Observational Descriptive Design Flashcards
What is Observational Descriptive Design?
An observational descriptive design is a research method that aims to measure and summarize natural phenomena without allocating treatments
(eg, surveys, case studies, and text analysis projects)
There are two key characteristics to Observational Descriptive Design, what are they?
- Descriptive Inquiry: The research question aims to describe a population or phenomenon, not to determine the impact of a treatment or intervention.
- Data Collection: Data is typically collected through surveys, official statistics, or qualitative methods like “thick description” (detailed, contextual descriptions).
What type of sampling is usually used in observational descriptive design?
Cluster Sampling and Intra-Cluster Correlation (ICC)
What is ICC?
This statistic measures the degree to which units within the same cluster are more similar to each other than units from different clusters.
What is Multi-level Regression?
This technique can account for hierarchical data structures (e.g., individuals nested within counties). It estimates the effects of variables at both individual and cluster levels.
What is Post-Stratification?
This method enhances the accuracy of estimates for both individual and cluster levels by correcting them according to known population proportions.
What are Latent variables?
These are underlying concepts that cannot be directly measured. We use observable proxy indicators to measure them, but each proxy has its own limitations.
What is an index creation, and why is it important?
This combines multiple proxy indicators into a single score to represent the latent variable (Y*). The goal is to reduce measurement error by “canceling out” some of the error associated with individual proxies.
What are the different appraoches to Index creation?
- Scale and Average
- Scaled Average
- Principal Component Analysis (PCA)
What is the Scale and Average approach in relation to index creation?
Standardize each proxy indicator and then take the average of the standardized scores.
What is the Scaled Average approach in relation to index creation?
Similar to the scale and average, but additionally, adjust the final score based on a covariate (e.g., adjust for the effect of gender on the index).
What is PCA in relation to Index Creation?
This technique identifies the first factor (component) that captures the most variance shared by all proxy indicators. This first factor is then used as the index score.
What is partial pooling?
Partial pooling is an approach in which estimates are adjusted by combining information from both individual and group levels, striking a balance between no pooling and full pooling strategies.
