### Materials and Methods

### Patient population

### Patient assessment

### Intraocular lens

### Surgical technique

### Measurement models

### Astigmatism analysis

### Statistical analysis

*t*-test was used. In order to compare the five models, we used repeated-measure one-way analysis of variance with a post hoc Bonferroni correction. All analyses were performed using IBM SPSS ver. 20.0 (IBM Corp., Armonk, NY, USA). Double-angle plots with 95% confidence ellipses were obtained using Sigmaplot ver. 12.5 (Systat Software Inc., San Jose, CA, USA) and R ver. 3.3.1 (R Foundation, Vienna, Austria). A

*p*-value <0.05 was considered statistically significant.

### Results

*p*< 0.001); however, the type of IOL toricity was not significantly different between the two groups (

*p*= 0.065). The mean refractive cylinders of all patients, the WTRA group, and ATRA group were −0.82 ± 0.63, −0.84 ± 0.73, and −0.80 ± 0.52 D, respectively, at one month postoperatively. These differences were not statistically significant between the two groups (

*p*= 0.800).The changes in corneal astigmatism between the preoperative and one month postoperative values (actual mean SICA) were 0.53 D @ 102.9° in the WTRA group (superior incision) and 0.20 D @ 4.4° in the ATRA group (temporal incision). Table 2 shows the results of the five models throughout the whole sample. The vector and arithmetic means of the DV in model 1 were significantly larger than were those in the other models (

*p*< 0.001). There were further improvements in DV when the actual mean SICA (model 3) was used. The mean vector of the DV was lowest in model 5 (0.15 D @ 48.5°), although the arithmetic mean of the DV was not significantly different between models 3 through 5. The CI in models 4 and 5 was significantly better than those in the other models. Table 3 shows the results of the five models in eyes with WTRA. The vector and arithmetic means of the DV in model 1 were significantly larger than those of the other models (

*p*≤ 0.001). The mean vector of the DV was similar in models 3 to 5. The arithmetic means of the DV were not significantly different between models 3 through 5. The CI progressively decreased from models 1 through 5; this was a significant trend, indicating that the PCA obtained using the regression model and the actual mean SICA had improved the problem of overcorrection.

*p*< 0.001). The mean vector of the DV was lowest in model 5 (0.19 D @ 81.0°), although the arithmetic mean of the DV was not significantly different across models 2 through 5. The CI showed a significantly increasing trend in models 1 through 5. This result indicates that both PCA obtained using the regression model and the actual mean SICA had improved the problem of undercorrection.