This paper introduces a method for testing metric invariance using network psychometrics within the Exploratory Graph Analysis (EGA) framework. The method uses network loadings, analogous to factor loadings in traditional structural equation modeling (SEM). The study compares this method to traditional SEM approaches and finds that it offers comparable or improved power, especially in scenarios with smaller or unequal sample sizes and lower noninvariance effect sizes. Metric invariance assesses whether a latent variable is measured equivalently across groups. Traditional methods for testing metric invariance include SEM, which involves four tests: configural, metric, scalar, and strict. However, these methods have limitations, such as reliance on data-driven methods and unequal group sample sizes, which reduce the power of traditional SEM methods. The proposed method uses permutation testing to assess the equivalence of network loadings across groups. This method is more conceptually consistent with the EGA framework and does not require intensive model specifications or comparisons. It is straightforward to implement in R and provides a way to test metric invariance in the EGA framework. The study compares the proposed method to traditional SEM methods and finds that it performs well in detecting noninvariant items, especially in conditions with small or unequal sample sizes and lower noninvariance effect sizes. The method uses a permutation test to assess the equivalence of network loadings across groups and provides a way to test metric invariance in the EGA framework. The study also discusses the multiple comparison problem in partial invariance testing and the use of the Benjamini-Hochberg procedure to control the false discovery rate. The results show that the proposed method has higher accuracy in detecting noninvariant items compared to traditional SEM methods, particularly in conditions with small or unequal sample sizes and lower noninvariance effect sizes. The method is more robust to these conditions and provides a more accurate assessment of metric invariance.This paper introduces a method for testing metric invariance using network psychometrics within the Exploratory Graph Analysis (EGA) framework. The method uses network loadings, analogous to factor loadings in traditional structural equation modeling (SEM). The study compares this method to traditional SEM approaches and finds that it offers comparable or improved power, especially in scenarios with smaller or unequal sample sizes and lower noninvariance effect sizes. Metric invariance assesses whether a latent variable is measured equivalently across groups. Traditional methods for testing metric invariance include SEM, which involves four tests: configural, metric, scalar, and strict. However, these methods have limitations, such as reliance on data-driven methods and unequal group sample sizes, which reduce the power of traditional SEM methods. The proposed method uses permutation testing to assess the equivalence of network loadings across groups. This method is more conceptually consistent with the EGA framework and does not require intensive model specifications or comparisons. It is straightforward to implement in R and provides a way to test metric invariance in the EGA framework. The study compares the proposed method to traditional SEM methods and finds that it performs well in detecting noninvariant items, especially in conditions with small or unequal sample sizes and lower noninvariance effect sizes. The method uses a permutation test to assess the equivalence of network loadings across groups and provides a way to test metric invariance in the EGA framework. The study also discusses the multiple comparison problem in partial invariance testing and the use of the Benjamini-Hochberg procedure to control the false discovery rate. The results show that the proposed method has higher accuracy in detecting noninvariant items compared to traditional SEM methods, particularly in conditions with small or unequal sample sizes and lower noninvariance effect sizes. The method is more robust to these conditions and provides a more accurate assessment of metric invariance.