This paper investigates how multilevel models(MLMs)handle hierarchical and longitudinal data,such as repeated measures nested in individuals,which are common in social science research.Effective sample size planning i...This paper investigates how multilevel models(MLMs)handle hierarchical and longitudinal data,such as repeated measures nested in individuals,which are common in social science research.Effective sample size planning is critical for MLMs,with power analysis serving to determine the necessary sample sizes.However,research on sample size planning for MLMs with ordinal outcomes is limited,despite its increasing popularity for applied researchers.Additionally,many studies examine the cross-lagged effects,which trace how changes in one variable at an earlier time influence another variable later.To address these issues,we conducted a simulation study to investigate how sample size,the autoregressive(AR)effect,and cross-lagged effects influence statistical power within a multilevel autoregressive framework.The results provide practical guidance for researchers designing longitudinal studies with ordinal outcomes.Furthermore,we developed OrdPower,an easy-to-use R package,which social science researchers can use to plan sample sizes for MLMs with ordinal outcomes,especially when cross-lagged effects are the primary focus.展开更多
文摘This paper investigates how multilevel models(MLMs)handle hierarchical and longitudinal data,such as repeated measures nested in individuals,which are common in social science research.Effective sample size planning is critical for MLMs,with power analysis serving to determine the necessary sample sizes.However,research on sample size planning for MLMs with ordinal outcomes is limited,despite its increasing popularity for applied researchers.Additionally,many studies examine the cross-lagged effects,which trace how changes in one variable at an earlier time influence another variable later.To address these issues,we conducted a simulation study to investigate how sample size,the autoregressive(AR)effect,and cross-lagged effects influence statistical power within a multilevel autoregressive framework.The results provide practical guidance for researchers designing longitudinal studies with ordinal outcomes.Furthermore,we developed OrdPower,an easy-to-use R package,which social science researchers can use to plan sample sizes for MLMs with ordinal outcomes,especially when cross-lagged effects are the primary focus.
