Using the Boxplot analysis in marketing research

• Author: Constantin C.
• Pages: 21-28

Bulletin of the Transilvania University of Braşov

Series V: Economic Sciences • Vol. 10 (59) No. 2 - 2017

Using the Boxplot analysis in marketing research

Cristinel CONSTANTIN1

Abstract: Taking into account the needs of decision makers inside the companies, marketing

research is meant to provide the best information that really can to help the adoption of the

best decisions. In this respect a lot of methods of data analysis can be used but the

researcher has to choose those results that minimize the errors. This paper proposes an

instrumental research regarding the using of Boxplot analysis to identify certain outliers that

can alter the information. In order to attain the objective of this research, we exemplified the

Boxplot analysis on data related to the GDP recorded in 2014 by Romanian counties.

Key-words: Boxplot analysis, marketing research, outlier identification, GDP analysis,

instrumental research, extreme values analysis

1. Introduction

Data analysis is a one of the most important part of every research as far as the

models used have to provide useful information for decision makers. A deep

analysis using different tools can help researchers to compare the results and to

avoid the errors that could be encountered due to poor analysis or bad methods.

Starting from this problem, in this research we emphasised the utility of using

Boxplot analysis especially when data are not normally distributed. Starting from an

example based on quantitative data regarding the Romanian GDP recorded at

territorial level we explain the benefits of using the Boxplot method for data

analysis.

2. The Boxplot analysis

Boxplot analysis is considered a very popular and easy method used for univariate

data analysis when an unimodal continuous variable is computed. Such variables are

measured with interval or ratio scales, which allow calculating the mean and the

variation statistics (variance, standard deviation etc.). But the value of these

indicators could be negatively influenced by extreme values, which are not

1 Transilvania University of Braşov, cristinel.constantin@unitbv.ro