Comparing Software Programs in Conducting Model Analysis With Sampling Weights

• In Event: Friday Roundtable Session 4:55 pm
  In Roundtable Session: Machine Learning, Statistical Software, and Data-Driven Analytics (Table 32)

Abstract

Large-scale Assessment and Survey (LSAS) data have enormous potential for addressing educational questions. Given the complexity of LSAS, researchers are recommended to incorporate sampling weights in data analysis. The current study aimed to provide up-to-date evidence regarding the capabilities of different software programs in weighted analyses. Six software programs (i.e., HLM, Mplus, SAS, SPSS, STATA, and R) were compared by fitting single- and multi- level models, and the weighted parameter estimates and their standard errors were compared. The results showed that four of the software programs, STATA, Mplus, SAS, and HLM are capable of accommodating weighted multi-level model analysis, whereas SPSS or R does not deal with sampling weights appropriately with respect to standard error or variance estimation.

Authors

• Ting Shen, Missouri University of Science & Technology

• Ya Zhang, Western Michigan University