Ronald H. Heck's research while affiliated with University of Hawaiʻi at Mānoa and other places

Publications (22)

Article
This is the first book to demonstrate how to use the multilevel and longitudinal modeling techniques available in IBM SPSS Version 18. The authors tap the power of SPSS's Mixed Models routine to provide an elegant and accessible approach to these models. Readers who have learned statistics using this software will no longer have to adapt to a new p...

Citations

... Evidence from our review showed varied analytical methods used in EMA studies. Though analytical approaches employed in studies are directed by the hypothesis being tested [18], mixed or multilevel models have been indicated to have considerable advantages for analyzing EMA data [51], including having the ability to handle 'correlated data and unequal variances' [52]. ...
... Preliminary to the main analyses, we tested each investigated scale score for differences in existing groups (i.e. groups due to countries of origin) by means of linear mixed models [40]. However, countries of origin could only explain 4%-7% of the variance in any investigated model. ...
... Thus, peer effects originate from the between-classroom variation. An analysis of peer effects therefore has to employ a multilevel approach to be able to handle both the selection problems in school systems (Manski, 1993;Sund, 2009) and the natural nested multilevel structure of school data (Heck et al., 2010;Hox, 2010). ...
... Breeding success, treated as a binomial proportion-dependent variable, the number of fledglings in relation to clutch size in individual broods as unit records, was initially modelled using generalized linear models, assuming binomial error structure and applying the logit link function, so that breeding success was expressed as natural log of the odds of fledged eggs to unfledged eggs (Crawley 2002;Heck et al. 2012). In exactly the same way, we modelled hatching success (the number of hatchlings in relation to clutch size) and fledging success (the number of fledglings in relation to the number of hatchlings). ...
... MLCFA evaluated the same factor structures (the same number of factors) at within (student) and between (school) levels in both scales. The weighted least squares means and variance (WLSMV) estimator was used for analysis of all structures, which included robust standard errors and adjustment to the χ2 test statistic, due to unbalanced group sizes and categorical indicators (Heck & Thomas, 2015). ...
... A linear mixed model (LMM) analysis was carried out for each of the eight variables using the MIXED procedure. LMM lets analyze data with a hierarchical structure in nesting units and has demonstrated its ability to cope with unbalanced and repeated-measures data [29]. For example, distance covered in matches are nested for players across time (i.e., each player has a record for any match played). ...
... With the exception of the individual therapy sample, all the collected data had a hierarchical structure, so the confirmatory factor analyses (CFA) were performed using multilevel modeling (clients nested within families or groups), which takes dependent data into account, avoiding errors in the estimates (Heck & Thomas, 2015). Therefore, the client data were modeled at Level 1, controlling for the variability due to the family or group at Level 2 for the corresponding samples. ...
... Compared to more traditional ANOVA analyses using aggregate data and repeated-measures, LMER controls for random effects variance without data aggregation, providing greater statistical power (Baayen, 2008; Judd, Westfall, & Kenny, 2012). Using SPSS version 24 MIXED procedure (Heck, Thomas, & Tabata, 2014), the fixed effects were group (control, DID/DDNOS) and target type (famous, self, stranger). Subjects were specified as a random factor to account for individual differences in predicting familiarity (random intercept). ...
... Therefore, the impact of community-level factors on deworming among preschool age children (pre SAC) remains understudied [16]. Moreover, analyzing the hierarchical nature data like the DHS data using single-level analysis leads to incorrect estimation of parameters and standard errors [22]. Therefore doing multi-level analyses using cluster effect can fill this gap. ...