Analyzing the Impacts of Alternated Number of Iterations in Multiple Imputation Method on Explanatory Factor Analysis
Abstract
-
Keywords
Analyzing the Impacts of Alternated Number of Iterations in Multiple Imputation Method on Explanatory Factor Analysis
Abstract
The study aims to identify the effects of iteration numbers used in multiple iteration method, one
of the methods used to cope with missing values, on the results of factor analysis. With this aim,
artificial datasets of different sample sizes were created. Missing values at random and missing
values at complete random were created in various ratios by deleting data. For the data in random
missing values, a second variable was iterated at ordinal scale level and datasets with different
ratios of missing values were obtained based on the levels of this variable. The data were generated
using “psych” program in R software, while “dplyr” program was used to create codes that would
delete values according to predetermined conditions of missing value mechanism. Different
datasets were generated by applying different iteration numbers. Explanatory factor analysis was
conducted on the datasets completed and the factors and total explained variances are presented.
These values were first evaluated based on the number of factors and total variance explained of
the complete datasets. The results indicate that multiple iteration method yields a better
performance in cases of missing values at random compared to datasets with missing values at
complete random. Also, it was found that increasing the number of iterations in both missing value
datasets decreases the difference in the results obtained from complete datasets
Details
| Other ID | JA78DV57KV |
|---|---|
| Journal Section | Research Article |
| Authors | |
| Publication Date | October 1, 2017 |
| Published in Issue | Year 2017 Volume: 7 Issue: 2 |