Measurement error in nonlinear models - a review

Schennach, Susanne M.

doi:10.1920/wp.cem.2012.4112

Cited by 31 publications

(20 citation statements)

References 85 publications

(44 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…My instrumental variable approach can be an an alternative strategy to deal with mismeasured endogenous treatment. It is worthwhile because, as mentioned in Schennach (2013), the availability of an auxiliary dataset is limited in empirical research. Furthermore, it is not always the case that the results from auxiliary datasets is transported into the primary dataset (Carroll, Ruppert, Stefanski, and Crainiceanu (2012, p. 10)), Some papers investigate mismeasured endogenous continuous variables, instead of binary variables.…”

Section: Literature Reviewmentioning

confidence: 99%

Heterogeneous treatment effects with mismeasured endogenous treatment

Ura¹

2018

Quantitative Economics

View full text Add to dashboard Cite

This paper studies the identifying power of an instrumental variable in the nonparametric heterogeneous treatment effect framework when a binary treatment is mismeasured and endogenous. Using a binary instrumental variable, I characterize the sharp identified set for the local average treatment effect under the exclusion restriction of an instrument and the deterministic monotonicity of the true treatment in the instrument. Even allowing for general measurement error (e.g., the measurement error is endogenous), it is still possible to obtain finite bounds on the local average treatment effect. Notably, the Wald estimand is an upper bound on the local average treatment effect, but it is not the sharp bound in general. I also provide a confidence interval for the local average treatment effect with uniformly asymptotically valid size control. Furthermore, I demonstrate that the identification strategy of this paper offers a new use of repeated measurements for tightening the identified set.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Heterogeneous treatment effects with mismeasured endogenous treatment

Ura¹

2018

Quantitative Economics

View full text Add to dashboard Cite

show abstract

“…As Th1 responses are associated with lesion development in bTB, the presence of liver fluke infection could promote a shift towards a Th2‐type immune response and potentially restrain the progression of lesion development. It should be noted that surveillance data have less than perfect sensitivity or specificity in detecting liver fluke (recent European studies suggest sensitivity of 0.63–0.68 and specificity 0.88–1.00; Mazeri, Sargison, Kelly, Bronsvoort, & Handel, ; Rapsch et al., ), which could impact on our inferences here (potentially conservatively biasing our inferences towards the null; Schennach, ). Indeed, similarly, the ability to detect bTB visible lesions at slaughter is less than perfect (though specificity is close to 1; Lahuerta‐Marin et al., ), which is a general limitation to these study types (e.g., Downs et al., ; O'Hagan et al., ; Olea‐Popelka et al., ; Wright et al., ).…”

Section: Discussionmentioning

confidence: 89%

Modelling the variation in skin-test tuberculin reactions, post-mortem lesion counts and case pathology in tuberculosis-exposed cattle: Effects of animal characteristics, histories and co-infection

Byrne

Graham

Brown

et al. 2018

Transbound Emerg Dis

View full text Add to dashboard Cite

Correctly identifying bovine tuberculosis (bTB) in cattle remains a significant problem in endemic countries. We hypothesized that animal characteristics (sex, age, breed), histories (herd effects, testing, movement) and potential exposure to other pathogens (co-infection; BVDV, liver fluke and Mycobacterium avium reactors) could significantly impact the immune responsiveness detected at skin testing and the variation in post-mortem pathology (confirmation) in bTB-exposed cattle. Three model suites were developed using a retrospective observational data set of 5,698 cattle culled during herd breakdowns in Northern Ireland. A linear regression model suggested that antemortem tuberculin reaction size (difference in purified protein derivative avium [PPDa] and bovine [PPDb] reactions) was significantly positively associated with post-mortem maximum lesion size and the number of lesions found. This indicated that reaction size could be considered a predictor of both the extent (number of lesions/tissues) and the pathological progression of infection (maximum lesion size). Tuberculin reaction size was related to age class, and younger animals (<2.85 years) displayed larger reaction sizes than older animals. Tuberculin reaction size was also associated with breed and animal movement and increased with the time between the penultimate and disclosing tests. A negative binomial random-effects model indicated a significant increase in lesion counts for animals with M. avium reactions (PPDb-PPDa < 0) relative to non-reactors (PPDb-PPDa = 0). Lesion counts were significantly increased in animals with previous positive severe interpretation skin-test results. Animals with increased movement histories, young animals and non-dairy breed animals also had significantly increased lesion counts. Animals from herds that had BVDV-positive cattle had significantly lower lesion counts than animals from herds without evidence of BVDV infection. Restricting the data set to only animals with a bTB visible lesion at slaughter (n = 2471), an ordinal regression model indicated that liver fluke-infected animals disclosed smaller lesions, relative to liver fluke-negative animals, and larger lesions were disclosed in animals with increased movement histories.

show abstract

“…With finite samples we may also worry about measurement error bias, although it is not clear how it affects our estimates. In nonlinear models the usual attenuation intuition fails, as discussed in Chesher (1991) and the literature reviewed in Schennach (2013).…”

Section: Estimating the Mtementioning

confidence: 99%

Average and Marginal Returns to Upper Secondary Schooling in Indonesia

Carneiro¹,

Lokshin²,

Ridao-Cano³

et al. 2011

Average and Marginal Returns to Upper Secondary Schooling in Indonesia

View full text Add to dashboard Cite

This paper estimates average and marginal returns to schooling in Indonesia using a semiparametric selection model. Identification of the model is given by geographic variation in access to upper secondary schools. We find that the return to upper secondary schooling varies widely across individuals: it can be as high as 50% per year of schooling for those very likely to enroll in upper secondary schooling, or as low as À10% for those very unlikely to do so. Average returns for the student at the margin are substantial, but they are also well below those for the average student attending upper secondary schooling. 1 We would have liked to use instead the distance between the community of residence in childhood and the nearest school in childhood. It is, in theory, possible to construct such a variable because there should be information about the opening date for all schools in the sampled communities. Unfortunately, many of these dates are missing, and using this information would result in a drop in our sample size of more than 50%, and in hopelessly imprecise estimates. Our assumption is that current residence and current school availability are good approximations to the variables we need (as in Card, 1995). We show below that this measure of distance to school is a good predictor of upper secondary school attendance. We discuss in detail how likely it is that our assumption holds. P. CARNEIRO, M. LOKSHIN AND N. UMAPATHI 18 Note: This table presents estimates of various returns to upper secondary school attendance for the semiparametric and normal selection models: average treatment on the treated (ATT), average treatment effect (ATE), treatment on the untreated (ATU), marginal policy relevant treatment effect (MPRTE) and policy-relevant treatment effect (PRTE) corresponding to a 10% reduction in distance to upper secondary school. Returns to upper school are annualized to show returns for each additional year. Bootstrapped highest posterior density 95% intervals are reported in parentheses. AVERAGE AND MARGINAL RETURNS TO SCHOOLING IN INDONESIA 29

show abstract

Measurement error in nonlinear models - a review

Cited by 31 publications

References 85 publications

Heterogeneous treatment effects with mismeasured endogenous treatment

Heterogeneous treatment effects with mismeasured endogenous treatment

Modelling the variation in skin-test tuberculin reactions, post-mortem lesion counts and case pathology in tuberculosis-exposed cattle: Effects of animal characteristics, histories and co-infection

Average and Marginal Returns to Upper Secondary Schooling in Indonesia

Contact Info

Product

Resources

About