Logistic or linear? Estimating causal effects of experimental treatments on binary outcomes using regression analysis.

Gomila, Robin

doi:10.1037/xge0000920

Cited by 325 publications

(200 citation statements)

References 39 publications

Supporting

Mentioning

197

Contrasting

Unclassified

Order By: Relevance

“…The copyright holder for this preprint this version posted April 13, 2021. ; https://doi.org/10.1101/2021.04.09.21255159 doi: medRxiv preprint (sample positivity) was coded as a binary variable (1=detected, 0=not-detected). A linear model was used to predict the outcome variable as it provided improved interpretability of coefficients and also model convergence stability 18 .…”

Section: Discussionmentioning

confidence: 99%

SARS-CoV-2 Detection from the Built Environment and Wastewater and Its Use for Hospital Surveillance

Hinz

Liu

Doukhanine

et al. 2021

Preprint

View full text Add to dashboard Cite

Background. SARS-CoV-2 causes significant morbidity and mortality in health care settings. Our understanding of the distribution of this virus in the built healthcare environment and wastewater, and relationship to disease burden, is limited. Methods. We performed a prospective multi-center study of environmental sampling of SARS-CoV-2 from hospital surfaces and wastewater and evaluated their relationships with regional and hospital COVID-19 burden. We developed and validated a qPCR-based approach to surface sampling, and swab samples were collected weekly from different locations and surfaces across two tertiary care hospital campuses for a 10-week period during the pandemic, along with wastewater samples. Results. Over a 10-week period, 963 swab samples were collected and analyzed. We found 61 (6%) of swabs were positive for SARS-CoV-2, with the majority of these (n=51) originating from floor samples. Wards that actively managed patients with COVID-19 had the highest frequency of positive samples (p<0.01). Detection frequency in built environment swabs was significantly associated with active cases in the hospital throughout the study (p<0.025). Wastewater viral signal changes appeared to predate change in case burden. Conclusions. Environment sampling for SARS-CoV-2, in particular from floors, may offer a unique and resolved approach to surveillance of COVID-19.

show abstract

Section: Discussionmentioning

confidence: 99%

SARS-CoV-2 Detection from the Built Environment and Wastewater and Its Use for Hospital Surveillance

Hinz

Liu

Doukhanine

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…To take into account the heteroscedasticity and non-normality the method implies, we estimate robust standard errors [e.g. 61,84]. All data curation and statistical analyses were performed with SAS version 9.4 [85].…”

Section: Plos Onementioning

confidence: 99%

The effect of pharmaceutical co-payment increase on the use of social assistance—A natural experiment study

Rättö

Aaltonen

2021

PLoS ONE

View full text Add to dashboard Cite

Health care out-of-pocket payments can create barriers to access or lead to financial distress. Out-of-pocket expenditure is often driven by outpatient pharmaceuticals. In this nationwide register study, we study the causal relationship between an increase in patients’ pharmaceutical expenses and financial difficulties by exploiting a natural experiment design arising from a 2017 reform, which introduced higher co-payments for type 2 diabetes medicines in Finland. With difference-in-differences estimation, we analyze whether the reform increased the use of social assistance, a last-resort financial aid. We found that after the reform the share of social assistance recipients increased more among type 2 diabetes patients than among a patient group not affected by the co-payment increase, suggesting the reform increased the use of social assistance among those subject to it. The results indicate that increases in patients’ pharmaceutical expenses can lead to serious financial difficulties even in countries with a comprehensive social security system.

show abstract

“…One outcome variable was ordinal (Fear of Protesters) and the other was continuous (Support Police Control). However, recent methodological research demonstrated that in experiments with categorical outcomes, linear regression is appropriate (Gomila, 2020;Huang, 2019). Additionally, linear regression is advantageous for testing moderation (Mood, 2010).…”

Section: Analytic Strategymentioning

confidence: 99%

Public fear of protesters and support for protest policing: An experimental study

Metcalfe¹,

Pickett²

2021

Preprint

View full text Add to dashboard Cite

As protests erupted across the United States in recent years over politically polarized issues (e.g., Black Lives Matter, COVID-19 restrictions), so too did questions about when and how police should respond. The context of these protests and how police reacted to them varied substantially, with limited understanding of relevant public attitudes. Public opinion is double-edged; it is critical for police legitimacy and influences criminal justice policy, but it also often reflects racial animus. We hypothesized that disruptive, dangerous, or unlawful protest tactics would increase support for police control, by elevating public fear, but also that support for repression would be higher when protest goals conflict with preexisting racial beliefs. To test our hypotheses, we embedded an experiment in a nationwide survey fielded in 2020, after George Floyd’s killing sparked the broadest protests in U.S. history. We randomized protest tactics (e.g., weapon carrying, violence) and goals, as well as other contextual characteristics (e.g., protest size). We found that the public generally opposed repressive protest policing. However, certain protest tactics increased support for repression by increasing public fear. Protest goals (e.g., pro-Black Lives Matter, anti-COVID-19 restrictions, or pro-confederate monuments) also impacted support for repression, but the effect depended on respondents’ racial beliefs.

show abstract

Logistic or linear? Estimating causal effects of experimental treatments on binary outcomes using regression analysis.

Cited by 325 publications

References 39 publications

SARS-CoV-2 Detection from the Built Environment and Wastewater and Its Use for Hospital Surveillance

SARS-CoV-2 Detection from the Built Environment and Wastewater and Its Use for Hospital Surveillance

The effect of pharmaceutical co-payment increase on the use of social assistance—A natural experiment study

Public fear of protesters and support for protest policing: An experimental study

Contact Info

Product

Resources

About