China
Africa
Explore chapters and articles related to this topic
Random Parameters Models
Published in Simon Washington, Matthew Karlaftis, Fred Mannering, Panagiotis Anastasopoulos, Statistical and Econometric Methods for Transportation Data Analysis, 2020
Simon Washington, Matthew Karlaftis, Fred Mannering, Panagiotis Anastasopoulos
Traditional random effects model can be viewed as a restrictive form of a true random parameters model where only the constant term is allowed to vary across groups of observations.
Coupling shared E-scooters and public transit: a spatial and temporal analysis
Published in Transportation Letters, 2023
Mohammadjavad Javadiansr, Amir Davatgari, Ehsan Rahimi, Motahare Mohammadi, Abolfazl (Kouros) Mohammadian, Joshua Auld
In addition, because we have panel data, there are likely correlations within the data that need to be taken into account. We addressed this by employing a random-effects version of the negative binomial count model. The random-effects model assumes that the data are randomly sampled across a larger population and provides the opportunity to generalize the findings better. Moreover, negative binomial regression uses a Poisson-gamma mixture distribution, which is especially suitable for modeling variables that are non-negative integers. The Poisson part of this mixture captures the count nature of the data, and the gamma part accounts for overdispersion.
Insights into Older Adults’ Technology Acceptance through Meta-Analysis
Published in International Journal of Human–Computer Interaction, 2021
Qi Ma, Alan H. S. Chan, Pei-Lee Teh
We used version 2.0 of the Comprehensive Meta-Analysis (CMA) software (Borenstein et al., 2005) to perform the statistical analysis. CMA is relatively rich in features and widely used in meta-analysis studies (e.g., Cai et al., 2017; Toril et al., 2014). Moreover, the software can perform fixed effects and random effects analyses. In addition, it can report key statistics, such as the summary effect and confidence intervals (CI), and measures of heterogeneity and provide adequate information (Borenstein et al., 2009). The fixed effects model assumes that one true effect size underlies all the studies in the analysis, whereas the random effects model assumes that the true effect varies from one study to another (Field, 2003). In this study, the random effects model was used to estimate the distribution of the observed effect sizes, because the possible differential effects of moderators such as sample age, gender proportion, geographic region, and technology domain demonstrated that effect sizes were not identical across studies. To estimate publication bias and detect heterogeneity, funnel plots and fail-safe N statistics were used (Egger et al., 1997). Funnel plots were generated as a visual aid to detect publication bias or systematic heterogeneity in the meta-analysis for the three pairs of correlations (i.e., PU–BI, PEOU–BI, and SI–BI). Fail-safe N statistics (or availability bias) were also calculated, which provided the number of nonsignificant studies that must be included in the sample to reverse the conclusion that a significant relationship existed. A meta-regression, which refers to the regression analysis in a meta-analysis, was used to detect the moderating effects and possible sources of heterogeneity across the primary studies. In a meta-regression, the Q value is a test statistic that indicates whether the regression coefficient (not including the intercept) differs from zero. If the Q value is significant when compared with a critical chi-square statistic (;), then the regression coefficient should differ from zero (Borenstein et al., 2005). Before a meta-analysis, the basic characteristics of the primary studies were extracted, classified, and coded. Size-weighted correlations as effect sizes were calculated based on the available statistics from the primary studies.