A regression estimation procedure based on known coefficient of variation is proposed for the estimation of the population mean. The bias and mean square error of the proposed estimator are found. A comparative study with the usual regression estimator of the population mean has been made. Keywords: Regression estimator, Coefficient of variation, Bias, Mean square error, Efficiency.

  We consider two consecutive nearby occasions over which a finite survey population changes little in composition. The problem is to estimate the current population total on surveying the population on the previous occasion through a well-designed sampling scheme. Then retaining by probability sampling a part of it as a matched sample for which both the past and the current values are ascertained. A Double-sampling theoretic estimate is first obtained for the current total, a current sample is then gathered independently of what precedes yielding another estimate for the current total. These two estimates are then appropriately combined into a pooled estimate as an improved one. Horvitz and Thompson?s (HT, 1952) and generalized regression methods due to Cassel, Sarndal and Wretman (CSW, 1976) provide basic estimation procedures. As these do not yield a variance explicitly in terms of the sample-size we conclude non-availability of an optimal ?Matching Sampling Fraction? (MSF) formula. Estimated coefficient of variation (CV) is derived to guide ?Rotational policy formulation?. For specific unequal probability sampling scheme due for example to Rao, Hartley and Cochran (RHC, 1962) however MSF may be worked out as we have shown in a separate conference paper. In this paper a solution is worked out under a postulated model. Keywords: Double sampling, Matching sampling fraction, Sampling on successive occasions, Unequal probability sampling.

  The problem of estimation of finite population mean in the presence of the random response has been considered when the sampling design is two-stage with two phases at the second stage. Three different types of estimators, based on subsampling of nonrespondents, collecting data on the subsample through specialized efforts, are developed. Expressions for the variances of the estimators along with unbiased variance estimators are developed. Optimum values of sample sizes are obtained by considering a suitable cost function. The percentage reduction in the expected cost of the proposed estimators is studied empirically. Keywords: Cost function, Nonresponse, Random response, Population mean, Subsampling, Two-stage sampling, Percentage reduction in the expected cost.

  Agricultural performance of a country, generally, depends to a large extent on the quantum and distribution of rainfall. So its accurate forecasting is vital for planning and policy purposes. An attempt is made here for modelling and forecasting of Indian monsoon rainfall time-series data by using the promising nonparametric methodology of ?wavelet analysis in frequency domain?. Maximal overlap discrete wavelet transform (MODwT) which, unlike discrete wavelet transform (DwT), does not require the number of data points to be a power of two is employed. Haar wavelet filter is used for computing the same in order to analyze the behaviour of time-series data in terms of different times and scales. wavelet methodology in frequency domain and Autoregressive integrated moving average (ARIMA) methodologies are applied for describing the data and for computing one-step ahead forecasts for hold-out data. Relevant computer programs are developed in SAS, Ver. 9.3 and R, Ver. Keywords: ARIMA, Forecasting, MODwT, Monsoon rainfall, wavelet, Frequency domain.

  Using the calibration approach, the Hansen and Hurwitz (1946) technique based estimator is developed for the situation where the information on auxiliary variable is assumed known for the entire sampled units. Expressions for the estimator of population total, its variance and variance estimator are developed. The theoretical results are illustrated with the help of simulation studies. Simulation results show that proposed calibration approach based estimator outperforms the Hansen and Hurwitz estimator. Keywords: Calibration approach, Hansen and Hurwitz estimator, Non-response, Population total, Sub-sampling of non- respondents.

  This paper is concerned with the recursive construction of balanced ternary (BT), ternary group divisible (TGD) and nested ternary group divisible (NTGD) designs through a set of balanced incomplete block (BIB) designs. An illustrative example in each case has been added separately. The efficiency of BT design has also been computed. Keywords: Balanced incomplete block design (BIBD), Balanced Ternary design (BTD), Ternary group divisible (TGD) design, Nested ternary group divisible (NTGD) design, Intercropping experiments.

  When measuring an observation is expensive, but ranking a small subset of observations is relatively easy, ranked set sampling (RSS) can be used to increase the precision of the estimators. Estimating the variance in case of RSS has been found to be cumbersome in the context of finite population. Therefore, in this paper, we propose two different variance estimation procedures using Jackknife method in RSS under finite population framework. We compare the efficiency of these proposed variance estimation procedures with each other through a simulation study. Keywords: Variance estimation, Ranked set sampling, Jackknife method, Strata based approach, Cycle based approach.

  The small area models make use of explicit linking models based on random area specific effects that account for between areas variation apart from variations explained by auxiliary variables included in the model. The basic small area model considers the random area effects as independent. In practice, it should be more reasonable to assume that the random effects between the neighbouring areas are correlated. In this context, many models have been developed in recent past (Singh et al. 2005 and Pratesi and Salvati 2008, Salvati et al. 2012). In the present study, a spatial unit level small area model is obtained using Geographically Weighted Regression (GWR) approach. Further, the spatial model is studied under Hierarchical Bayes (HB) framework to improve small area estimates. Small area HB estimates are obtained using Gibbs sampling. The effects of incorporating spatial information in the model through three spatial weighting procedures are compared. Results show that estimates from new spatial model in HB framework are more efficient than the empirical approach. Keywords: Spatial unit level model, Geographically Weighted Regression, Hierarchical Bayes, Gibbs sampling.

  Ranked Set Sampling (RSS) provides efficient estimation of population mean as compared to Simple Random Sampling (SRS). The procedure of RSS generates virtual stratification of the population, as a consequence of this, it provides better representation of the population in a selected sample as compared to SRS. Published literatures related to RSS are either based on infinite population theory or super population framework. In this article, an attempt has been made to examine the RSS procedure in randomization frame-work of survey sampling. It has been observed that, this procedure pertains to the category of equal probability sampling method, i.e., the probability of including every unit of the population in a sample is equal. The estimator for estimating population mean has been proved to be unbiased and an expression of its variance has been derived in terms of variability among the sampling units of individual ranks. Statistical properties of this sampling strategy have been studied using simulation under two different cases, i.e., (i) usual case when N = mn2, where N is the population size, n denotes the RSS sample size from each cycle and m is the number of cycle. (ii) two phase sampling when N>mn2. It was found that the proposed RSS estimator is always better than SRS estimator for both the cases. Gain of 20 to 40 percent is achieved in RSS over SRS for equivalent sample size. Keywords: Ranked set sampling, Randomization approach, Simulation, Finite population.

  Crop yield forecasts are quite useful in formulation of policies regarding stock and distribution of agricultural produce to different areas of any country. One among the various statistical approaches in vogue includes models based on Markov chains for providing objective forecasts of crop yields well in advance before harvest for taking timely decisions. A situation which takes the form of a chain of stages with a limited number of possible states (plant condition classes) within each stage is called a Markov chain, if there is a case of simple dependence that any state of a particular stage depends directly on any of the states of the preceding stage. However, for dealing with the key features of Markov chains like estimation of transition probability matrices, predicted yield distributions etc. to get final forecasts, the computational efforts are tedious. One has to either take recourse to writing programs or use statistical packages. Many standard statistical software packages cater to analyze data and obtain forecasts either by using the regression or time series approaches. No single software has tailor-made and customized module to get forecast using the stochastic approach viz. Markov chain modeling. Hence a userfriendly software has been developed based on Markov chain model. It can be used in any platform having Java Virtual Machine (JVM), Java being a platform independent language, programming has been done in Core Java (as back end) and Java Swing (as front end). For testing the software, two years data on biometrical characters and crop yield collected by Indian Agricultural Statistics Research Institute (IASRI), New Delhi under the pilot study on pre-harvest forecasting of sugarcane in Meerut district were utilized. The salient features of the software are highlighted briefly. The software builds first order finite Markov chain (hereinafter referred to as FOMC) model. The software allows up to twenty stages (excluding the harvest stage for first year) depending upon the crop, in the Markov chain model, up to 16 states within each stage and up to four variables can be considered within each stage. Minimum ten records are required for performing analysis through this software. The software is having full description of Markov chain based forecasting model used under the help menu. It also has online help at each screen, which enables the user to decide what to do next and about the details of data files. The software shows output in terms of expected crop yield (forecast) at various stages. Keywords: First order Markov Chain, Java swing, Predicted yield distribution, Transition probability matrix.

  Productivity growth in agriculture is both necessary and sufficient condition for its development. Total Factor Productivity (TFP) is that part of growth in output, which cannot be explained by growth in factor inputs like land, labour and capital. The purpose of this paper is to describe process model, features and functional details of a web based software for computation of total factor productivity (wBSTFP). Software is based on standard three tiers of web architecture. Development is carried out using .NET technology. Software is completely menu driven and user-friendly. Being online system, web browser is the only requirement at user end. The software provides output in the form of TFP index, output index, input index, growth and growth curve of each index. This software is useful for economists, statisticians and other agricultural researchers working in the area of agricultural productivity. Keywords: Agricultural productivity, TFP, web based software, wBSTFP.

   N.A

   N.A

   N.A