A data science team is analyzing data from a longitudinal study on a new hypertension drug's effectiveness. The dataset includes repeated blood pressure measurements for numerous participants over one year. The primary objective is to model the change in blood pressure over time, specifically accounting for the inherent correlation in measurements from the same individual. The chosen model must differentiate between fixed effects, such as the overall impact of the drug, and the random effects that represent individual variability in blood pressure trajectories. Which analytical approach is MOST suitable for this scenario?
Autoregressive Integrated Moving Average (ARIMA) model
The correct answer is a Linear Mixed-Effects Model (LMM). This model is specifically designed for longitudinal or hierarchical data where observations are not independent. It properly accounts for the within-subject correlation by modeling both fixed effects (population-level trends, like the average effect of the drug) and random effects (subject-specific deviations from the average trend).
Repeated Measures ANOVA (RMANOVA) is incorrect because, while it handles repeated measures, it is far less flexible than LMMs. RMANOVA requires balanced data (measurements at the same fixed time points for all subjects) and handles missing data poorly, often by deleting the entire subject's data if one measurement is missing (listwise deletion).
Pooled Ordinary Least Squares (OLS) Regression is incorrect because it violates a key assumption of OLS: the independence of observations. By treating all measurements as independent, it ignores the within-subject correlation, which leads to underestimated standard errors and an inflated risk of Type I errors (false positives).
An Autoregressive Integrated Moving Average (ARIMA) model is incorrect because it is designed for univariate time-series analysis, where there is a single sequence of observations over time (e.g., a company's total monthly sales). It is not suited for panel or longitudinal data that involves tracking multiple individual subjects simultaneously.
Ask Bash
Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.
What are fixed and random effects in a Linear Mixed-Effects Model (LMM)?
Open an interactive chat with Bash
Why is Repeated Measures ANOVA less flexible than a Linear Mixed-Effects Model?
Open an interactive chat with Bash
Why is independence of observations crucial in Ordinary Least Squares (OLS) regression, and why does LMM overcome this limitation?