Toward Statistics on Construction Engineering and Management Research

Surveys and other opinion-based data collection techniques are prevalent across various disciplines. However, a lengthy controversy over employing the traditional descriptive and inferential statistics for this type of data continues. In the field of construction engineering and management (CEM), conducting experimental research is a challenging task due to the dynamic and transient nature of the construction industry. As a result, CEM researchers studying topics that involve human behaviors in the construction process must commonly employ research methods that use categorical level measurements to collect data based on expert judgment and opinion. While the interval between ordinal values is not always equal, researchers often perform parametric descriptive and inferential statistics for their data. This analysis may lead to inappropriate results or indefensible conclusions. This paper reviews measurement scales to obtain expert opinion and associated appropriate statistical analyses. The paper analyzes 151 manuscripts published in the 12 issues of the Journal of Construction Engineering and Management (JCEM) from January to December 2012 to identify the typical statistical applications used in CEM research. The paper describes interconnection between scales of measurement and statistical applications and then provides recommendations for CEM researchers to select appropriate statistics tests corresponding to a given data set. In particular, the paper discusses when and how researchers can employ traditional descriptive and inferential statistics to analyze categorical and ordinal data to enhance validity and reliability of the given study. INTRODUCTION Construction engineering and management (CEM) research commonly works to advance the body of knowledge based on latent constructs and variables for regularly uncollected or unshared data. Construction projects often deal with multilingual work crews, dynamic and transient work environments, industry fragmentation, weather, and other factors. These characteristics lead to challenge in conducting research based on the experimental setting. As a result, CEM researchers often rely on non-experimental research methodologies. In fact, through examining 1,102 manuscripts published over the period from 1993 to 2007 in the Journal of 1139 Construction Research Congress 2014 ©ASCE 2014