China
Africa
Explore chapters and articles related to this topic
Introduction
Published in Mark Stamp, Introduction to Machine Learning with Applications in Information Security, 2023
To sample a finite spatial population, the population units are listed in a data frame. This data frame contains the spatial coordinates of the population units and other information needed for selecting sampling units according to a specific design. Think, for instance, of the labels of more or less homogeneous subpopulations (used as strata in stratified random sampling, see Chapter 4) and the labels of clusters of population units, for instance, all units in a polygon of a map (used in cluster random sampling, see Chapter 6). Besides, if we have information about covariates possibly related to the study variable, which we would like to use in selecting the population units, these covariates are added to the list. The list used for selecting sampling units is referred to as the sampling frame.
Inference from probability and nonprobability samples
Published in Uwe Engel, Anabel Quan-Haase, Sunny Xun Liu, Lars Lyberg, Handbook of Computational Social Science, Volume 2, 2021
Rebecca Andridge, Richard Valliant
There are many types of sampling frames, and often in practice there can be a mismatch between observation units available on the sampling frame and the observation units in the population. An address-based sampling frame or sampling frame consisting of telephone numbers might be incomplete or contain outdated information. This gives rise to the notion of the sampled population, which is the group of observation units that could actually be included in the sample. Mismatch between the target population and sampled population is referred to as a coverage error. This mismatch can take two forms: undercoverage is when some of the target population does not appear on the sampling frame, and overcoverage is when the sampling frame contains units that are not part of the target population. Coverage is just one type of error that can occur in probability sampling; for a complete treatment of the so-called total survey error framework, see Biemer et al. (2017).
Decision-Making and Monitoring Strategies in Natural Resource Management and Conservation
Published in Yeqiao Wang, Terrestrial Ecosystems and Biodiversity, 2020
Brian D. Gerber, Perry J. Williams
Probability-based sampling methods assume the underlining spatial process of interest is fixed, but randomness is introduced in sample selection using a probability sampling design. A probability sampling design consists of both a sampling frame, a set of all elements in the population of interest, and a probability measure defined on the frame; a systematic method for assigning a probability to the selection of any subset (i.e., any sample) from the frame. The statistical properties (e.g., unbiased-ness) of the estimators used to make inference on the population are based entirely on the probability measure that defines the design. For this reason, inference from probability-based sampling methods is often referred to as design-based inference. The objective of design-based inference is to quantify a population parameter of interest. Typically, these include summary statistics such as the mean, total, variance, or proportion. Probability-based sampling methods are one of the most common and widely used sampling methods in natural resources. When probability-based sampling designs are combined with the appropriate estimators, investigators obtain objective, unbiased estimates of the population. Both [26] and [28] provide in-depth discussion of probability-based sampling methods and numerous extensions.
Does collaboration pay in agricultural supply chain? An empirical approach
Published in International Journal of Production Research, 2018
Stella Despoudi, Grammatoula Papaioannou, George Saridakis, Samir Dani
In this research, the fresh peach agricultural product was selected as a representative product of the Greek fresh produce ASC. The data of this research were collected at the beginning of 2013, and the overall production and PHFL constructs are averaged over the past three years – covering periods of upward and downward fluctuations in peach production – to eliminate fluctuations in yearly production and PHFL levels and thus, reflect typical values. The actual number of Greek ASC peach producers is not registered anywhere, as producers in Greece are not classified as for example either peach or orange producers. Elstat (2011) provides information about the numbers of peach trees in different regions in Greece. The majority of peach trees are based in Central Macedonia (699,731 trees), Thessaly (29,376 trees), Western Macedonia (30,402 trees) and Eastern Macedonia (245 trees). Thus, the target population of this study is all the peach producers operating in the aforementioned geographical regions as these areas are representative of the whole population of peach producers in Greece. This means that the sample size of this study can be perceived as representative. The sampling frame is created by identifying lists of companies or customers lists (Lee and Lings 2008). However, due to the non-existence of register with the contact details of all the Greek peach producers, the non-probability sampling technique was selected and the producers were approached through gatekeepers (i.e. members of cooperatives) at 180 different cooperatives located in the above regions.
An Empirical Study on Factors Impacting the Adoption of Digital Technologies in Supply Chain Management and What Blockchain Technology Could Do for the Manufacturing Sector of Bangladesh
Published in Information Systems Management, 2023
Zerin Tasnim, Mahmud Akhter Shareef, Abdullah M. Baabdullah, Abu Bakar A. Hamid, Yogesh K Dwivedi
The questionnaires were sent to respondents and the researchers personally supervise this process to collect high-quality survey data. The respondents were all related to supply chain activities in their designated organizations in different departments like procurement, supply chain planning, logistics, and operations, and working as plant heads. All respondents had 3 or more than 3 years of working experience in supply chain-related activities and have some knowledge about BCT. Before sending the questionnaires, the respondents were contacted by the researcher personally over the phone, and after sending the survey questionnaire, a time-to-time follow-up was maintained by the researchers. After completing the survey questionnaires, the questionnaires were collected by the researchers. All survey companies were listed in the Dhaka Stock Exchange list. To maintain the anonymity of the respondents, a random sampling method was applied (Wong et al., 2020). Simple random sampling is preferable to use when there is an accurate sampling frame of the target population (Saunders et al., 2019). In this study, the sampling frame is available from the Dhaka Stock Exchange list and respondents are easily available as the list is available. Hence, random sampling method is suitable for this study at the organizational level. A total of 275 questionnaires were disseminated and only 189 were used for the data analysis process. This sample size matches the minimum sample size requirement for analysis of partial least squares structural equation modeling (PLS-SEM) (J. F. J. Hair et al., 2015) and meets up the criteria of the five-times rule of sample size selection (J. F. Hair et al., 2007).