Statistical Power to Detect the Correct Number of Classes in Latent Profile Analysis

By J. Tein, Stefany Coxe, Heining Cham. Structural equation modeling : a multidisciplinary journal, 2013

1,719 citations recorded by the source corpus.

Little research has examined factors influencing statistical power to detect the correct number of latent classes using latent profile analysis (LPA). This simulation study examined power related to interclass distance between latent classes given true number of classes, sample size, and number of indicators. Seven model selection methods were evaluated. None had adequate power to select the correct number of classes with a small (Cohen's d = .2) or medium (d = .5) degree of separation. With a very large degree of separation (d = 1.5), the Lo–Mendell–Rubin test (LMR), adjusted LMR, bootstrap likelihood ratio test, Bayesian Information Criterion (BIC), and sample-size-adjusted BIC were good at selecting the correct number of classes. However, with a large degree of separation (d = .8), power depended on number of indicators and sample size. Akaike's Information Criterion and entropy poorly selected the correct number of classes, regardless of degree of separation, number of indicators, or sample size.

Subfields: Molecular Biology, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/cc2fb25644330c040d413cedebd2f5260fb50f86

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

1. Margarita Olivera‐Aguilar
2. E. W. G. Moore
3. Kevin Delucchi
4. D. Hull
5. Sarah Ferguson
6. Alexandre J. S. Morin
7. Samuel Rikoon
8. P. Sinha

In Statistics and Probability

1. David Litalien
2. Alexandre J. S. Morin
3. Marylène Gagné
4. I. Madore
5. Véronique Dagenais‐Desmarais
6. Margarita Olivera‐Aguilar
7. E. W. G. Moore
8. K. Schnitzler

Citations in the corpus

• Ten Frequently Asked Questions About Latent Class Analysis (2018, neutral): “…studies have shown that beyond sample size per se, the likelihood of accurately recovering the correct number of mixture classes depends heavily on class separation (Depaoli, 2013; Lubke & Muthén, 2007; Lubke & Neale, 2006; Tein et al., 2013; Tofighi & Enders, 2008; Tueller & Lubke, 2010).”
• Motivation profiles at work: A self-determination theory approach (2016, result): “Overall, these studies converge in supporting the efficacy of the CAIC, the BIC, the ABIC, and the BLRT in choosing the model which best recovers the sample's true parameters (e.g., Henson, Reise, & Kim, 2007; McLachlan & Peel, 2000; Morgan, 2015; Nylund, Asparouhov, & Muthén, 2007; Peugh & Fan, 201”
• All better than being disengaged: Student engagement patterns and their relations to academic self-concept and achievement (2020, background): “Despite this reduction, the sample size is still comparable to the median sample size in other LPAs, which is seen as sufficient (Tein et al. 2013).”
• Sensitivity and Specificity of Information Criteria (2018, neutral): “In latent profile analysis, Tein et al. (2013) found that BIC and ABIC did well for large sample sizes and easily distinguishable classes, but AIC chose too many classes, and no method performed well for especially challenging scenarios.”
• Complementary Variable- and Person-Centered Approaches to the Dimensionality of Psychometric Constructs: Application to Psychological Wellbeing at Work (2016, background): “Simulation studies indicate that four of these indicators (CAIC, BIC, ABIC, and BLRT) are particularly effective and that when the indicators fail to retain the optimal model, the ABIC and BLRT tend to overestimate the number of classes, whereas the BIC, CAIC, and aLMR tend to underestimate it (Nylu”
• Performance of the entropy as an index of classification accuracy in latent profile analysis: A Monte Carlo simulation study (2017, neutral): “…情况 明 Entropy 在不同条件下的表现。据此, 本研究系 统考察了样本量、类别距离、指标数和类别数对 Entropy 及相关指数影响的情况。 4.1 样本量和指标数对 Entropy 的影响 正如我们预实验发现的那样, 样本量越大, 分 类精确性越差。尽管 Tein 及其同事(2013)主要研究 在什么条件下以及采用什么样的评价指标有利于 正确选择潜剖面模型的类别数, 但从他们拟合真模 型(k = 5)的结果可以发现, 样本量越大(N = 250, 500, 1000), Entropy 的值越小。Lubke 和 Neale (2006)…”
• Measurement Invariance Testing with Many Groups: A Comparison of Five Approaches (2017, result): “Several studies supported the BLRT for different models (Kim et al., 2016; Nylund et al., 2007; Tein et al., 2013), but it was reported that the execution time was extremely long with a large sample, nonconvergence frequently occurred, and the performance of BLRT was often comparable to that of…”
• Finding latent groups in observed data: A primer on latent profile analysis in Mplus for applied researchers (2020, methodology): “Values of .80 or greater provide supporting evidence that profile classification of individuals in the model occurs with minimal uncertainty (Celeux & Soromenho, 1996; Tein et al., 2013).”
• Assessing Measurement Invariance in Multiple-Group Latent Profile Analysis (2018, methodology): “Tein et al. (2013) found that the statistical power to detect the correct number of classes in LPA depended on the distance between profiles.”
• Practitioner’s Guide to Latent Class Analysis: Methodological Considerations and Common Pitfalls. (2020, methodology): “Again, in simulation studies with latent profile analysis (continuous indicators), Tein et al (56) found that the BLMR had higher statistical power than LMR, although both tests performed well at detecting the true class.”