Sampling survey methodology issues of SBS- survey

• Author: Liljana Boçi, Elona Berberi
• Position: Albanian Institute of Statistics
• Pages: 287-296

ISSN 2410-3918 Academic Journal of Business, Administration, Law and Social Sciences Vol 1 No 2

Acces online at www.iipccl.org IIPCCL Publishing, Tirana-Albania July 2015

287

Sampling survey methodology issues of SBS- survey

Liljana BOÇI

Albanian Institute of Statistics

Elona BERBERI

Albanian Institute of Statistics

Abstract

is paper aims at providing an insight on what is required to build an ecient and high quality

business statistics from sample survey procedures, and on the eective and appropriate use of

survey data in analysis. It aims at describing a general overview of what is required to have a

good survey estimate. It shows in practice how to estimate characteristics of the population in

SBS considering: weighting, non-response adjustments, post stratication, estimating a population

totals, the identication and treatment of outliers, and analyses of coecient of variation. It

provides sources of errors and gives recommendations of how to improve them throw sample

survey techniques.

Keywords: Estimation, sample, non-response, outlier, coecient of variation.

Introduction

is paper is focused in a concrete sample survey such as SBS- Statistical Business Survey.

INSTAT (Albanian Institute of Statistics) aims at providing important economic statistics.

Apart from this survey a complement consisting of Balance sheets and the VAT register

is used for the compilation of statistics. e Balance sheets are documented from the

Annual Reports of enterprises. Estimation is the means by which the statistical institution

obtains values for the population of interest so that it can draw conclusions about that

population based on information gathered from only a sample of the population (Moore

& Notz, 2006, 426).

e principle behind estimation in a probability survey like SBS is that each sample unit

represents not only itself, but also several units of the survey population, by taking the

design weight. So, starting from the frame population determining the nal weight are

important parts of the estimation process through a survey.

Once the nal estimation weights have been calculated, they are applied to the sample

data in order to compute estimates. Summary measures of the population such as totals,

are typically estimated for a wide range of characteristics collected from the sample unit –

for example, total turnover or total number of employees.

Taken into consideration the SBS-survey, the purpose of this paper is to describe the SBS-

procedures, to analyze what kinds of problems are phased on that process and the way to

improve them. is paper dedicates attention also to estimations of survey and magnitude

of the sampling error. is provides a measure of the quality of the survey’s estimation in

terms of accuracy.

Sampling Frame and Design

e sampling design used for the SBS consists of systematic random sampling (SRS)