engleski

Evaluation of mean imputation method appropriateness in business surveys: Case of statistical methods use survey in Croatian enterprises

The missing data problem is quite common in business surveys. In order to solve this problem, researchers can use different imputation methods. In this paper the mean imputation method is observed because of its very easy implementation. The four variables from business survey on statistical methods use in Croatian enterprises are used to test if mean imputation method results are satisfactory accurate in estimation of sample means or not. Full data sets of four observed variables are used to simulate five different missing data levels. The analysis has shown that, despite of differences in statistical characteristics of data distributions between observed variables, there was no statistically significant difference at any usually used significance level between real sample means and calculated sample means after conducted imputations. This research confirmed usefulness of the mean imputation method in business surveys with the purpose to make data manipulation easier.

business web survey ; Croatian enterprises ; mean imputation method ; statistical methods use

Ovaj rad je djelomično podržan od strane Hrvatske zaklade za znanost kao dio znanstvenog projekta „Statistical Modelling for Response to Crisis and Economic Growth in Western Balkan Countries“ (STRENGTHS) ; Broj projekta: HRZZ-IP-2013-11-9402, Voditeljica: prof. dr. sc. Ksenija Dumičić. Sažetak ovoga rada objavljen je u Book of Abstracts of the ISCCRO - International Statistical Conference in Croatia - Zagreb, Croatia, 05-06 May 2016 ; Conference Topic: „New Challenges of Official and Applied Statistics in European Union“, Vol. 1, No. 1. Chief Editors: Ksenija Dumičić ; Nataša Erjavec ; Mirjana Pejić Bach. Croatian Statistical Association Zagreb, 2016. ISSN: 1849- 9864. p. 28.