Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model
Abstract
We compared three statistical analyses over binary outcomes. As applying ANOVA over proportions violates at least two classical assumptions of linear models, two alternatives are described: the binary logistic regression and the mixed logit model. Firstly, we compared the effects obtained by the three methods over the same data from a previous memory research. All three methods gave similar results: the effects of the tasks and the number of sensory modalities were observed, but not their interaction. Secondly, by using the bootstrap estimates of the parameters, the efficacy of each method was explored. As predicted, the bootstrap parameter estimates of the ANOVA had large bias and standard errors, and consequently wide confidence intervals. On the other hand, the bootstrap parameter estimates of the binary logistic regression and the mixed logit models were similar - both had low bias and standard errors and narrow confidence intervals.
Keywords:
mixed logit model / logistic regression / cued/free recall / bootstrapping / ANOVASource:
Psihologija, 2018, 51, 4, 469-+Publisher:
- Društvo psihologa Srbije, Beograd
Funding / projects:
- Fundamental cognitive processes and functions (RS-MESTD-Basic Research (BR or ON)-179033)
- Psychological foundations of mental health: hereditary and environmental factors (RS-MESTD-Basic Research (BR or ON)-179006)
DOI: 10.2298/PSI170615023P
ISSN: 0048-5705
WoS: 000453737800005
Scopus: 2-s2.0-85059258563
Institution/Community
Psihologija / PsychologyTY - JOUR AU - Popović-Stijacić, Milica AU - Mihić, Ljiljana AU - Filipović Đurđević, Dušica PY - 2018 UR - http://reff.f.bg.ac.rs/handle/123456789/2608 AB - We compared three statistical analyses over binary outcomes. As applying ANOVA over proportions violates at least two classical assumptions of linear models, two alternatives are described: the binary logistic regression and the mixed logit model. Firstly, we compared the effects obtained by the three methods over the same data from a previous memory research. All three methods gave similar results: the effects of the tasks and the number of sensory modalities were observed, but not their interaction. Secondly, by using the bootstrap estimates of the parameters, the efficacy of each method was explored. As predicted, the bootstrap parameter estimates of the ANOVA had large bias and standard errors, and consequently wide confidence intervals. On the other hand, the bootstrap parameter estimates of the binary logistic regression and the mixed logit models were similar - both had low bias and standard errors and narrow confidence intervals. PB - Društvo psihologa Srbije, Beograd T2 - Psihologija T1 - Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model EP - + IS - 4 SP - 469 VL - 51 DO - 10.2298/PSI170615023P ER -
@article{ author = "Popović-Stijacić, Milica and Mihić, Ljiljana and Filipović Đurđević, Dušica", year = "2018", abstract = "We compared three statistical analyses over binary outcomes. As applying ANOVA over proportions violates at least two classical assumptions of linear models, two alternatives are described: the binary logistic regression and the mixed logit model. Firstly, we compared the effects obtained by the three methods over the same data from a previous memory research. All three methods gave similar results: the effects of the tasks and the number of sensory modalities were observed, but not their interaction. Secondly, by using the bootstrap estimates of the parameters, the efficacy of each method was explored. As predicted, the bootstrap parameter estimates of the ANOVA had large bias and standard errors, and consequently wide confidence intervals. On the other hand, the bootstrap parameter estimates of the binary logistic regression and the mixed logit models were similar - both had low bias and standard errors and narrow confidence intervals.", publisher = "Društvo psihologa Srbije, Beograd", journal = "Psihologija", title = "Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model", pages = "+-469", number = "4", volume = "51", doi = "10.2298/PSI170615023P" }
Popović-Stijacić, M., Mihić, L.,& Filipović Đurđević, D.. (2018). Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model. in Psihologija Društvo psihologa Srbije, Beograd., 51(4), 469-+. https://doi.org/10.2298/PSI170615023P
Popović-Stijacić M, Mihić L, Filipović Đurđević D. Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model. in Psihologija. 2018;51(4):469-+. doi:10.2298/PSI170615023P .
Popović-Stijacić, Milica, Mihić, Ljiljana, Filipović Đurđević, Dušica, "Analyzing data from memory tasks - comparison of ANOVA, logistic regression and mixed logit model" in Psihologija, 51, no. 4 (2018):469-+, https://doi.org/10.2298/PSI170615023P . .