bayesian, regularized, trimmed meta-regression approach called MR-BRT to analyze complex datasets with between-study heterogeneity, as it allows for relaxing the assumption of a single true point estimate and incorporating random effects with a gamma term to account for variability in the data. (Hay 2020)
Consider using simulations instead of mathematical inference when studying rapidly spreading diseases like COVID-19 because simulations can better capture real-world complexities and provide a more accurate picture of the diseases potential impact. (D. Li et al. 2020)
Carefully consider the role of undetected or asymptomatic infections in driving disease transmission, especially in the context of emerging diseases like COVID-19 where such infections may constitute a significant proportion of total cases and contribute significantly to the basic reproductive number (R0). (R. Li et al. 2020)
Consider using Bayesian networks (BNs) to analyze COVID-19 data, as BNs allow for the incorporation of multiple sources of data and the quantification of uncertainty in a coherent framework. (Neil et al. 2020)
Consider incorporating endogenous behavioral responses to risk when modeling infectious diseases, as they can significantly alter the dynamics and produce more accurate projections. (Rahmandad, Lim, and Sterman 2020)
Consider using a Bayesian hierarchical semi-mechanistic framework to model complex phenomena such as the spread of COVID-19, allowing them to incorporate multiple sources of uncertainty and make accurate predictions based on limited data. (Unwin et al. 2020)
Carefully consider the impact of variability in case reporting times when estimating the serial interval distribution, as it can lead to significant smoothing effects in the time series of estimated reproduction numbers and potentially biased estimates if reporting practices change over time. (Worden et al. 2020)
Consider collecting real-time data on social behaviors during a pandemic, as traditional post-facto data may not accurately capture the nuanced dynamics of human interactions, particularly in relation to social distancing practices. (Marchiori 2020)
Consider both incomplete testing and imperfect diagnostic accuracy when estimating the true burden of infectious diseases like SARS-CoV-2, as these factors can lead to significant underestimates of actual infections. (Wu et al. 2020)
Carefully consider and address potential biases in reported COVID-19 deaths before using them in your analyses, such as by incorporating data on COVID-19 cases and hospitalizations as leading indicators of deaths. (“Modeling COVID-19 Scenarios for the United States” 2020)
Exercise extreme caution when interpreting observational studies due to the potential for confounding variables, autocorrelation, and the risk of mistaking correlation for causation, particularly in light of the complex and dynamic nature of COVID-19 spread and the limitations of existing data sources. (Douglass, Scherer, and Gartzke 2020)
Prioritize molecular diagnostic testing for SARS-CoV-2 for all patients with acute respiratory illness, especially during periods of co-circulation, to accurately differentiate between influenza and COVID-19, guide patient care, and inform public health interventions. (Belongia and Osterholm 2020)
2.2 Vaccine Efficacy
Use propensity score methods like Marginal Structural Models (MSM) with inverse probability weighting (IPW) to account for confounding when estimating causal effects in observational studies, particularly when evaluating the impact of vaccination on disease outcomes. (Belayneh et al. 2023)
Consider using a test negative case control design to estimate vaccine effectiveness, as it is a powerful study design that has high concordance with randomized controlled trials and can help control for factors that are typically difficult to estimate in observational studies. (Bernal et al. 2021)
Use appropriate statistical methods, such as Poisson and Cox regressions, to account for confounding factors and accurately estimate vaccine effectiveness in observational studies. (Moustsen-Helms et al. 2021)
Consider using multiple methods, including observational studies and randomized controlled trials, to investigate the effectiveness of COVID-19 vaccines in reducing asymptomatic viral carriage and other proxies of possible transmission, such as cycle threshold (Ct) values and viral load. (Abbas et al. 2021)
Carefully consider the potential benefits and risks of interventions at different levels of exposure to infection, as exposure affects the likelihood of adverse outcomes and thus the balance of benefits and risks. (Gurdasani et al. 2021)
2.3 Mask Study
Carefully consider and address potential sources of bias, including self-reporting biases, when evaluating the effectiveness of public health interventions. (Abaluck et al. 2022)
Consider conducting experiments to directly examine the impact of mask-wearing on the spread of respiratory droplets, particularly in enclosed spaces like public transportation, where the risk of transmission is higher. (Ku et al. 2021)
Carefully consider the potential for selection biases introduced by unblinded staff during participant recruitment, as such biases can lead to significant imbalances in denominators between treatment and control groups, potentially affecting the validity of statistical inferences drawn from the study. (McCoy 2017)
2.4 Treatment for Covid
Consider using inverse probability of treatment weighting (IPTW) to emulate randomization and balance differences in baseline variables between treatment groups when analyzing observational data. (Mahévas et al. 2020)
Carefully consider the safety implications of administering high doses of chloroquine for treating COVID-19, as the study found that a higher dosage of CQ (12 g total dose over 10 days) in COVID-19 should not be recommended due to safety concerns regarding QTc prolongation and increased lethality, particularly among older patients taking medications like azithromycin and oseltamivir that also prolong Q (Borba et al. 2020)
Employ a multicenter, open-label, randomized controlled trial design to evaluate the efficacy and safety of hydroxychloroquine (HCQ) in treating COVID-19 patients, while carefully considering potential confounders and adverse events. (Tang et al. 2020)
2.5 Covid Testing
Consider using both PCR and rapid antigen tests concurrently in your studies to accurately capture the early infectious period of Omicron variant cases, as rapid antigen tests alone may produce false negatives during this critical window. (Adamson et al. 2022)
Consider using antigen tests as a practical alternative to RT-PCR for identifying contagious individuals, particularly within the first five days of symptom onset, as antigen tests demonstrate high sensitivity (96.2%) and negative predictive value (99.1%) for contagiousness, while being faster, cheaper, and easier to administer than traditional methods like viral culture. (Lopera et al. 2021)
2.6 Effects of Covid
Utilize natural experiments, like the unexpected cancellation of professional sports games due to the COVID-19 pandemic, to identify the causal effects of mass gatherings on disease spread. (Ahammer, Halla, and Lackner 2023)
Utilize a two-step process when analyzing the causal effects of lockdown policies on health and macroeconomic outcomes: first, estimate an epidemiological model with time-varying parameters to capture changes in individual behavior and shifts in disease transmission, then use the outputs from this model to conduct a causal assessment of nonpharmaceutical policy interventions using techniques such as structural vector autoregression (SVAR) and local projections (Plagborg-Møller and Wolf 2020)
2.7 Policy Impact on Covid
Consider incorporating measures of both geographic and social spillovers when analyzing the impact of public policies, particularly in the context of infectious diseases like COVID-19, as failure to account for these spillovers may result in biased estimates of policy effectiveness. (Aronow et al. 2020)
Utilize a conditional independence assumption to account for unobserved pre-treatment Covid-19 cases when evaluating the impact of early-pandemic policies, allowing for the estimation of bounds on the effect of the policy on actual Covid-19 cases using differences in confirmed cases among treated and untreated groups with similar pre-treatment characteristics. (Callaway and Li 2020)
2.8 Transmission of Covid
Exercise caution when interpreting findings related to viral risk during flight, safe flight times, or seating capacity, as the viral aerosol production rates, infectious dose, and general assumptions used to estimate a flight time of 54 hours to produce an infection are hypothetical and not intended to provide actionable information. (Bae et al. 2020)
Exercise caution when interpreting results from models that rely heavily on assumptions about variables such as incubation periods, especially when those assumptions differ significantly from empirical findings, as they can lead to implausible predictions about infectiousness profiles. (Slifka and Gao 2020)
2.10 Covid Data
- Carefully consider the limitations and biases present in open data sources for COVID-19, including issues with data format variability, time-varying nature, lack of accuracy in confirmed cases and mortality rates, absence of individual case data, changing and non-uniform criteria for reporting, and shifting database structures and locations. (Alamo et al. 2020)
2.11 Covid Demographics
- Carefully match participant characteristics when comparing immune responses between males and females, as significant differences exist in age, BMI, and other factors that could potentially confound results. (Takahashi et al. 2020)
2.12 Covid Origins
- Employ a multi-disciplinary and collaborative approach, involving both local and international experts, to comprehensively investigate the origins of SARS-CoV-2 using multiple lines of evidence, such as epidemiological, molecular, and environmental data, while carefully considering alternative hypotheses and conducting rigorous risk assessments. (Nishiura, Linton, and Akhmetzhanov 2020)
2.13 Economic Impact of Covid
- Consider leveraging private sector data to supplement traditional survey-based statistics, as it provides granular and high-frequency insights into economic activity, while being mindful of privacy concerns and ensuring representativeness through careful data processing and benchmarking against public statistics. (Carman and Nataraj 2020)
2.14 Fatality Rate of Covid
- Carefully consider and address sources of uncertainty in serological test characteristics and delays from onset of infection to seroconversion, death, and antibody waning when estimating the infection fatality ratio (IFR) of COVID-19. (Brazeau et al. 2020)
2.15 Mask Policy
- Consider the behavioral, societal, and contextual factors affecting mask-wearing adherence, alongside the material and construction properties of masks, when studying the effectiveness of face masks and coverings in reducing virus transmission. (Teasdale et al. 2014)
2.16 Monitoring Covid
- Consider implementing a flexible time series framework using a variety of models including linear regression, generalized linear models (GLMs), or Bayesian regression, combined with outlier detection inspired by classical Shewhart control charts, to identify recent anomalous data points and signal ongoing changes in COVID-19 incidence patterns. (Jombart et al. 2020)
2.17 Mortality from Covid
- Carefully consider the potential impact of missing or incomplete data when analyzing mortality statistics, particularly in light of the COVID-19 pandemic, and take steps to address this issue through strategies such as excluding certain data points or using weighted mortality counts. (Karlinsky and Kobak 2021)
2.18 Symptoms of Covid
- Utilize a two-stage weighting process to minimize errors of representation in survey data, specifically by incorporating inverse propensity score weighting to mitigate non-response biases followed by post-stratification to account for coverage errors. (“Gridded Population of the World, Version 4 (GPWv4): Population Density, Revision 11” 2017)
2.19 Transmission Of Covid
- Employ rigorous epidemiological investigations to establish the likelihood of airborne transmission of SARS-CoV-2 at distances greater than two meters, taking into consideration factors such as air replacement rates, air flow patterns, and duration of exposure. (Palmer et al. 2021)
2.20 Variants of Covid
- Conduct clinical evaluations of Ag-RDT performance in Omicron-infected individuals, as analytical testing with cultured virus may provide useful insights into sensitivity but cannot replace actual clinical trials. (Bekliz et al. 2021)
3 Political science
3.1 Political Events
Leverage recent advancements in natural language processing (NLP) to efficiently produce custom event data without relying on traditional, labor-intensive dictionary-based methods. (Halterman et al. 2023)
Consider framing political event coding as a text-to-text sequence generation problem, leveraging transformer language models and overcoming data scarcity challenges through synthetic data generation techniques. (Dai, Radford, and Halterman 2022)
Consider adopting the Plover ontology for event data analysis due to its simpler and more flexible structure, which includes distinct components for event type, mode, and context, allowing for more accurate and efficient coding of political events. (Azar 1984)
3.2 Ideal Point Estimation
Consider using Expectation Maximization (EM) algorithms for estimating ideal points in spatial voting models, which can significantly reduce computation time while producing nearly identical results compared to traditional methods such as Markov Chain Monte Carlo (MCMC). (IMAI, LO, and OLMSTED 2016)
Carefully consider identifiability issues in logistic regression models, particularly in the presence of additive and multiplicative aliasing, and employ methods such as hierarchical modeling, linear transformations, and informative regression predictors to address these challenges. (Bafumi et al. 2005)
3.3 Voter Turnout
Utilize difference-in-differences identification strategies to address the confounding effects of individual characteristics when studying the impact of neighborhood characteristics on political participation. (Imai and Khanna 2016)
Consider incorporating social preferences into your models of voter behavior, as doing so allows for the possibility of rational voting in large elections due to the non-diminishing expected utility of social benefits as the size of the electorate increases. (Edlin, Gelman, and Kaplan 2007)
3.4 Historical Persistence
- Carefully define and measure your causal variable of interest, characterize its assignment mechanism, and estimate its effect on a relevant outcome variable using an appropriate research design that accounts for potential confounding factors, in order to draw valid causal conclusions about historical persistence in political science. (Cirone and Pepinsky 2022)
3.5 Incumbency Advantage
- Avoid using incumbent votes as a proxy for incumbency advantage, as it leads to biased estimates due to the confounding effect of partisan predispositions. Instead, researchers should aim to measure incumbency advantage directly and use exogenous measures of constituency service, such as legislative operating budgets, to estimate its impact accurately. (King 1991)
3.6 Nonpartisan Programmatic Policies
- Utilize large-scale randomized experiments whenever feasible to establish causality in the absence of discretionary spending, as demonstrated by the finding that programmatic policies have no discernible effect on voter support for incumbents. (Imai, King, and Rivera 2019)
3.7 Partisan Symmetry
- Consider using the partisan symmetry criterion as a useful tool for measuring the burden on representational rights in the context of partisan gerrymandering claims, although it should not be the only factor considered and its implementation may require careful consideration of whether it should be applied retrospectively or prospectively. (Grofman and King 2007)
3.8 Research Design
- Strive to maximize leverage - explaining as much as possible with as little as possible - while minimizing bias and reporting estimates of uncertainty in your conclusions. (King, Keohane, and Verba 1995)
3.9 Simulation-based Inference
- Utilize statistical simulation techniques to extract and communicate the most relevant and interpretable information from your statistical results, thereby improving the transparency and accessibility of your findings. (King, Tomz, and Wittenberg 2000)
4 Warfare and conflicts
4.1 Civil War
Prioritize out-of-sample predictive performance over in-sample statistical significance when developing models for explaining complex phenomena like civil wars. (Ward, Greenhill, and Bakke 2010)
Employ a structured-focused comparison design in conducting case studies to complement existing theories, refine empirical measures, explore interactions among explanatory variables, identify endogenous and exogenous variables, and identify causal mechanisms underlying the theory. (Sambanis 2004)
Prioritize analyzing the conditions that facilitate insurgency, such as poverty, slow growth, rough terrain, and large populations, instead of assuming that ethnic or religious diversity alone drives civil war onset. (FEARON and LAITIN 2003)
Strive to bridge the gap between academia and policy by making your work more accessible to policymakers, including translating complex statistical analyses into actionable insights, and addressing misconceptions about probabilistic theories. (MACK 2002)
4.2 Conflict Trap
- Combine quantitative and qualitative methods to gain a comprehensive understanding of civil war dynamics, as quantitative analysis excels at identifying correlations among potential determinants while qualitative analysis provides insights into the temporal processes and endogeneity of variables. (“Understanding Civil War (Volume 1: Africa)” 2005)
4.3 Gray Zone Conflict
- Distinguish between internal and external constraints on actors behavior in gray zone conflicts, as this differentiation allows for more accurate predictions and interpretations of observed patterns. (Gannon et al. 2023)
4.4 Greed And Grievance In Civil War
- Carefully evaluate the appropriateness of your chosen proxies, as using inappropriate proxies can lead to incorrect inferences about causal relationships. (Collier, Hoeffler, and Söderbom 2004)
4.5 Predictive Analytics
- Incorporate both static and dynamic information in your models for predicting civil war, as combining structural variables with event data can significantly improve the predictive power of the model. (Chiba and Gleditsch 2017)
4.6 Specific Civil Wars
- Be mindful of the complex interplay between political, economic, and social factors in understanding the causes and dynamics of civil wars, rather than solely focusing on quantitative models like the Collier-Hoeffler (CH) model. (NA?)
4.7 Understanding Civil War Evidence And Analysis
- Carefully match your choice of statistical model to the nature of your data, taking into account issues such as sample size, variable measurement error, and omitted variable bias. (“Understanding Civil War (Volume 1: Africa)” 2005)
5 Macroeconomics
5.1 Fiscal Multiplier
Consider using regime-switching models to capture the heterogeneity of fiscal policy effects across different stages of the business cycle, as they find significant differences in the size of spending multipliers in recessions compared to expansions. (Auerbach and Gorodnichenko 2012)
Carefully consider the impact of monetary policy responses when estimating the government expenditure multiplier in New Keynesian models, particularly when prices or wages are sticky and the zero lower bound is a binding constraint on monetary policy. (Woodford 2011)
5.2 Monetary Policy Surprises
Carefully decompose monetary policy shocks into policy changes and contemporaneous information shocks, as failing to do so may lead to biased estimates of the effects of monetary policy on the economy. (Jarociński and Karadi 2020)
Carefully consider the potential endogeneity issues arising from the simultaneity between policy decisions and financial variables when estimating the impact of monetary policy on credit costs and economic activity. (Gertler and Karadi 2015)
5.3 Economic Policy Uncertainty
- Employ multiple causal inference techniques, such as Granger & Diks-Wolski nonlinear test, Breitung-Candelon test, and convergent cross-mapping, to establish robust causal links between variables like systemic risk, economic policy uncertainty, and firm bankruptcies. (Stolbov and Shchepeleva 2019)
5.4 Inequality And Growth
- Account for potential confounding factors, such as reverse causality and omitted variable bias, when examining the relationship between income inequality and economic growth. (Besley, Persson, and Sturm 2005)
5.5 Secular Stagnation
- Consider using overlapping generations (OLG) models to study secular stagnation, as these models allow for a wider range of factors influencing the natural rate of interest compared to traditional infinite-horizon models, and avoid counterfactual dynamics predicted by existing New Keynesian models. (Eggertsson, Mehrotra, and Robbins 2019)
5.6 Taxable Income Elasticity
- Carefully distinguish between resource costs and transfer costs when estimating the efficiency impact of tax policies, as failing to do so can result in incorrect conclusions regarding the deadweight loss of taxation. (Chetty 2009)
6 Conflict studies
6.1 International Crisis Behavior Project
Aim to create comprehensive and accurate event datasets for studying international relations, incorporating multiple dimensions such as actors, actions, and locations, and utilizing both human coders and natural language processing techniques to ensure high coverage and precision. (Beardsley et al. 2020)
Use a combination of expert and novice coders to annotate data, requiring at least one expert vote per annotation, accepting annotations with a majority of either expert or novice votes, and resolving any remaining ambiguity by selecting the tag with the most votes among the coders. (Lewis et al. 2019)
6.2 ACLED
- Strive for disaggregation in conflict data, particularly in terms of resolution, agency, strategy, and non-violent aspects, in order to better understand the complexities of conflict and improve the validity of inferences drawn from statistical models. (Gleditsch, Metternich, and Ruggeri 2013)
6.3 Coethnic Bias
- Account for the potential impact of coethnic bias - the systematic tendency to favor cooperation with coethnics - when studying attitudes towards high-risk behaviors like informing in wartime settings. (Lyall, Shiraito, and Imai 2015)
6.4 Conflict Analysis
- Collect fine-grained, disaggregated data and incorporate contextual information beyond just prior violence to improve the predictive accuracy of models for forecasting insurgent attacks. (Hirose, Imai, and Lyall 2017)
6.5 State Failure Task Force
- Account for selection bias in case-control designs by applying prior correction, which involves adjusting estimates based on the true population proportion of the outcome variable, to avoid inflated predictions and biased causal inferences. (King and Zeng 2001)
7 International relations theory
7.1 Democratic Peace Theory
Employ nonparametric sensitivity analysis to evaluate the robustness of empirical evidence for the democratic peace hypothesis, as this approach allows for the direct examination of the existence of unobserved confounders without making assumptions about the correct regression model. (Imai and Lo 2021)
Carefully consider the inclusion of multiple independent variables in regression analyses, balancing the need for parsimony with the desire to accurately capture complex relationships. (Oneal and Russett 2005)
Avoid controlling for intervening variables in multivariate models, as doing so can lead to misleading results by artificially eliminating the statistical association between the original key explanatory factor and the outcome variable. (Ray 2003)
Carefully consider the inclusion of fixed effects in your statistical models to account for unobserved heterogeneity among units, particularly when analyzing panel data with rare events. (Oneal and Russett 2001)
Avoid attributing cross-temporal variation in dispute rates solely to differences in regime type, and instead consider how changing constellations of interests might drive observed patterns. (NA?)
7.2 Capitalist Peace Theory
- Consider alternative explanatory variables beyond just regime type when studying the democratic peace phenomenon, such as economic development, free markets, and shared economic interests, as these may account for the observed reduction in conflicts between democracies. (Mansfield 2021)
9 Wikipedia
9.1 NA
- Consider focusing on explicit links added by editors in the text of Wikipedia articles rather than all links, including automatically generated ones, to better understand semantic relationships between concepts and avoid biases introduced by transcluded links. (Consonni, Laniado, and Montresor 2019)
9.2 Automatically Generating Wikipedia Infoboxes From Wikidata
- Carefully consider the relative importance of attributes and values when generating Wikipedia info-boxes from Wikidata, and propose combining frequency and PageRank measures to rank attribute-value pairs. (Sáez and Hogan 2018)
9.3 Bias In Wikipedia
- Employ a supervised classification approach to detect biased statements in Wikipedia, utilizing an automatically created lexicon of bias words along with syntactical and semantic characteristics of biased statements, achieving state-of-the-art performance compared to alternative approaches. (Hube and Fetahu 2018)
9.4 Politics Of Memory
- Employ critical discourse analysis of intertextuality to examine the connections between texts through hyperlinking and other shared patterning, moving beyond micro-level practices to uncover macro and meta-level insights on the organization of Wikipedia and its interactions with other institutions. (Appleby-Arnold et al. 2018)
9.5 Reliability Of Wikipedia
- Pay close attention to the impact of the presence and absence of specific features in Googles Knowledge Panels on users perceptions of news source credibility, as demonstrated by the finding that the presence of an Awards panel and Wikipedia information significantly influences users judgements. (K. Lam et al. 2011)
9.6 Semantics Of Wikipedia Categories
- Consider combining both lexical and statistical information when attempting to derive high-quality axioms from categories in a knowledge graph, as demonstrated by the proposed hybrid approach “Cat2Ax”. (Heist and Paulheim 2019)
10 Genetics
10.1 Animal Genetics
- Consider implementing a complete linear model instead of using pre-adjustment factors for ultrasound scanned carcass traits in large scale sheep genetic evaluations, as it provides significant improvements in regression slopes and is therefore likely to yield more accurate predictions of future progeny performance. (Brito et al. 2017)
10.2 Epigenetics
- Use genetic instruments as causal anchors when investigating the relationship between gene expression and DNA methylation, allowing them to establish directed and specific associations while controlling for linkage disequilibrium and pleiotropy among neighboring genes. (Hop et al. 2020)
10.3 Family Studies
- Consider using twin study designs to control for genetic and shared environmental confounders when attempting to establish causal relationships between exposures and outcomes, particularly in cases where traditional methods of identifying and measuring confounders are impractical or impossible. (McAdams et al. 2020)
10.4 Genomewide Association Study
- Utilize the digital twin test, a novel approach for finding causal regions in genomic data that leverages black-box models and subject matter knowledge, provides provable localization of causal variants, handles multiple comparisons accurately, and enables rigorous causal inferences in the context of family studies. (Bates et al. 2020)
10.5 Quantitative Trait Loci
- Carefully consider the potential for horizontal pleiotropy when conducting Mendelian randomization studies using molecular QTLs as instrumental variables, as this can lead to biased or false causal inferences. (Neumeyer, Hemani, and Zeggini 2020)
11 Redistricting
11.1 Redistricting
Consider using Markov chain Monte Carlo (MCMC) algorithms to simulate redistricting plans, specifically the proposed Swendsen-Wang algorithm extended with simulated tempering and divide-and-conquer approaches, to generate a representative sample of redistricting plans under various constraints. (Fifield et al. 2020)
Carefully account for incumbency advantage when studying electoral responsiveness and partisan bias in US House elections, as it significantly impacts the latter while having a smaller effect on the former. (King and Gelman 1991)
Differentiate between theoretical concepts and your empirical measurements, allowing them to estimate uncertainty around your estimates using statistical techniques. (NA?)
11.2 Gerrymandering
Utilize simulation algorithms to generate alternative redistricting plans that conform to federal and state laws, allowing for rigorous evaluation of potential biases in enacted plans through comparison with these alternatives. (McCartan et al. 2022)
Consider the complex trade-offs involved in redistricting, including the multiple goals of incumbents, the impact of political parties, and the various legal and practical constraints, when evaluating the effects of redistricting on electoral responsiveness and partisan bias. (Gelman and King 1994)
8 Social psychology
8.1 Altruistic Punishment
8.2 Conformism
8.3 Endorsement Experiments
8.4 Inequity Aversion
8.5 Moral Foundations Theory
8.6 Universal Patterns Of Prosocial Behavior Across Cultures