Multiple Imputation with Massive Data: An Application to the Panel Study of Income Dynamics
Authors:
Yajuan Si,
Steve Heeringa,
David Johnson,
Roderick Little,
Wenshuo Liu,
Fabian Pfeffer,
Trivellore Raghunathan
Abstract:
\Multiple imputation (MI) is a popular and well-established method for handling missing data in multivariate data sets, but its practicality for use in massive and complex data sets has been questioned. One such data set is the Panel Study of Income Dynamics (PSID), a longstanding and extensive survey of household income and wealth in the U.S. Missing data for this survey are currently handled usi…
▽ More
\Multiple imputation (MI) is a popular and well-established method for handling missing data in multivariate data sets, but its practicality for use in massive and complex data sets has been questioned. One such data set is the Panel Study of Income Dynamics (PSID), a longstanding and extensive survey of household income and wealth in the U.S. Missing data for this survey are currently handled using traditional hot deck methods because of the simple implementation; however, the univariate hot deck results in large random wealth fluctuations. MI is effective but faced with operational challenges. We use a sequential regression/ chained-equation approach, using the software IVEware, to multiply impute cross-sectional wealth data in the 2013 PSID, and compare analyses of the resulting imputed data with those from the current hot deck approach. Practical difficulties, such as non-normally distributed variables, skip patterns, categorical variables with many levels, and multicollinearity, are described together with our approaches to overcoming them. We evaluate the imputation quality and validity with internal diagnostics and external benchmarking data. MI produces improvements over the existing hot deck approach by helping preserve correlation structures, such as the associations between PSID wealth components and the relationships between the household net worth and socio-demographic factors, and facilitates completed data analyses with general purposes. MI incorporates highly predictive covariates into imputation models and increases efficiency. We recommend the practical implementation of MI and expect greater gains when the fraction of missing information is large.
△ Less
Submitted 17 August, 2021; v1 submitted 6 July, 2020;
originally announced July 2020.
The Two Faces of Capacitance: New Interpretations for Electrical Impedance Measurements of Perovskite Solar Cells and Their Relation to Hysteresis
Authors:
Daniel A. Jacobs,
Heping Shen,
Florian Pfeffer,
Jun Peng,
Thomas P. White,
Fiona J. Beck,
Kylie R. Catchpole
Abstract:
Perovskite solar cells are notorious for exhibiting transient behaviour not seen in conventional inorganic semiconductor devices. Significant inroads have been made into understanding this fact in terms of rapid ion migration, now a well-established property of the prototype photovoltaic perovskite MAPbI$_3$ and strongly implicated in the newer mixed compositions. Here we study the manifestations…
▽ More
Perovskite solar cells are notorious for exhibiting transient behaviour not seen in conventional inorganic semiconductor devices. Significant inroads have been made into understanding this fact in terms of rapid ion migration, now a well-established property of the prototype photovoltaic perovskite MAPbI$_3$ and strongly implicated in the newer mixed compositions. Here we study the manifestations of ion migration in frequency-domain small-signal measurements, focusing on the popular technique of Electrical Impedance Spectroscopy (EIS). We provide new interpretations for a variety of previously puzzling features, including giant photo-induced low-frequency capacitance and negative capacitance in a variety of forms. We show that these apparently strange measurements can be rationalized by the splitting of AC current into two components, one associated with charge-storage, and the other with the quasi-steady-state recombination current of electrons and holes. The latter contribution to the capacitance can take either a positive or a negative sign, and is potentially very large when slow, voltage-sensitive processes such as ion migration are at play. Using numerical drift-diffusion semiconductor models, we show that giant photo-induced capacitance, inductive loop features, and low-frequency negative capacitance all emerge naturally as consequences of ion migration via its coupling to quasi-steady-state electron and hole currents. In doing so, we unify the understanding of EIS measurements with the comparably well-developed theory of rate dependent current-voltage (I-V) measurements in perovskite cells. Comparing the two techniques, we argue that EIS is more suitable for quantifying I-V hysteresis than conventional methods based on I-V sweeps, and demonstrate this application on a variety of cell types.
△ Less
Submitted 4 December, 2018; v1 submitted 2 July, 2018;
originally announced July 2018.
