Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
1 Introduction Developing Effective Experiment Designs and Data Analysis Plans Personnel in state departments of transportation (DOTs), county road commissions, and other transportation agencies often engage in evaluations of different products or want to know whether a treatment has resulted in an improvement in their system, such as whether some crack sealer is better than the one that was used in the past or whether changing signs on a number of horizontal curves really led to a crash reduction. Some agencies also engage in more comprehen- sive and formal studies through ongoing research programs. The questions asked in informal and formal investigations are significantly different, but the common element is that some sort of experiment design should be done that includes specifying the appropriate statistical analysis. In some cases, the required design and analysis are straightforward; in others, complex. The pur- pose of these guidelines is to help the practitioner ask the right questions and design an analysis that is appropriate to address the research problem. Throughout NCHRP Report 727, the terms research and experiment design are used fairly loosely. A large, state-funded project undertaken by a local university to evaluate the effectiveness of some statewide highway safety improvement program clearly qualifies as research. In contrast, com- paring the effectiveness of two products that are routinely used may not seem like research that requires a rigorous experiment design. For the purposes of this guide, however, such a comparison is classified as research, and determining the approach to take to make that comparison is, more or less, designing the experiment. What This Guide Is and Is Not This guide is not intended to turn practitioners into statisticians. For the expert, let alone the uninitiated, designing experiments and understanding and undertaking complex statistical analysis can be truly daunting. It is unnecessary (and would be ill-advised) to try to turn all traffic engineers, pavement designers, and other transportation researchers into qualified statisticians. However, transportation professionals can learn some of the terminology, understand some of the pitfalls in conducting research, learn to ask effective questions, interact more productively with researchers and statisticians, and obtain understandable and valid test results. Practitioners need to be able to understand what needs to be done and what the results really mean with reference to practice. This guide will help transportation practitioners become better engaged in undertaking any kind of experiment design and analysis, improve interactions with contracted consultants and researchers, know when to engage statistical experts, and discern more useful information from transportation studies. C h a p t e r 1
2 effective experiment Design and Data analysis in transportation research Organization of the Guide This report has three chapters. Chapter 2 focuses on how to get organized, ask questions, develop an experiment design, and understand how an appropriate statistical technique/ approach is selected. Chapter 3 presents a variety of common examples of experiments in differ- ent areas of interest (e.g., traffic engineering, maintenance, and planning). The examples empha- size working through various questions related to experiment design. The questions lead to the selection of an analysis technique that is appropriate for the research question formulated. The examples are generally short, although some are more in-depth to better illustrate the issues that must be addressed. It should be noted that NCHRP Project 20-45, âScientific Approaches to Transportation Research,â has already resulted in the publication of NCHRP CD-22, a CD-ROm containing a two-volume primer on statistical methods. Throughout this report, when more details on spe- cific analysis techniques are required, excerpts from or references to NCHRP 20-45 are included. For readersâ convenience, these references also have been summarized in the Resources section at the end of Chapter 3. more information about NCHRP Project 20-45 and NCHRP CD-22 can be found at: http://www.trb.org.
