Table 6 Results of Monte Carlo simulation for sensitivity analysis of Model 3B with N = 2500, 4000, 5000, and 7000.

From: Evolving trajectories of educational expectations and science performance during middle school and STEM degree attainment of youth in adulthood

 

Sample Size

Assessment Criteria

N = 2500

N = 4000

N = 5000

N = 7000

CFI

1.000

1.000

1.000

1.000

RMSEA

0.004

0.003

0.003

0.002

SRMR

0.004

0.003

0.003

0.003

Range of Parameter Estimate Bias %

−2.2203 to 0.9901

−1.9201 to 1.3079

−1.2931 to 0.1964

−0.7942 to 0.1964

Range of Standard Error Bias %

−1.0978 to 2.5970

−2.9059 to 2.5714

−1.7857 to 0.1949

−2.1276 to 0.9237

Range of 95% Coverage

0.943 to 0.960

0.9460 to 0.9600

0.943 to 0.956

0.944 to 0.957

Range of Power (\(\hat{P}\))

0.829 to 1.000

0.9550 to 1.000

0.985 to 1.000

0.997 to 1.000

  1. The input of parameter estimates in Monte Carlo sensitivity analysis is unstandardized coefficients, which are needed for simulations. As observed in the sensitivity analysis, a sample size of N ≥ 2500 is needed to preserve adequate statistical power (\(\hat{P}\) ≥ 0.80) for model reliability and stability.
Back to article page