Compositional data statistics improves smart tourism data analytics: profound managerial decisions through reduced statistical bias and increased power

Date

2024-01-03

Contributor

Advisor

Department

Instructor

Depositor

Speaker

Researcher

Consultant

Interviewer

Narrator

Transcriber

Annotator

Journal Title

Journal ISSN

Volume Title

Publisher

Volume

Number/Issue

Starting Page

1701

Ending Page

Alternative Title

Abstract

Psychographics of tourists, customer satisfaction and consumer experience are the basis of managerial decisions. Customer surveys measuring the contentment, the interests and the psychometric profiles of tourists often make use of bipolar econometric and psychometric scales. The statistical analysis of smart tourism data can be improved when considering bipolar scales survey data as compositional data. To date, only the order of magnitude of agreement (OMA) towards an item assertion is considered. Regarding the complement (i.e., the order of magnitude of disagreement (OMD)) increases the amount of information available improving the quality of statistical results. Ignoring the OMD induces serious bias to subsequent descriptive (e.g., computation of Pearson correlation and Cohen's $d$) and inferential statistical analyses (e.g., correlation tests based on Student's t-distribution and two-sample t-tests) possibly leading to ambiguous results. Together, the OMA and the OMD yield a bivariate compositional data point which can be transformed towards the univariate real valued interval scale using the isometric log-ratio (ilr) transformation. Conducting statistical analyses with ilr transformed data avoids statistical biases in the process psychometric tourist profiling. Via simulation, we provide evidence that the ilr approach increases the statistical power of two-sample t-tests (paired and unpaired) affecting managerial decisions based on pre-post-comparisons and comparisons of independent tourist samples.

Description

Keywords

Smart Tourism and Data-driven Decision Making, bipolar scale, isometric log-ratio, normal distribtion, statistical power, t-test

Citation

Extent

10 pages

Format

Geographic Location

Time Period

Related To

Proceedings of the 57th Hawaii International Conference on System Sciences

Related To (URI)

Table of Contents

Rights

Attribution-NonCommercial-NoDerivatives 4.0 International

Rights Holder

Local Contexts

Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.