The Blinder-Oaxaca decomposition for linear regression models

The counterfactual decomposition technique popularized by Blinder (1973, Journal of Human Resources, 436–455) and Oaxaca (1973, International Economic Review, 693–709) is widely used to study mean outcome differences between groups. For example, the technique is often used to analyze wage gaps by sex or race. This article summarizes the technique and addresses several complications, such as the identification of effects of categorical predictors in the detailed decomposition or the estimation of standard errors. A new command called oaxaca is introduced, and examples illustrating its usage are given.

Standard Errors for the Blinder-Oaxaca Decomposition

The decomposition technique introduced by Blinder (1973) and Oaxaca (1973) is widely used to study outcome differences between groups. For example, the technique is commonly applied to the analysis of the gender wage gap. However, despite the procedure's frequent use, very little attention has been paid to the issue of estimating the sampling variances of the decomposition components. We therefore suggest an approach that introduces consistent variance estimators for several variants of the decomposition. The accuracy of the new estimators under ideal conditions is illustrated with the results of a Monte Carlo simulation. As a second check, the estimators are compared to bootstrap results obtained using real data. In contrast to previously proposed statistics, the new method takes into account the extra variation imposed by stochastic regressors.

OAXACA: Stata module to compute the Blinder-Oaxaca decomposition

-oaxaca- computes the so-called Blinder-Oaxaca decomposition, which is often used to analyze wage gaps by sex or race. Older versions of this routine are available as -oaxaca9- and -oaxaca8-.

A Stata implementation of the Blinder-Oaxaca decomposition

The counterfactual decomposition technique popularized by Blinder (1973) and Oaxaca (1973) is widely used to study mean outcome differences between groups. For example, the technique is often used to analyze wage gaps by sex or race. The present paper summarizes the technique and addresses a number of complications such as the identification of effects of categorical predictors in the detailed decomposition or the estimation of standard errors. A new Stata command called -oaxaca- is introduced and examples illustrating its usage are given.

DEVCON: Stata module to apply the deviation contrast transform to estimation results

devcon transforms the coefficients of 0/1 dummy variables so that they reflect deviations from the "grand mean" rather than deviations from the reference category (the transformed coefficients are equivalent to those obtained by the so called "effects coding") and adds the coefficient for the reference category. The variance-covariance matrix of the estimates is transformed accordingly. The transformed estimated can be used with post estimation procedures. In particular, devcon can be used to solve the identification problem for dummy variable effects in the so-called Blinder-Oaxaca decomposition (see the oaxaca package).

No adverse effect of genetically modified antifungal wheat on decomposition dynamics and the soil fauna community - a field study

The cultivation of genetically modified (GM) plants has raised several environmental concerns. One of these concerns regards non-target soil fauna organisms, which play an important role in the decomposition of organic matter and hence are largely exposed to GM plant residues. Soil fauna may be directly affected by transgene products or indirectly by pleiotropic effects such as a modified plant metabolism. Thus, ecosystem services and functioning might be affected negatively. In a litterbag experiment in the field we analysed the decomposition process and the soil fauna community involved. Therefore, we used four experimental GM wheat varieties, two with a race-specific antifungal resistance against powdery mildew (Pm3b) and two with an unspecific antifungal resistance based on the expression of chitinase and glucanase. We compared them with two non-GM isolines and six conventional cereal varieties. To elucidate the mechanisms that cause differences in plant decomposition, structural plant components (i.e. C:N ratio, lignin, cellulose, hemicellulose) were examined and soil properties, temperature and precipitation were monitored. The most frequent taxa extracted from decaying plant material were mites (Cryptostigmata, Gamasina and Uropodina), springtails (Isotomidae), annelids (Enchytraeidae) and Diptera (Cecidomyiidae larvae). Despite a single significant transgenic/month interaction for Cecidomyiidae larvae, which is probably random, we detected no impact of the GM wheat on the soil fauna community. However, soil fauna differences among conventional cereal varieties were more pronounced than between GM and non-GM wheat. While leaf residue decomposition in GM and non-GM wheat was similar, differences among conventional cereals were evident. Furthermore, sampling date and location were found to greatly influence soil fauna community and decomposition processes. The results give no indication of ecologically relevant adverse effects of antifungal GM wheat on the composition and the activity of the soil fauna community.

A multi-level decomposition of variance in somatic symptom reporting in families with adolescent children

OBJECTIVES: This paper examines four different levels of possible variation in symptom reporting: occasion, day, person and family. DESIGN: In order to rule out effects of retrospection, concurrent symptom reporting was assessed prospectively using a computer-assisted self-report method. METHODS: A decomposition of variance in symptom reporting was conducted using diary data from families with adolescent children. We used palmtop computers to assess concurrent somatic complaints from parents and children six times a day for seven consecutive days. In two separate studies, 314 and 254 participants from 96 and 77 families, respectively, participated. A generalized multilevel linear models approach was used to analyze the data. Symptom reports were modelled using a logistic response function, and random effects were allowed at the family, person and day level, with extra-binomial variation allowed for on the occasion level. RESULTS: Substantial variability was observed at the person, day and occasion level but not at the family level. CONCLUSIONS: To explain symptom reporting in normally healthy individuals, situational as well as person characteristics should be taken into account. Family characteristics, however, would not help to clarify symptom reporting in all family members.

From EEG dependency multichannel matching pursuit to sparse topographic EEG decomposition

In this work, we present a multichannel EEG decomposition model based on an adaptive topographic time-frequency approximation technique. It is an extension of the Matching Pursuit algorithm and called dependency multichannel matching pursuit (DMMP). It takes the physiologically explainable and statistically observable topographic dependencies between the channels into account, namely the spatial smoothness of neighboring electrodes that is implied by the electric leadfield. DMMP decomposes a multichannel signal as a weighted sum of atoms from a given dictionary where the single channels are represented from exactly the same subset of a complete dictionary. The decomposition is illustrated on topographical EEG data during different physiological conditions using a complete Gabor dictionary. Further the extension of the single-channel time-frequency distribution to a multichannel time-frequency distribution is given. This can be used for the visualization of the decomposition structure of multichannel EEG. A clustering procedure applied to the topographies, the vectors of the corresponding contribution of an atom to the signal in each channel produced by DMMP, leads to an extremely sparse topographic decomposition of the EEG.

How does spatial extent of fMRI datasets affect independent component analysis decomposition?

Spatial independent component analysis (sICA) of functional magnetic resonance imaging (fMRI) time series can generate meaningful activation maps and associated descriptive signals, which are useful to evaluate datasets of the entire brain or selected portions of it. Besides computational implications, variations in the input dataset combined with the multivariate nature of ICA may lead to different spatial or temporal readouts of brain activation phenomena. By reducing and increasing a volume of interest (VOI), we applied sICA to different datasets from real activation experiments with multislice acquisition and single or multiple sensory-motor task-induced blood oxygenation level-dependent (BOLD) signal sources with different spatial and temporal structure. Using receiver operating characteristics (ROC) methodology for accuracy evaluation and multiple regression analysis as benchmark, we compared sICA decompositions of reduced and increased VOI fMRI time-series containing auditory, motor and hemifield visual activation occurring separately or simultaneously in time. Both approaches yielded valid results; however, the results of the increased VOI approach were spatially more accurate compared to the results of the decreased VOI approach. This is consistent with the capability of sICA to take advantage of extended samples of statistical observations and suggests that sICA is more powerful with extended rather than reduced VOI datasets to delineate brain activity.