Skip to main content

Table 3 Influence on self-esteem in complete and impute data by multiple logistic regression models

From: The ability of different imputation methods for missing values in mental measurement questionnaires

Missingness
proportion
Original
OR (95% CI)
HD
OR (95% CI)
MI
OR (95% CI)
Complete data
 Gender 0.754 (0.648–0.877)
 Grade 1.067 (1.022–1.115)
 Record 0.674 (0.619–0.733)
 Residence 1.054 (0.955–1.164)a
 Communication 0.656 (0.589–0.731)
5%
 Gender 0.727 (0.625–0.845) 0.739 (0.634–0.861)
 Grade 1.060 (1.015–1.106) 1.067 (1.021–1.115)
 Record 0.680 (0.626–0.740) 0.674 (0.617–0.736)
 Residence 1.030 (0.931–1.140) a 1.045 (0.943–1.157) a
 Communication 0.686 (0.617–0.762) 0.665 (0.594–0.745)
10%
 Gender 0.718 (0.618–0.834) 0.712 (0.611–0.830)
 Grade 1.062 (1.018–1.109) 1.059 (1.014–1.107)
 Record 0.700 (0.645–0.760) 0.684 (0.627–0.746)
 Residence 1.048 (0.950–1.157)a 1.052 (0.940–1.177)a
 Communication 0.670 (0.603–0.745) 0.675 (0.604–0.754)
 15%
 Gender 0.743 (0.639–0.864) 0.746 (0.639–0.871)
 Grade 1.067 (1.022–1.114) 1.059 (1.010–1.110)
 Record 0.691 (0.636–0.751) 0.696 (0.639–0.759)
 Residence 1.082 (0.982–1.191)a 1.038 (0.929–1.159)a
 Communication 0.708 (0.638–0.786) 0.683 (0.599–0.779)
20%
 Gender 0.748 (0.642–0.870) 0.726 (0.619–0.853)
 Grade 1.074 (1.029–1.122) 1.053 (1.006–1.102)
 Record 0.698 (0.642–0.759) 0.710 (0.645–0.781)
 Residence 1.075 (0.975–1.185)a 1.057 (0.914–1.221)a
 Communication 0.710 (0.639–0.789) 0.692 (0.618–0.774)
  1. HD Hot-deck imputation, MI multiple imputation
  2. aOR was not statistically significant