Study Notes on Statistical Studies and Experimental Issues

Focus on issues related to running statistical studies, observational studies, and experiments.

Importance of a representative sample to accurately reflect the population.
- Key to obtaining useful and valid results.
Random Sampling
- Definition: A method where every individual in the population has the same chance of being selected.
- Example: Rolling a die repeatedly ensures each side has equal representation.
Different Sampling Types
- Convenience Sampling: Choosing a sample that is easy to access, which may lead to bias.
  - Example: Measuring grass height directly in a well-mowed area may lead to results skewed toward a specific height (around 3 inches).
- Systematic Sampling: Choosing every nth individual, but can inherit bias because of overlaps.
  - Example: Mowing the lawn and capturing measurements every five feet may lead to repeated measurements from overlapped areas.
- Voluntary Sampling: Sampling where individuals choose to respond, which can create bias towards strong opinions.
  - Example: Asking for opinions on Donald Trump leads to responses from only those who feel strongly, neglecting moderate perspectives.
- Cluster Sampling: Dividing the population into clusters and randomly selecting whole clusters.
- Stratified Sampling: Dividing the population into strata or groups based on shared characteristics and randomly sampling from those groups.

Sampling Bias: When the sample does not represent the population adequately due to selection bias or measurability issues.
Participation Bias: Occurs when certain groups do not participate, leading to a skewed perspective based on who is included.
Non-response Bias: When individuals chosen for the sample do not respond or participate, potentially due to sensitive/questionable subjects.
Researcher Bias: When the researcher's knowledge affects how they conduct and interpret the study.
Question Wording Bias: Loaded questions can lead respondents towards certain answers, influencing results.
- Example: Asking if respondents really like pizza implies a positive response, influencing answers to conform to social norms.

Confounding Variables: Variables that can obscure the true relationship between the independent and dependent variables.
- Example: Sunlight impacting grass growth, which must be controlled in studies.
Types of Experiments
- Single Blind Experiment: The subject does not know which treatment they are receiving, but the researcher does. This may introduce bias.
- Double Blind Experiment: Neither the subjects nor the researchers know which treatment is being administered, minimizing bias from both ends.

Non-adherers are participants who do not follow the study protocols or drop out.
- Impact: Leads to incomplete data and potentially invalid outcomes.
Historical Example: Use of hot toddies for colds, leading to varied subjective experiences impacting study results.

Acknowledgment that various forms of bias can affect observational studies and experiments.
Importance of controlling these variables and making accommodations for potential issues discussed.
Encouragement to anticipate confounding variables and their potential impact on study validity.