scholarly journals Asymmetric Fixed Effects Models for Panel Data

2018 ◽  
Author(s):  
Paul D Allison
Keyword(s):  
Logistic Regression ◽  
Panel Data ◽  
Opposite Direction ◽  
Difference Method ◽  
Fixed Effects ◽  
Regression Models ◽  
Social Phenomena ◽  
Logistic Regression Models ◽  
Fixed Effects Models ◽  
The Difference

Standard fixed effects methods presume that effects of variables are symmetric: the effect of increasing a variable is the same as the effect of decreasing that variable but in the opposite direction. This is implausible for many social phenomena. York and Light (2017) showed how to estimate asymmetric models by estimating first-difference regressions in which the difference scores for the predictors are decomposed into positive and negative changes. In this paper, I show that there are several aspects of their method that need improvement. I also develop a data generating model that justifies the first-difference method but can be applied in more general settings. In particular, it can be used to construct asymmetric logistic regression models.

scholarly journals Asymmetric Fixed-effects Models for Panel Data

2019 ◽  
Vol 5 ◽  
pp. 237802311982644 ◽  
Author(s):  
Paul D. Allison
Keyword(s):  
Logistic Regression ◽  
Panel Data ◽  
Opposite Direction ◽  
Difference Method ◽  
Fixed Effects ◽  
Regression Models ◽  
Social Phenomena ◽  
Logistic Regression Models ◽  
Fixed Effects Models ◽  
The Difference

Standard fixed-effects methods presume that effects of variables are symmetric: The effect of increasing a variable is the same as the effect of decreasing that variable but in the opposite direction. This is implausible for many social phenomena. York and Light showed how to estimate asymmetric models by estimating first-difference regressions in which the difference scores for the predictors are decomposed into positive and negative changes. In this article, I show that there are several aspects of their method that need improvement. I also develop a data-generating model that justifies the first-difference method but can be applied in more general settings. In particular, it can be used to construct asymmetric logistic regression models.

Understanding Charter School Leaders’ Perceptions of Competition in Arizona

2021 ◽  
pp. 0013161X2110373
Author(s):  
Benjamin Creed ◽  
Huriya Jabbar ◽  
Michael Scott
Keyword(s):  
Logistic Regression ◽  
School Choice ◽  
Charter School ◽  
Fixed Effects ◽  
Regression Models ◽  
School Leaders ◽  
State Level ◽  
Local Market ◽  
School Practices ◽  
Logistic Regression Models

Purpose: School choice policies are expected to generate competition leading to improvement in school practices. However, little is known about how competition operates in public education—particularly in charter schools. This paper examines charter-school leaders’ competitive perception formation and the actions taken in response to competition. Research Methods: Using Arizona charter-school leaders’ responses to an original survey, Arizona Department of Education data, and the Common Core of Data, we examined the factors predicting the labeling of a school as a competitor. We estimated fixed effects logistic regression models which examine factors predicting the labeling of competitor schools and of top competitors. We used logistic regression models to understand charter-school leaders’ responses to competition. Findings: We find charter-school leaders in Arizona perceived at least some competition with other schools, and their perceptions vary by urbanicity. While distance between schools mattered generally for labeling a school as a competitor, distance did not factor into labeling “top competitor” schools. Student outcomes did not predict competition between schools, but student demographics were associated with labeling a school a competitor. Charter-school leaders responded to competition through changes in outreach and advertising rather than curriculum and instruction. Competitive responses were related to the respondent school’s quality and the level of perceived competition. Implications for Research and Practice: We found charter-school leaders perceive competition and respond by changing school practices. Responses typically focus on marketing activities over productive responses. The novel state-level analysis allows us to test the effects of local market conditions typically absent in the literature.

Performance simulations for categorical mediation: Analyzing khb estimates of mediation in ordinal regression models

2019 ◽  
Vol 19 (4) ◽  
pp. 913-930 ◽  
Author(s):  
E. Keith Smith ◽  
Michael G. Lacy ◽  
Adam Mayer
Keyword(s):  
Logistic Regression ◽  
Nonlinear Regression ◽  
Regression Models ◽  
Binary Logistic Regression ◽  
Experimental Conditions ◽  
Logistic Regression Models ◽  
Nonlinear Regression Models ◽  
Size Number ◽  
Confidence Interval Coverage ◽  
The Difference

Standard mediation techniques for fitting mediation models cannot readily be translated to nonlinear regression models because of scaling issues. Methods to assess mediation in regression models with categorical and limited response variables have expanded in recent years, and these techniques vary in their approach and versatility. The recently developed khb technique purports to solve the scaling problem and produce valid estimates across a range of nonlinear regression models. Prior studies demonstrate that khb performs well in binary logistic regression models, but performance in other models has yet to be investigated. In this article, we evaluate khb‘s performance in fitting ordinal logistic regression models as an exemplar of the wider set of models to which it applies. We examined performance across 38,400 experimental conditions involving sample size, number of response categories, distribution of variables, and amount of mediation. Results indicate that under all experimental conditions, khb estimates the difference (mediation) coefficient and its associated standard error with little bias and that the nominal confidence interval coverage closely matches the actual. Our results suggest that researchers using khb can assume that the routine reasonably approximates population parameters.

scholarly journals On the Use of Two-Way Fixed Effects Regression Models for Causal Inference with Panel Data

2020 ◽  
pp. 1-11 ◽  
Author(s):  
Kosuke Imai ◽  
In Song Kim
Keyword(s):  
Panel Data ◽  
Causal Inference ◽  
Fixed Effects ◽  
Regression Models ◽  
Applied Research ◽  
Causal Effects ◽  
Difference In Differences ◽  
The Difference ◽  
Fixed Effects Regression ◽  
Time Specific

Abstract The two-way linear fixed effects regression (2FE) has become a default method for estimating causal effects from panel data. Many applied researchers use the 2FE estimator to adjust for unobserved unit-specific and time-specific confounders at the same time. Unfortunately, we demonstrate that the ability of the 2FE model to simultaneously adjust for these two types of unobserved confounders critically relies upon the assumption of linear additive effects. Another common justification for the use of the 2FE estimator is based on its equivalence to the difference-in-differences estimator under the simplest setting with two groups and two time periods. We show that this equivalence does not hold under more general settings commonly encountered in applied research. Instead, we prove that the multi-period difference-in-differences estimator is equivalent to the weighted 2FE estimator with some observations having negative weights. These analytical results imply that in contrast to the popular belief, the 2FE estimator does not represent a design-based, nonparametric estimation strategy for causal inference. Instead, its validity fundamentally rests on the modeling assumptions.

scholarly journals Influence of year-on-year performance on final degree classification in a chiropractic master's degree program

2016 ◽  
Vol 30 (1) ◽  
pp. 14-19 ◽  
Author(s):  
Philip Dewhurst ◽  
Jacqueline Rix ◽  
David Newell
Keyword(s):  
Logistic Regression ◽  
Problem Solving ◽  
Linear Regression ◽  
Regression Models ◽  
Linear Regression Models ◽  
Logistic Regression Models ◽  
The Difference ◽  
Degree Classification ◽  
Grade Performance ◽  
Final Degree

Objective: We explored if any predictors of success could be identified from end-of-year grades in a chiropractic master's program and whether these grades could predict final-year grade performance and year-on-year performance. Methods: End-of-year average grades and module grades for a single cohort of students covering all academic results for years 1–4 of the 2013 graduating class were used for this analysis. Analysis consisted of within-year correlations of module grades with end-of-year average grades, linear regression models for continuous data, and logistic regression models for predicting final degree classifications. Results: In year 1, 140 students were enrolled; 85.7% of students completed the program 4 years later. End-of-year average grades for years 1–3 were correlated (Pearson r values ranging from .75 to .87), but the end-of-year grades for years 1–3 were poorly correlated with clinic internship performance. In linear regression, several modules were predictive of end-of-year average grades for each year. For year 1, logistic regression showed that the modules Physiology and Pharmacology and Investigative Imaging were predictive of year 1 performance (odds ratio [OR] = 1.15 and 0.9, respectively). In year 3, the modules Anatomy and Histopathology 3 and Problem Solving were predictors of the difference between a pass/merit or distinction final degree classification (OR = 1.06 and 1.12, respectively). Conclusion: Early academic performance is weakly correlated with final-year clinic internship performance. The modules of Anatomy and Histopathology year 3 and Problem Solving year 3 emerged more consistently than other modules as being associated with final-year classifications.

Antipsychotic Medication Dispensing and Associated Odds Ratios of Death In Elderly Veterans and War Widows, 2001

2006 ◽  
Vol 40 (11-12) ◽  
pp. 981-986 ◽  
Author(s):  
Jean Hollis ◽  
Stephen Touyz ◽  
David Grayson ◽  
Loelle Forrester
Keyword(s):  
Logistic Regression ◽  
Drug Toxicity ◽  
Regression Models ◽  
Reference Group ◽  
Antidepressant Medication ◽  
Mortality Rates ◽  
Elderly Subjects ◽  
Odds Ratios ◽  
Logistic Regression Models ◽  
The Difference

Objectives: To explore the odds ratios (ORs) of death associated with antipsychotic (AP) medications dispensed to elderly subjects. Method: Subjects were veterans and war widows 65 years and older dispensed an AP drug in 2001 in NSW or ACT. For all subjects, dispensing records for AP medication, benzodiazepines, lithium, carbamazepine, sodium valproate and antidepressant medication were extracted and combined with age, gender and date of death. A study date was allocated, either the date of death or a random date from 1.5.01 to 31.12.01. Subjects dispensed an AP in 2001, but not dispensed an AP or other psychotropic medication in the 120 days prior to their study date, formed a reference group. Psychotropic dispensing in the 120 days prior to the study date was analysed using nested logistic regression models to produce ORs of death associated with various AP drugs. The ORs for risperidone, olanzapine and pericyazine were compared. Haloperidol ORs were established for those dispensed the drug 0–30 days prior to study date or 31–120 days prior to the study date. Results: The ORs associated with haloperidol, olanzapine, risperidone, pericyazine, thioridazine and chlorpromazine were significant when compared with the reference group. Odds ratios for all three haloperidol periods were significant when compared with olanzapine, risperidone and pericyazine 120 day ORs. Although there was a trend favouring olanzapine when compared with risperidone, the difference in the ORs failed to reach significance (p = 0.066). Conclusions: Haloperidol is associated with significantly higher mortality rates than other AP medication but it is not clear whether this represents drug toxicity or the medical conditions for which it was dispensed. There was no evidence that the conventional AP pericyazine was associated with a higher mortality rate than olanzapine or risperidone.

Care Practices and Bronchopulmonary Dysplasia Debate

PEDIATRICS ◽  
1987 ◽  
Vol 80 (6) ◽  
pp. 969-970
Author(s):  
JACOB KELLER ◽  
MARY ELLEN AVERY
Keyword(s):  
Logistic Regression ◽  
Birth Weight ◽  
Null Hypothesis ◽  
Regression Models ◽  
Multiple Logistic Regression ◽  
Dummy Variables ◽  
Logistic Regression Models ◽  
Independent Variables ◽  
Care Practices ◽  
The Difference

In Reply.— We appreciate the opportunity to clarify the issues raised by Barry et al. We tested the significance of the differences among centers in the incidence of chronic lung disease, adjusted for sex, race, and birth weight, by computing two multiple logistic regression models. The first model included only sex, race, and birth weight as independent variables. The second model additionally included seven dummy variables representing the eight centers. Twice the difference in the log likelihoods in the two models has a χ2 distribution with 7 df under the null hypothesis of no center differences.

Association of Body Composition and Intradialytic Hypotension in Hemodialysis Patients

2020 ◽  
Vol 49 (3) ◽  
pp. 334-340
Author(s):  
Maolu Tian ◽  
Yan Zha ◽  
Shuwen Qie ◽  
Xin Lin ◽  
Jing Yuan
Keyword(s):  
Body Composition ◽  
Logistic Regression ◽  
Regression Models ◽  
Density Lipoprotein ◽  
Intradialytic Hypotension ◽  
Logistic Regression Models ◽  
Multifrequency Bioimpedance ◽  
Dialysis Vintage ◽  
The Difference ◽  
Adjusted Logistic Regression

Background/Aim: The relationship between body mass index (BMI) and intradialytic hypotension (IDH) has been inconsistently reported, but no further research has investigated the correlation between body composition and IDH so far. This study aimed to determine whether the lean tissue index (LTI), fat tissue index (FTI), or both derived from body composition monitoring (BCM) is associated with IDH defined as a nadir intradialytic systolic blood pressure of <90 mm Hg and ≥3 episodes hypotension per 10 hemodialysis (HD) treatments in patients undergoing prevalent HD. Methods: The observational cohort study comprised 1,463 patients receiving thrice-weekly HD from 13 dialysis centers. LTI and FTI were assessed using a BCM machine, a multifrequency bioimpedance spectroscopy device. Unadjusted and multivariable adjusted logistic regression models were fit to estimate the association of body composition with the odds of developing IDH. Results: One hundred and seven patients (7.3%) were diagnosed as IDH. The difference in dialysis vintage, BMI, FTI, LTI, high-density lipoprotein cholesterol, and C-reactive protein between IDH and non-IDH groups was statistically significant (all p < 0.05). The prevalence of diabetes among IDH patients was slightly higher than among non-IDH patients. In logistic regression models, low LTI and high FTI, but not high BMI were associated with greater odds of IDH (“high” as above median and “low” as below median). When patients were further stratified into 4 distinct body composition groups based on both the LTI and FTI, only the low LTI/high FTI group was connected with a significantly higher odds of IDH (OR 2.686, 95% CI 1.072–6.734; reference: low LTI/low FTI group). Conclusions: The LTI and FTI can provide better correlation of IDH occurrence than the BMI alone in prevalent HD patients. The low LTI/high FTI appears to be most associated with IDH. An optimal body composition for preventing the occurrence of IDH needs to be determined.

VAAs Users’ Voting Behavior and the Role of Partisanship. Evidence from the 2014 Belgian Federal Elections

2018 ◽  
Vol 9 (2) ◽  
pp. 87-104 ◽  
Author(s):  
Marta Gallina
Keyword(s):  
Logistic Regression ◽  
Panel Data ◽  
Voting Behavior ◽  
Regression Models ◽  
Vote Choice ◽  
Logistic Regression Models ◽  
The Impact ◽  
Voting Advice Applications

AbstractIn this article, I investigate the effects of Voting Advice Applications (VAAs) on voting behavior of their users. It has been already demonstrated that voters are more likely to follow VAAs recommendation when this latter is consistent with their previous vote intentions. However, the role of partisan attachments in this process has been generally overlooked. The basic idea that I intend to test, indeed, is that partisanship works as strong attitude in voters’ minds, making their preferences less amenable to VAAs advices if compared to those of non-partisan citizens. By implementing logistic regression models on panel data from the 2014 Belgian Federal elections, I show that it is actually unlikely that citizens decide to switch their vote after having played the test, if the advice is not consistent with pre-existing vote intentions. More importantly, I find that the impact of VAAs advice on vote choice is even weaker among citizens that declare to feel attached to a party.

Sign in / Sign up

Export Citation Format

Share Document