Stats to the Point (E3) | What is Multilevel Modeling (sub FR/EN)

Conducting two-level regression using R, Stata, Mplus, or SPSS Nicolas Sommet – SNSF Ambizione Lecturer - Université de Lausanne 𝗖𝗵𝗮𝗽𝘁𝗲𝗿𝘀 00:00 Introduction 00:39 How Two-Level Modeling Works 04:05 A Three-Step Procedure for Conducting Two-Level Analysis 05:23 Step #0 Centering Predictors 07:08 Step #1 Empty Model 09:07 Step #2 Intermediate Models 11:34 Step #3 Final Model 12:56 Conclusion This video is based on the following article (if you've found this video useful, feel free to cite it!): Sommet, N., & Morselli, D. (2021). Keep calm and learn multilevel linear modeling: A three-step procedure using SPSS, Stata, R, and MPlus. 𝘐𝘯𝘵 𝘙𝘦𝘷 𝘚𝘰𝘤 𝘗𝘴𝘺𝘤𝘩𝘰𝘭, 34(1). https://doi.org/10.5334/irsp.555 The data/scripts from the video are available at: https://doi.org/10.17605/OSF.IO/4YHBM • CSV: https://osf.io/ch3sj • R: https://osf.io/9hmjn/ • Stata: https://osf.io/k2mp7/ • Mplus: https://osf.io/64eu2/ • SPSS: https://osf.io/fycvx/ Satisfaction survey → https://unilaps.fra1.qualtrics.com/jf... 𝗔𝗥𝗧𝗜𝗖𝗟𝗘𝗦 𝗦𝗨𝗣𝗣𝗢𝗥𝗧𝗜𝗡𝗚 𝗧𝗛𝗘 𝗖𝗟𝗔𝗜𝗠𝗦 𝗠𝗔𝗗𝗘 𝗜𝗡 𝗧𝗛𝗘 𝗩𝗜𝗗𝗘𝗢 ~ 3m27. If you choose to use traditional regression with hierarchical data, you will obtain biased standard errors. 👉 Scariano, S. M., & Davenport, J. M. (1987). The effects of violations of independence assumptions in the one-way ANOVA. 𝘈𝘮 𝘚𝘵𝘢𝘵, 41, 123–129. DOI: https://doi.org/10.1080/00031305.1987... ~ 5m47. The distinction between grand-mean centering and cluster-mean centering. 👉 Enders, C. K., & Tofighi, D. (2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. 𝘗𝘴𝘺𝘤𝘩𝘰𝘭 𝘔𝘦𝘵𝘩𝘰𝘥𝘴, 12, 121–128. https://doi.org/10.1037/1082-989x.12.... ~ 8m51. A design effect greater than 1.5 suggests that multilevel modeling is needed. 👉 Lai, M. H., & Kwok, O. M. (2015). Examining the rule of thumb of not using multilevel modeling: The “design effect smaller than two” rule. 𝘑 𝘌𝘹𝘱 𝘌𝘥𝘶𝘤, 83, 423–438. https://doi.org/10.1080/00220973.2014... ~ 9m56. To estimate the between-cluster variation of an effect, one needs to compare a constrained model and an augmented model without and with the random slope, respectively. 👉 Aguinis, H., Gottfredson, R. K., & Culpepper, S. A. (2013). Best-practice recommendations for estimating cross-level interaction effects using multilevel modeling. 𝘑 𝘔𝘢𝘯𝘢𝘨𝘦, 39, 1490–1528. https://doi.org/10.1177/0149206313478188 ~ 10m36. In a random slope model, the covariance parameter cannot be presumed to be zero and must be included in the model. 👉 Wang, L., Yang, M., & Liu, X. (2019). The impact of over-simplifying the between-subject covariance structure on inferences of fixed effects in modeling nested data. 𝘚𝘵𝘳𝘶𝘤𝘵 𝘌𝘲𝘶 𝘔𝘰𝘥𝘦𝘭𝘪𝘯𝘨, 26, 1–11. https://doi.org/10.1080/10705511.2018... ~11m07 If the LR χ² is significant at α = 0.20, the random slope variance and covariance terms need to be kept. 👉 Matuschek, H., Kliegl, R., Vasishth, S., Baayen, H., & Bates, D. (2017). Balancing Type I error and power in linear mixed models. 𝘑 𝘔𝘦𝘮 𝘓𝘢𝘯𝘨, 94, 305–315. https://doi.org/10.1016/j.jml.2017.01... ~11m07 It is also important to note that it is advisable to always include the random slope when testing a cross-level interaction. 👉 Heisig, J. P., & Schaeffer, M. (2019). Why you should always include a random slope for the lower-level variable involved in a cross-level interaction. 𝘌𝘶𝘳 𝘚𝘰𝘤𝘪𝘰𝘭 𝘙𝘦𝘷, 35, 258–279. https://doi.org/10.1093/esr/jcy053 ~12m40. A mock paper offering a good example of how to report multilevel analyses. 👉 Abelkermit, J. R., Hazesc, J. S., Prikkitrack, C. A., Etafon, J. A., & Ssab, J. B. (2021b). The Justin Timberlake effect. 𝘐𝘑𝘉𝘚𝘚𝘙, 2, 8. https://osf.io/n62ph 𝗔𝗗𝗗𝗜𝗧𝗜𝗢𝗡𝗔𝗟 𝗥𝗘𝗦𝗢𝗨𝗥𝗖𝗘𝗦 𝙈𝙪𝙡𝙩𝙞𝙡𝙚𝙫𝙚𝙡 𝙇𝙤𝙜𝙞𝙨𝙩𝙞𝙘 𝙍𝙚𝙜𝙧𝙚𝙨𝙨𝙞𝙤𝙣 👉 Sommet, N., & Morselli, D. (2017). Keep calm and learn multilevel logistic modeling: A simplified three-step procedure using Stata, R, Mplus, and SPSS. 𝘐𝘯𝘵 𝘙𝘦𝘷 𝘚𝘰𝘤 𝘗𝘴𝘺𝘤𝘩𝘰𝘭, 30, 203–218. https://doi.org/10.5334/irsp.90 𝙈𝙪𝙡𝙩𝙞𝙡𝙚𝙫𝙚𝙡 𝙈𝙤𝙙𝙚𝙡𝙞𝙣𝙜 𝙄𝙣𝙫𝙤𝙡𝙫𝙞𝙣𝙜 𝙏𝙝𝙧𝙚𝙚 𝙤𝙧 𝙈𝙤𝙧𝙚 𝙇𝙚𝙫𝙚𝙡𝙨 👉 Peugh, J. L. (2014). Conducting three-level cross-sectional analyses. 𝘑 𝘌𝘢𝘳𝘭𝘺 𝘈𝘥𝘰𝘭𝘦𝘴𝘤, 34, 7-37. https://doi.org/10.1177/0272431613498646 𝘾𝙧𝙤𝙨𝙨-𝘾𝙡𝙖𝙨𝙨𝙞𝙛𝙞𝙚𝙙 𝙈𝙤𝙙𝙚𝙡𝙨 👉 Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed 247 random effects for subjects and items. 𝘑 𝘔𝘦𝘮 𝘓𝘢𝘯𝘨, 59, 390-412. https://doi.org/10.1016/j.jml.2007.12... 𝘼𝙥𝙥𝙡𝙞𝙘𝙖𝙩𝙞𝙤𝙣 𝙤𝙛 𝙈𝙪𝙡𝙩𝙞𝙡𝙚𝙫𝙚𝙡 𝙈𝙤𝙙𝙚𝙡𝙞𝙣𝙜 𝙩𝙤 𝙍𝙚𝙥𝙚𝙖𝙩𝙚𝙙 𝙈𝙚𝙖𝙨𝙪𝙧𝙚𝙨 👉 Quené, H., & Van den Bergh, H. (2004). On multi-level modeling of data from repeated measures designs: A tutorial. 𝘚𝘱𝘦𝘦𝘤𝘩 𝘊𝘰𝘮𝘮𝘶𝘯, 43, 103-121. https://doi.org/10.1016/j.specom.2004... 𝙎𝙖𝙢𝙥𝙡𝙚 𝙎𝙞𝙯𝙚 𝙞𝙣 𝙈𝙪𝙡𝙩𝙞𝙡𝙚𝙫𝙚𝙡 𝙈𝙤𝙙𝙚𝙡𝙞𝙣𝙜 👉 Arend, M. G., & Schäfer, T. (2019). Statistical power in two-level models: A tutorial based on 244Monte Carlo simulation. 𝘗𝘴𝘺𝘤𝘩𝘰𝘭 𝘔𝘦𝘵𝘩𝘰𝘥𝘴, 24, 1-19. https://doi.org/10.1037/met0000195 Steering : https://www.adrips.org/ Funding: https://u-paris.fr/aap-innovations-pe... Production : https://www.lacasquette.fr/