Search

Supplementary files for manuscript https://osf.io/preprints/metaarxiv/7jqtv/ include three project components, one for each reproduction study

A survey of researcher perceptions of replication in geography.

Mollalo et al. (2020) investigated county-level variations of COVID-19 incidence across the continental United States using spatial lag and spatial error models to investigate spatial dependence as well as geographically weighted regression (GWR) and multiscale GWR (MGWR) models to locally examine spatial non-stationarity. The original analyses are retrospective and use observational data collected from federal and other public sources. Although not publicly available, we were able to obtain the original data based on the authors's description. However, the analysis code was not made available. At the outset of a larger project on reproducibility and replicability in the human-environment and geographic sciences, we chose this study to reproduce on account of its application of spatial statistical methods common in spatial epidemiology and its compatibility with graduate student training in spatial statistics (e.g., spatial regression and pattern analysis) as well as its data and code availability. As a reproduction study, we aimed to independently generate identical results from the original publication.

The original study is one of the first spatial statistical analyses of of COVID-19 in the United States. A main contribution of the study is to provide more detail into the interplay between COVID-19 and healthcare-related characteristics of the population using readily available datasets. At the outset of a larger project on reproducibility and replicability in the human-environment and geographic sciences, we chose this study to reproduce on account of its application of spatial statistical methods common in spatial epidemiology and compatibility with graduate student training in spatial statistics (e.g., spatial regression and pattern analysis) as well as its data and code availability. As a reproduction study, we aimed to independently generate identical results from the original publication. We found an exact reproduction to be impossible due to missing variables in the published data and ambiguities in the narrative methodology.

Vijayan et al. (2021) examined whether spatial patterns existed in SARS-CoV-2 age-adjusted testing rates, age-adjusted diagnosis rates, and crude positivity rates in Los Angeles County (LAC), and used a spatial regression model to explore associations between COVID-19 crude positivity rates and a series of predictor variables. The original analyses are retrospective and use observational data collected from federal and private sources. Although not publicly available, we were able to obtain the original study data after contacting the authors. However, the analysis code was not made available, nor was information about the computational environment used. At the outset of a larger project on reproducibility and replicability in the human-environment and geographic sciences, we chose this study to reproduce on account of its application of spatial statistical methods common in spatial epidemiology and its compatibility with graduate student training in spatial statistics (e.g., spatial regression and pattern analysis) as well as its data and code availability. As a reproduction study, we aimed to independently generate identical results from the original publication.

A survey of geographic research community designed to gather information about the current understanding and use of reproducible research practices. This project will be updated too include anonymized data, results, and preprints as they are completed.

Researchers in the human-environment and geographical sciences study a world of incredible complexity and enduring change. As such a persistent challenge for the field is to develop credible and reliable explanations of geographic phenomena. We examine a range of problems that pose potential challenges to the credibility of geographic research (e.g., publication bias, specification searching). Our overarching goals are to 1) describe the extent of these problems in the field, 2) explain their causes, and 3) develop interventions to mitigate their impacts on the credibility and reliability of geographic research.