Abstract
The topics, themes and subject matter of Chapters 1 thru 5 are of a reference nature. Besides the complexities discussed in these preceding chapters, the field of spatial statistics also embraces several prominent difficult, bothersome, and unresolved problems. One concern has to do with incomplete data, which is the topic of this chapter. Geographical data sets sometimes contain missing observations that need to be estimated. An exact maximum likelihood solution for this problem is discussed, both in terms of parameter and missing value estimation, for multivariate normal spatial data sets satisfying the first-order spatial Markov property with constant mean. Moreover, information at neighboring or contiguous observed sites is used to estimate the missing values, and then the complete spatial distribution is used to estimate model parameters. The solution procedure is iterative, and is akin to the Orchard and Woodbury missing information principle. Results are reported for extensions to a second-order, simultaneous model, and from an empirical example used to explore the behavior of these estimates. Also, tentative simulation experiment findings are reviewed.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Abraham, B., 1981, Missing observations in time series, Communications in Statistics A, Vol. 10: 1643–1653.
Afifi, A., and R. Elashoff, 1966, Missing observations in multivariate statistics, I: review of the literature, Journal of the American Statistical Association, Vol. 61: 595–604.
Akaike, H., and M. Ishiguro, 1980, Trend estimation with missing observations, Annals of the Institute of Statistical Mathematics, Vol. 32: 481–488.
Anderson, A., A. Basilevsky, and D. Hum, 1983, Missing data: a review of the literature, Handbook of Survey Research, Vol. 4: 415–494.
Beale, E., and R. Little, 1975, Missing values in multivariate analysis, Journal of the Royal Statistical Society B. Vol. 37: 129–146.
Bennett, R., D. Griffith, and R. Haining, 1984, The problem of missing data on spatial surfaces, Annals, Association of American Geographers, Vol. 74: 138–156.
Besag, J., 1974, Spatial interaction and the statistical analysis of lattice systems, Journal of the Royal Statistical Society B, Vol. 36: 192–236.
Bhoj, D., 1984, On difference of means of correlated variâtes with incomplete data on both responses, Journal of Statistical Computation and Simulation, Vol. 19: 275–290.
Bhoj, D., 1984, On testing equality of variances of correlated variates with incomplete data, Biometrika, Vol. 71: 639–641.
Boyles, R., 1983, On the convergence of the EM algorithm, Journal of the Royal Statistical Society B, Vol. 45: 47–50.
Bouza, C., 1983, Estimation of a difference in finite populations with missing observations, Biometrical Journal, Vol. 25: 123–128.
Box, M., 1971, A parameter estimation criterion for multiresponse models applicable when some observations are missing, Applied Statistics, Vol. 20: 1–7.
Brooks, R., 1982, On the loss of information through censoring, Biometrika, Vol. 69: 137–144.
Brouwer, U., and P. Vijn, 1980, A program to estimate the correlation coefficient in incomplete datasets, COMPSTAT, Vol. 4, Proceedings in Computational Statistics, edited by M. Barritt and D. Wishart, Vienna: Physica, pp. 194–200.
Campbell, N., 1984, Canonical variate analysis — a general model formulation, Australian Journal of Statistics, Vol. 26: 86–96.
Campbell, G., 1984, Testing equality of proportions with incomplete correlated data, Journal of Statistical Planning and Inference, Vol. 10: 311–321.
Chan, L., and O. Dunn, 1974, A note on the asymptotic aspects of the treatment of missing values in discriminant analysis, Journal of the American Statistical Association, Vol. 69: 672–673.
Chapman, D., 1982, Substitution for missing units, Proceedings of Survey Research Methods Section, American Statistical Association, 76–84.
Cheng, S., and K. Ling, 1983, On the BLUE’s of location and scale parameters based on incomplete samples, Soochow Journal of Mathematics, Vol. 9: 35–45.
Chow, G., and A. Lin, 1976, Best linear unbiased estimation of missing observations in an economic time series, Journal of the American Statistical Association, Vol. 71: 719–721.
Dagenais, M., and J. Dufour, 1984, Durbin-Watson tests with missing observations: applications and comparisons, Proceedings, American Statistical Association Business and Economic Statistics Section, pp. 525–530.
Dahiya, R., and R. Korwar, 1980, Maximum likelihood estimates for a bivariate normal distribution with missing data, Annals of Statistics, Vol. 8: 687–692.
Damsleth, E., 1980, Interpolating missing values in a time series, Scandinavian Journal of Statistics, Vol. 7: 33–39.
de Ligny, C. et al., 1981, An application of factor analysis with missing data, Technometrics, Vol. 23: 91–95.
del Pino, G., 1984, Linear restrictions and two step least squares with applications, Statistics and Probability Letters, Vol. 2: 245–248.
Dempster, A., N. Laird, and D. Rubin, 1977, Maximum likelihood from incomplete data via the EM algorithm, Journal of the Royal Statistical Society B, Vol. 39: 1–38.
Donner, A., 1982, The relative effectiveness of procedures commonly used in multiple regression analysis for dealing with missing values, American Statistician, Vol. 36: 378–381.
Donner, A., and B. Rosner, 1982, Missing value problems in multiple linear regression with two independent variables, Communications in Statistics A, Vol. 11: 127–140.
Drygas, H., 1976, Gauss-Markov estimation for multivariate linear models with missing observations, Annals of Statistics, Vol. 4: 779–787.
Dunsmuir, W., and P. Robinson, 1981, Estimation of time series models in the presence of missing data, Journal of the American Statistical Association, Vol. 76: 560–568.
Englman, L., 1982, An efficient algorithm for computing covariance matrices from data with missing values, Communications in Statistics B, Vol. 11: 113–121.
Eubank, R., and V. LaRiccia, 1982, Location and scale parameter estimation from randomly censored data, Communications in Statistics A, Vol. 11: 2869–2888.
Feingold, M., 1982, Missing data in linear models with correlated errors, Communications in Statistics A, Vol. 11: 2831–2843.
Gleason, T., and R. Staelin, 1975, A proposal for handling missing data, Psychometrika, Vol. 40: 229–252.
Gokhale, D., and B. Sirtonik, 1984, On tests for correlated proportions in the presence of incomplete data, Psychometrika, Vol. 49: 147–152.
Greenlees, J. et al., 1982, Imputation of missing values when the probability of response depends on the variable being imputed, Journal of the American Statistical Association, Vol. 77: 251–261.
Haining, R., D. Griffith, and R. Bennett, 1984, A statistical approach to the problem of missing spatial data using a first-order Markov model, The Professional Geographer, Vol. 36: 338–345.
Hamdan, M., W. Pirie, and A. Khuri, 1976, Unbiased estimation of the common mean based on incomplete bivariate normal samples, Biometrische Zeitschrift, Vol. 18: 245–249.
Hartley, H., and R. Hocking, 1971, An analysis of incomplete data, Biometrics, Vol. 27: 783–823.
Harvey, A., 1981, The Kalman filter and its applications in econometrics and time series analysis, Methods of Operations Research, Vol. 44: 3–18.
Harvey, A., and C. McKenzie, 1984, Missing observations in dynamic econometric models, in Time Series Analysis of Irregularly Observed Data, edited by E. Parzen. New York: Springer-Verlag, pp. 108–133.
Harvey, A., and R. Pierse, 1984, Estimating missing observations in economic time series, Journal of the American Statistical Association, Vol. 79: 125–131.
Hill, M., and W. Dixon, 1981, Missing data: search for patterns, Proceedings of the Statistical Computing Section, American Statistical Association, pp. 57–60.
Hinich, M., and W. Weber, 1984, A method for estimating distributed lags when observations are randomly missing, Journal of the American Statistical Association, Vol. 79: 368–373.
Hinkins, S., 1980, RFACTOR — a program to create Rubin’s factorization when there are incomplete multivariate data, American Statistician, Vol. 34: 182–183.
Hocking, R., and D. Marx, 1979, Estimation with incomplete data: an improved computational method and the analysis of nested data, Communications in Statistics — Theory and Methods A, Vol. 8: 1155–1181.
Hocking, R., and H. Oxspring, 1971, Maximum likelihood estimation with incomplete multinomial data, Journal of the American Statistical Association, Vol. 66: 65–70.
Hocking, R., and W. Smith, 1968, Estimation of parameters in the multivariate normal distribution with missing observations, Journal of the American Statistical Association, Vol. 63: 159–173.
Hocking, R., and W. Smith, 1972, Optimum incomplete normal samples, Technometrics, Vol. 14: 299–307.
Hoffmann, R., and J. Anderson, 1982, The effect of missing values on estimators for very short AR(1) time series, Proceedings of the Statistical Computing Section, American Statistical Association, pp. 224–227.
Honohan, P., and C. McCarthy, 1982, On the use of Durbin-Watson type statistics where there are missing observations, The Statistician, Vol. 31: 149–152.
Hosking, J., 1981, Missing data in multivariate linear models: a comparison of several estimation techniques, Proceedings of the SAS Users Group International Conference, Vol. 6: 46–51.
Huseby, J., N. Schwertman, and D. Allen, 1980, Computation of the mean vector and dispersion matrix for incomplete multivariate data, Communications in Statistics B, Vol. 9: 301–309.
Iwase, K., and N. Seto, 1984, A construction of incomplete sufficient unbiased estimators of the normal correlation coefficient, Journal of the Japan Statistical Society, Vol. 14: 49–61.
John, J., and P. Prescott, 1975, Estimating missing values in experiments, Applied Statistics, Vol. 24: 190–192.
Jones, B., and R. Facer, 1982, CORRMAT/PROB, a program to create and test a correlation coefficient matrix from data with missing values, Computers and Geosciences, Vol. 8: 191–198.
Jones, R., 1980, Maximum likelihood fitting of ARMA models to time series with missing observations, Technometrics, Vol. 22: 389–395.
Kemp, W., D. Burnell, D. Eberson, and A. Thomson, 1983, Estimating missing daily maximum and minimum temperatures, Journal of Climate and Applied Meteorology, Vol. 22: 1587–1593.
Kennedy, S., and W. Tobler, 1983, Geographic interpolation, Geographical Analysis, Vol. 15: 151–156.
Korwar, R., and R. Dahiya, 1982, Estimation of a bivariate distribution function from incomplete observations, Communications in Statistics A, Vol. 11: 887–897.
Koul, H., and V. Susarla, 1980, Testing for new better than used in expectation with incomplete data, Journal of the American Statistical Association, Vol. 75: 952–956.
Koziol, J., 1980, Goodness-of-fit tests for randomly censored data, Biometrika, Vol. 67: 693–696.
Laird, N., and T. Louis, 1982, Approximate posterior distributions for incomplete data problems, Journal of the Royal Statistical Society B, Vol. 44: 190–200.
Limonard, C., 1978, Missing values in time series and the implications on autocorrelation analysis, Analytica Chimica Acta, Vol. 103: 133–140.
Lin, P., and L. Stivers, 1974, On difference of means with incomplete data, Biometrika, Vol. 61: 325–334.
Lin, P., 1971, Estimation procedures for differences of means with missing data, Journal of the American Statistical Association, Vol. 66: 634–636.
Little, R., 1976, Inference about means from incomplete multivariate data, Biometrika, Vol. 63: 593–604.
Little, R., and D. Rubin, 1983, Missing data in large data sets, in Statistical Methods and the Improvement of Data Quality, edited by T. Wright. New York: Academic Press, pp. 215–243.
Liung, G., 1982, The likelihood function for a stationary Gaussian autoregressive-moving average process with missing observations, Biometrika, Vol. 69: 265–268.
Marshall, R., 1980, Autocorrelation estimation of time series with randomly missing observations, Biometrika, Vol. 67: 567–570.
Martin, R., 1984, Exact maximum likelihood for incomplete data from a correlated Gaussian process, Communications in Statistics A, Vol. 13: 1275–1288.
Milhoej, A., 1984, Bias correction in the frequency domain estimation of time series models, Biometrika, Vol. 71: 91–99.
Miller, R., and J. Halpern, 1982, Regression with censored data, Biometrika, Vol. 69: 521–531.
Morgan, B., and D. Titterington, 1977, A comparison of iterative methods for obtaining maximum likelihood estimates in contingency tables with a missing diagonal, Biometrika, Vol. 64: 265–270.
Morrison, D., 1971, Expectations and variances of maximum likelihood estimates of the multivariate normal distribution parameters with missing data, Journal of the American Statistical Association, Vol. 66: 602–604.
Morrison, D., and D. Bhoj, 1973, Power of the likelihood ratio test on the mean vector of the multivariate normal distribution with missing observations, Biometrika, Vol. 60: 365–368.
Murry, G., 1979, The estimation of multivariate normal density functions using incomplete data, Biometrika, Vol. 66: 375–380.
Nelson, F., 1981, A test for misspecification in the censored normal model Econometrica, Vol. 49: 1317–1330.
Ord, K., 1975, Estimation methods for models of spatial interaction, Journal of the American Statistical Association, Vol. 70: 120–126.
Papaioannou, T., and S. Loukas, 1984, Inequalities of rank correlation with missing data, Journal of the Royal Statistical Society B, Vol. 46: 68–71.
Papaioannou, T., and T. Speevak, 1977, Rank correlation inequalities with missing data, Communications in Statistics A, Vol. 6: 67–72.
Preece, D., 1971, Iterative procedures for missing values in experiments, Technometrics, Vol. 13: 743–754.
Press, S., and A. Scott, 1976, Missing variables in Bayesian regression II, Journal of the American Statistical Association, Vol. 71: 366–369.
Radhakrishnan, R., 1982, Inadmissibility of the maximum likelihood estimator for a multivariate normal distribution when some observations are missing, Communications in Statistics A, Vol. 11: 941–955.
Ratkowsky, D., 1974, Maximum likelihood estimation in small incomplete samples from the bivariate normal distribution, Applied Statistics, Vol. 23: 180–189.
Redner, R., and H. Walker, 1084, Mixture densities, maximum likelihood and the EM algorithm, SIAM Review, Vol. 26: 195–202.
Richardson, S., and K. White, 1981, The power of tests for autocorrelation with missing observations, Econometrica, Vol. 47: 785–788.
Robinson, P., 1980, Estimation and forecasting for time series containing censored or missing observations, in Time Series, edited by O. Anderson, Amsterdam: North-Holland, pp. 167–182.
Rubin, D., 1974, Characterizing the estimation of parameters in incomplete-data problems, Journal of the American Statistical Association, Vol. 69: 467–474.
Rubin, D., 1972, A noniterative algorithm for least squares estimation of missing values in any analysis of variance design, Applied Statistics, Vol. 21: 136–141.
Rubin, D., 1976, Inference and missing data, Biometrika, Vol. 63: 581–592.
Rubin, D., and T. Szatrowski, 1982, Finding maximum likelihood estimates of patterned covariance matrices by the EM algorithm, Biometrika, Vol. 69: 657–660.
Ryan, T., B. Joiner, and B. Ryan, 1982, Minitab Reference Manual. State College, Pa.: Minitab, Inc.
Selvin, S., 1980, Maximum likelihood estimation for complete or incomplete discrete data, Computer Programs in Biomedicine, Vol. 11: 83–87.
Shumway, R., and D. Stoffer, 1982, An approach to time series smoothing and forecasting using the EM algorithm, Journal of Time Series Analysis, Vol. 3: 253–264.
Singh, R., 1977, A note on the use of incomplete multi-auxiliary information in sample surveys, Australian Journal of Statistics, Vol. 19: 105–107.
Smith, W., and M. Riggs, 1984, Likelihood ratio testing on partial multinormal data, Statistics and Probability Letters, Vol. 2: 337–343.
Sundberg, R., 1974, Maximum likelihood theory for incomplete data from an exponential family, Scandinavian Journal of Statistics, Vol. 1: 49–58.
Susarla, V., and J. Van Ryzin, 1976, Nonparametnc Bayesian estimation of survival curves from incomplete observations, Journal of the American Statistical Association, Vol. 71: 897–902.
Tabony, R., 1982, The estimation of missing values in highly correlated data, COMPSTAT, Vol. 5, Proceedings in Computational Statistics, edited by H. Caussinus, P. Ettinger and R. Tomassone. Vienna: Physica, pp. 425–430.
Titterington, D., 1977, Analysis of incomplete multivariate binary data by the kernel method, Biometrika, Vol. 64: 455–460.
Titterington, D., 1984, Recursive parameter estimation using incomplete data, Journal of the Royal Statistical Society B, Vol. 46: 257–267.
Titterington, D., and J. Jiang, 1983, Recursive estimation procedures for missing-data problems, Biometrika, Vol. 70: 613–624.
Titterington, D., and G. Mill, 1983, Kernel-based density estimates from incomplete data, Journal of the Royal Statistical Society B, Vol. 45: 258–266.
Tobler, W., 1979, Smooth pycnophlyactic interpolation for geographical regions, Journal of the American Statistical Association, Vol. 74: 519–530.
Tobler, W., and S. Kennedy, 1985, Smooth multidimensional interpolation, Geographical Analysis, Vol. 17: 251–257.
Upton, G., 1985, Distance-weighted geographic interpolation, Environment and Planning A, Vol. 17: 667–671.
U.S. National Academy of Sciences, 1980, Panel on Incomplete Data, Washington, D.C.: NAS.
Vacek, P., and T. Ashikaga, 1980, An examination of the nearest neighbor rule for imputing missing values, Proceedings of the Statistical Computing Section, American Statistical Association, pp. 326–331.
van Guilder, M., and S. Azen, 1981, Conclusions regarding algorithms for handling incomplete data, Proceedings of the Statistical Computing Section, American Statistical Association, pp. 53–56.
Vo-Dai, T., 1980, Time series analysis with missing or aberrant data, COMPSTAT, Vol. 4, Proceedings in Computational Statistics, ed. by M. Barritt and D. Wishart. Vienna: Physica, pp. 594–601.
Wei, L., 1983, Tests for interchangeability with incomplete paired observations, Journal of the American Statistical Association, Vol. 78: 725–729.
Weier, D., and A. Basu, 1980, An investigation of Kendall’s t modified for censored data with applications, Journal of Statistical Planning and Inference, Vol. 4: 381–390.
Wingo, D., 1982, Unimodality of the Pareto distribution likelihood function for multicensored samples and implications for estimation, Communications in Statistics A, Vol. 11: 1129–1138.
Woolson, R., J. Leeper, and W. Clarke, 1978, Analysis of incomplete data from longitudinal and mixed longitudinal studies, Journal of the Royal Statistical Society A, Vol. 141: 242–252.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1988 Kluwer Academic Publishers, Dordrecht
About this chapter
Cite this chapter
Griffith, D.A. (1988). The Missing Data Problem for a Two-Dimensional Surface. In: Advanced Spatial Statistics. Advanced Studies in Theoretical and Applied Econometrics, vol 12. Springer, Dordrecht. https://doi.org/10.1007/978-94-009-2758-2_6
Download citation
DOI: https://doi.org/10.1007/978-94-009-2758-2_6
Publisher Name: Springer, Dordrecht
Print ISBN: 978-94-010-7739-2
Online ISBN: 978-94-009-2758-2
eBook Packages: Springer Book Archive