Monday, March 15
1:00 pm - 2:00 pm

Speaker

Description

Authors: Yujiao Mai, To be determined; Deo Kumar Srivastava, St. Jude Children's Research Hospital; Kevin R. Krull, St. Jude Children's Research Hospital

The sum score of multiple items is commonly used in clinical practice to measure an intended variable/construct but, often, is not very reliable and could lead to incorrect diagnosis or statistical inference. It is important to assess the composite reliability (CR) of the sum score before using it instead of its alternatives, factor-score and latent-variable methods. Cronbach alpha has been popular in estimating CR regardless of its often under- or over-estimation of the CR. MacDonald omega is more accurate but less easy-to-use. However, both methods are not appropriate for multidimensional scales with overlapping items that are related to more than one construct. This study proposes a new CR coefficient, Omega-ESEM, based on exploratory structural equation modeling (ESEM) to overcome this limitation. Extensive Monte Carlo simulations show that Omega-ESEM generally outperforms MacDonald Omega and Cronbach alpha in terms of accuracy, precision, and coverage rates of confidence interval. We implement Omega-ESEM in an R package and apply it to evaluate the CR of PedsQL Multidimensional Fatigue Scale among young adult cancer survivors and provide a guideline for scoring the scale.

Add to my calendar

Add to Google Add to Outlook (.ics)

Create your personal schedule through the official app, Whova!

Get Started