Abstract
This methodological note makes two novel contributions to spatial political and conflict research using grid data. First, it develops a methodological theory of how uncertainty specific to grid data affects inference. Second, it introduces a comprehensive robustness test on sensitivity to this uncertainty, implemented in R. The uncertainty stems from (1) establishing the correct size of grid cells, (2) deciding the correct locations where the dividing lines of grid data are drawn, and (3) a greater effect of measurement errors due to finer grid cells. The proposed test diversifies grid cell sizes, by aggregating original grid cells into a multiple of these grid cells. The test also varies the locations of the diving lines, by using different starting points of grid cell aggregation (e.g., starting the aggregation from the corner of the entire map or one grid cell of the original size away from the corner). I apply the test to Theisen et al. (Int. Secur. 36(3):79–106, 2011), who utilize the PRIO-GRID data (Tollefsen et al., J. Peace Res. 49(2):363–374, 2012), to substantiate its use.
Similar content being viewed by others
Notes
Depending on the setting, after the aggregation a few grid cells might remain unaggregated or partially aggregated because the aggregated size of grid cells might not evenly divide the entire map. This point is discussed further in the next section.
In theory, both “x” and “y” can be observed exactly in the same location, but such a case is not included here for legibility.
Note the difference between the correct (range of) grid cell size, which should be justified theoretically, and the locations where the dividing lines of grid data are drawn, which cannot be justified theoretically.
This requirement is in addition to the general requirement for the identification of the average treatment effect: both areas with “x” and those without “x” are (conditionally) exchangeable as treatment/control groups. See, for example, Hernán and Robins (2020).
For simplicity, I focus on the case of grid cells with “x” (i.e., where the presence of X is observed), but the same logic applies to counterfactual states (i.e., whether the grid cells would correctly capture the causal relationship between X and Y if “x” were observed).
Again, for simplicity, I focus on the case of grid cells with observed “x,” but the same logic applies to how to define grid cells for the correct specification of counterfactual states.
More specifically, the test is applied to Model 2 in Theisen et al. (2011). Their replication dataset can be obtained from the Peace Research Institute Oslo website at https://www.prio.org/publications/5109 (accessed on January 7, 2022).
As explained in Sect. 2, the exceptions are grid cells at the periphery when an aggregated grid cell size cannot divide the entire map evenly, and/or when the starting point of grid cell aggregation is shifted.
Different grid cell sizes create different numbers of observations and change the baseline likelihood of the onset of civil armed conflict, although, in all grid cell sizes used here, the onsets of civil armed conflict remain rare events (less than \(0.5\%\) of the observations). Therefore, the comparison of effect sizes across models is more meaningful on the log odds ratio scale, the relative scale of an effect, than on the probability scale, the absolute scale of an effect.
Fuzzy-set qualitative comparative analysis (Ragin 2000) is another promising method. For examples in political and conflict research, see Bretthauer (2015), Haesebrouck (2017). For other examples, see Kusa et al. (2021), Medina-Molina et al. (2022), Romero-Castro et al. (2022). I thank an anonymous reviewer for this point.
References
Amrhein, V., Greenland, S., McShane, B.: Retire statistical significance. Nature 567, 305–307 (2019)
Bretthauer, J.M.: Conditions for peace and conflict: applying a fuzzy-set qualitative comparative analysis to cases of resource scarcity. J. Conflict Resolut. 59(4), 593–616 (2015)
Buhaug, H., Gleditsch, K.S., Holtermann, H., Ostby, G., Tollefsen, A.F.: It’s the local economy, stupid! Geographic wealth dispersion and conflict outbreak location. J. Conflict Resolut. 55(5), 814–840 (2011)
Fearon, J.D., Laitin, D.D.: Ethnicity, insurgency, and civil war. Am. Polit. Sci. Rev. 97(1), 75–90 (2003)
Gelman, A.: Causality and statistical learning. Am. J. Sociol. 117(3), 955–966 (2011)
Gross, J.H.: Testing what matters (if you must test at all): a context-driven approach to substantive and statistical significance. Am. J. Polit. Sci. 59(3), 775–788 (2015)
Haesebrouck, T.: NATO burden sharing in Libya: a fuzzy set qualitative comparative analysis. J. Conflict Resolut. 61(10), 2235–2261 (2017)
Hallberg, J.D.: PRIO conflict site 1989-2008 codebook: a geo-referenced dataset on armed conflicts. https://www.prio.org/download/datasetfile/57/ConflictSite%204-2010_v3%20Codebook.pdf (2011)
Hernán, M.A., Robins, J.M.: Causal Inference: What If. Chapman & Hall/CRC, Boca Raton, FL (2020)
Ito, G., Hinkkainen Elliott, K.: Battle diffusion matters: examining the impact of microdynamics of fighting on conflict termination. J. Conflict Resolut. 64(5), 871–902 (2020)
Keele, L., Stevenson, R.T., Elwert, F.: The causal interpretation of estimated associations in regression models. Polit. Sci. Res. Methods 8(1), 1–13 (2020)
Kikuta, K.: A new geography of civil war: a machine learning approach to measuring the zones of armed conflicts. Polit. Sci. Res. Methods 10(1), 97–115 (2022)
King, G., Zeng, L.: Logistic regression in rare events data. Polit. Anal. 9(2), 137–163 (2001)
Kruschke, J.K.: Rejecting or accepting parameter values in Bayesian estimation. Adv. Methods Pract. Psychol. Sci. 1(2), 270–280 (2018)
Kusa, R., Duda, J., Suder, M.: Explaining SME performance with fsQCA: the role of entrepreneurial orientation, entrepreneur motivation, and opportunity perception. J. Innov. Knowl. 6(4), 234–245 (2021)
Lee, D.W., Rogers, M.: Measuring geographic distribution for political research. Polit. Anal. 27(3), 263–280 (2019)
Lee, D.W., Rogers, M.Z., Soifer, H.D.: The modifiable areal unit problem in political science. Paper presented at the 116th APSA Annual Meeting, September 9th–13th (2020)
Lew, M.J.: Bad statistical practice in pharmacology (and other basic biomedical disciplines): you probably don’t know P. Br. J. Pharmacol. 166(5), 1559–1567 (2012)
Linke, A.M., Witmer, F.D.W., Holland, E.C., O’Loughlin, J.: Mountainous terrain and civil wars: geospatial analysis of conflict dynamics in the post-soviet caucasus. Ann. Am. Assoc. Geogr. 107(2), 520–535 (2017)
Lundberg, I., Johnson, R., Stewart, B.M.: What is your estimand? Defining the target quantity connects statistical evidence to theory. Am. Sociol. Rev. 86(3), 532–565 (2021)
Medina-Molina, C., Pèrez-Macías, N., Gismera-Tierno, L.: The multi-level perspective and micromobility services. J. Innov. Knowl. 7(2), 1–11 (2022)
Neumayer, E., Plüumper, T.: Robustness Tests for Quantitative Research. Cambridge University Press, Cambridge (2017)
O’Loughlin, J., Witmer, F.D.W.: The diffusion of violence in the North Caucasus of Russia, 1999–2010. Environ. Plann. A Econ. Space 44(10), 2379–2396 (2012)
Openshaw, S.: The Modifiable Areal Unit Problem. Geo Books, Norwich (1983)
Pearl, J.: On measurement bias in causal inference. In: Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, 425–432. AUAI Press, Arlington, VA (2010)
Pearl, J., Mackenzie, D.: The Book of Why: The New Science of Cause and Effect. Allen Lane, London (2018)
Pickering, S.: Introducing SpatialGridBuilder: a new system for creating geo-coded datasets. Confl. Manag. Peace Sci. 33(4), 423–447 (2016)
R Core Team: R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org (2021)
Ragin, C.C.: Fuzzy-Set Social Science. The University of Chicago Press, Chicago, IL (2000)
Reuveny, R.: Climate change-induced migration and violent conflict. Polit. Geogr. 26(6), 656–673 (2007)
Romero-Castro, N., López-Cabarcos, M.Á., Piñeiro-Chousa, J.: Uncovering complexity in the economic assessment of derogations from the European industrial emissions directive. J. Innov. Knowl. 7(1), 1–14 (2022)
RStudio Team: RStudio: integrated development for R. RStudio, PBC, Boston, MA. http://www.rstudio.com/ (2020)
Ruggeri, A., Dorussen, H., Gizelis, T.-I.: Winning the peace locally: UN peacekeeping and local conflict. Int. Organ. 71(1), 163–185 (2017)
Schutte, S., Donnay, K.: Matched wake analysis: finding causal relationships in spatiotemporal event data. Polit. Geogr. 41, 1–10 (2014)
Schutte, S., Weidmann, N.B.: Diffusion patterns of violence in civil wars. Polit. Geogr. 30(3), 143–152 (2011)
Shaver, A., Carter, D.B., Shawa, T.W.: Terrain ruggedness and land cover: improved data for most research designs. Confl. Manag. Peace Sci. 36(2), 191–218 (2019)
Soifer, H.D.: Units of analysis in subnational research. In: Giraudy, A., Moncada, E., Snyder, R. (eds.) Inside Countries: Subnational Research in Comparative Politics, pp. 92–112. Cambridge University Press, Cambridge (2019)
Sundberg, R., Melander, E.: Introducing the UCDP georeferenced event dataset. J. Peace Res. 50(4), 523–532 (2013)
Suzuki, A.: Presenting the probabilities of different effect sizes: towards a better understanding and communication of statistical uncertainty. arXiv: 2008.07478v3 [stat.AP] (2022)
Theisen, O.M., Holtermann, H., Buhaug, H.: Climate wars? Assessing the claim that drought breeds conflict. Int. Secur. 36(3), 79–106 (2011)
Tollefsen, A.F., Strand, H., Buhaug, H.: PRIO-GRID: a unified spatial data structure. J. Peace Res. 49(2), 363–374 (2012)
Von Uexkull, N., Croicu, M., Fjelde, H., Buhaug, H.: Civil conflict sensitivity to growing-season drought. Proc. Natl. Acad. Sci. 113(44), 12391–12396 (2016)
Wickham, H.: ggplot2: Elegant Graphics for Data Analysis, 2nd edn. Springer International Publishing, Cham (2016)
Wood, R.M., Sullivan, C.: Doing harm by doing good? The negative externalities of humanitarian aid provision during civil conflict. J. Polit. 77(3), 736–748 (2015)
Acknowledgements
I would like to thank seminar participants at Dublin City University, the editors, and the anonymous reviewers, for their helpful comments, and Johan A. Dornschneider-Elkink for encouraging me to independently develop his initial idea on this topic. The accompanying R package,“rbstgrid,” has been co-developed with Johan A. Dornschneider-Elkink at the School of Politics and International Relations, University College Dublin, and is available at https://akisatosuzuki.github.io/programs.html. The views expressed are my own unless otherwise stated, and do not necessarily represent those of the institutes/organizations to which I am/have been related.
Funding
I would like to acknowledge the receipt of funding from the Irish Research Council (the Grant Number: GOIPD/2018/328) for the development of this work.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing interests
The authors reports there are no competing interests to declare.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Suzuki, A. Uncertainty in grid data: a theory and comprehensive robustness test. Qual Quant 57, 4477–4491 (2023). https://doi.org/10.1007/s11135-022-01555-x
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11135-022-01555-x