scholarly journals Large Sample Randomization Inference of Causal Effects in the Presence of Interference

2014 ◽  
Vol 109 (505) ◽  
pp. 288-301 ◽  
Author(s):  
Lan Liu ◽  
Michael G. Hudgens
Keyword(s):  
Causal Effects ◽  
Large Sample ◽  
Randomization Inference

The bidirectional association between Internet addiction and depression: a large-sample longitudinal study among Chinese university students (Preprint)

2020 ◽  
Author(s):  
Wan-Jun Guo ◽  
Xia Yang ◽  
Yu-Jie Tao ◽  
Ya-Jing Meng ◽  
Hui-Yao Wang ◽  
...  
Keyword(s):  
Longitudinal Study ◽  
Internet Addiction ◽  
Causal Effects ◽  
Causal Relationships ◽  
Depression Severity ◽  
Large Sample ◽  
Panel Model ◽  
Bidirectional Association ◽  
Predictive Effect

BACKGROUND Evidence indicates that Internet addiction (IA) is associated with depression, but longitudinal studies have rarely been reported, and no studies have yet investigated potential common vulnerability or a possible specific causal relationship between these disorders. OBJECTIVE To overcome these gaps, the present 12-month longitudinal study based on a large-sample employed a cross-lagged panel model (CLPM) approach to investigate the potential common vulnerability and specific cross-causal relationships between IA and CSD (or depression). METHODS IA and clinically-significant depression (CSD) among 12 043 undergraduates were surveyed at baseline (as freshmen) and in follow-up after 12 months (as sophomores). Application of CLPM revealed two well-fitted design between IA and CSD, and between severities of IA and depression, adjusting for demographics. RESULTS Rates of baseline IA and CSD were 5.47% and 3.85%, respectively; increasing to 9.47% and 5.58%, respectively at follow-up. Among those with baseline IA and CSD, 44.61% and 34.48% remained stable at the time of the follow-up survey, respectively. Rates of new-incidences of IA and CSD over 12 months were 7.43% and 4.47%, respectively. Application of a cross-lagged panel model approach (CLPM, a discrete time structural equation model used primarily to assess causal relationships in real-world settings) revealed two well-fitted design between IA and CSD, and between severities of IA and depression, adjusting for demographics. Models revealed that baseline CSD (or depression severity) had a significant net-predictive effect on follow-up IA (or IA severity), and baseline IA (or IA severity) had a significant net-predictive effect on follow-up CSD (or depression severity). CONCLUSIONS These correlational patterns using a CLPM indicate that both common vulnerability and bidirectional specific cross-causal effects between them may contribute to the association between IA and depression. As the path coefficients of the net-cross-predictive effects were significantly smaller than those of baseline to follow-up cross-section associations, vulnerability may play a more significant role than bidirectional-causal effects. CLINICALTRIAL Ethics Committee of West China Hospital, Sichuan University (NO. 2016-171)

Randomization Inference for Accounting Researchers

2021 ◽  
Author(s):  
Roger M White ◽  
Matthew D. Webb
Keyword(s):  
Small Sample ◽  
Short Paper ◽  
Large Sample ◽  
Randomization Inference

In this short paper we summarize and promote randomization inference for accounting researchers. We discuss applications of randomization inference in both small sample and large sample settings, and we include several examples from our own work. We also provide guidance and sample code to researchers looking to implement randomization inference, as well as caveats to consider.

Estimation of Causal Effects via Principal Stratification When Some Outcomes are Truncated by “Death”

2003 ◽  
Vol 28 (4) ◽  
pp. 353-368 ◽  
Author(s):  
Junni L. Zhang ◽  
Donald B. Rubin
Keyword(s):  
Sample Space ◽  
Causal Effects ◽  
Outcome Variable ◽  
Principal Stratification ◽  
Randomized Experiments ◽  
Large Sample ◽  
Traditional Approaches

The topic of “truncation by death” in randomized experiments arises in many fields, such as medicine, economics and education. Traditional approaches addressing this issue ignore the fact that the outcome after the truncation is neither “censored” nor “missing,” but should be treated as being defined on an extended sample space. Using an educational example to illustrate, we will outline here a formulation for tackling this issue, where we call the outcome “truncated by death” because there is no hidden value of the outcome variable masked by the truncating event. We first formulate the principal stratification ( Frangakis & Rubin, 2002 ) approach, and we then derive large sample bounds for causal effects within the principal strata, with or without various identification assumptions. Extensions are then briefly discussed.

scholarly journals Limitations of Design-based Causal Inference and A/B Testing under Arbitrary and Network Interference

2018 ◽  
Vol 48 (1) ◽  
pp. 136-151 ◽  
Author(s):  
Guillaume W. Basse ◽  
Edoardo M. Airoldi
Keyword(s):  
Causal Inference ◽  
Nonparametric Inference ◽  
Causal Effects ◽  
Randomized Experiments ◽  
Interference Structure ◽  
Large Sample ◽  
Unbiased Estimators ◽  
Nonparametric Approach ◽  
Large Samples ◽  
Interference Mechanism

Randomized experiments on a network often involve interference between connected units, namely, a situation in which an individual’s treatment can affect the response of another individual. Current approaches to deal with interference, in theory and in practice, often make restrictive assumptions on its structure—for instance, assuming that interference is local—even when using otherwise nonparametric inference strategies. This reliance on explicit restrictions on the interference mechanism suggests a shared intuition that inference is impossible without any assumptions on the interference structure. In this paper, we begin by formalizing this intuition in the context of a classical nonparametric approach to inference, referred to as design-based inference of causal effects. Next, we show how, always in the context of design-based inference, even parametric structural assumptions that allow the existence of unbiased estimators cannot guarantee a decreasing variance even in the large sample limit. This lack of concentration in large samples is often observed empirically, in randomized experiments in which interference of some form is expected to be present. This result has direct consequences for the design and analysis of large experiments—for instance, in online social platforms—where the belief is that large sample sizes automatically guarantee small variance. More broadly, our results suggest that although strategies for causal inference in the presence of interference borrow their formalism and main concepts from the traditional causal inference literature, much of the intuition from the no-interference case do not easily transfer to the interference setting.

Within-treatment change in antisocial attitudes and reoffending in a large sample of custodial and community offenders.

2018 ◽  
Vol 42 (4) ◽  
pp. 321-335 ◽  
Author(s):  
Mark V. A. Howard ◽  
Gerard van Doorn
Keyword(s):  
Treatment Change ◽  
Large Sample ◽  
Antisocial Attitudes
Sign in / Sign up

Export Citation Format

Share Document