Math 338- Ch 1 intro to data

Six Steps of a Statistical Investigation

  • Step 1: Ask a Research Question

    • Formulate a question that can be addressed through data collection.

  • Step 2: Design a Study and Collect Data

    • Outline a study plan and gather relevant information.

  • Step 3: Explore the Data

    • Analyze the collected data for trends and patterns.

  • Step 4: Draw Inferences

    • Make conclusions or predictions based on the data analysis.

  • Step 5: Formulate Conclusions

    • Develop conclusions that take into account the scope of the findings from step 4.

  • Step 6: Look Back and Ahead

    • Identify limitations of the study and potential future research opportunities.

Example: Organ Donation Study

  • Research Question:

    • Does the way we ask questions about organ donation affect the choice to participate?

  • Study Design and Data Collection:

    • 161 participants divided into three groups:

      • Opt-in group: Check if you wish to be an organ donor. (23 out of 55 agreed)

      • Opt-out group: Check if you do not wish to be an organ donor. (41 out of 50 agreed)

      • Neutral group: Yes or No to organ donation. (44 out of 56 agreed)

  • Inferences:

    • Participants in the Opt-in group were less likely to agree to be organ donors.

  • Conclusions:

    • Recommend rephrasing organ donation questions to increase participation rates.

  • Limitations:

    • Consider possible biases in participant selection and question phrasing.

Observations, Variables, and Data Matrices

  • Statistics:

    • The science of reasoning from data, where data is defined as information.

  • Cases and Variables:

    • Cases: The objects described by a set of data (e.g., individuals).

    • Variables: Characteristics observed from each case.

Types of Variables

  • Categorical Variables: Defines categories for individuals.

    • Nominal Variable: Categories without natural ordering (e.g., Smoking Status).

    • Ordinal Variable: Categories with a natural order (e.g., Course Ratings).

  • Numerical Variables: Takes on numerical values.

    • Discrete Variable: Countable values (e.g., number of siblings).

    • Continuous Variable: Values in an interval (e.g., weight in pounds).

Matrix of Observations

  • Example: Organ Donation Study

    • Characteristics collected:

      • Group type (Opt-in, Opt-out, Neutral)

      • Indication of agreement to be a donor (Yes/No).

Explanatory and Response Variables

  • Explanatory Variable: Suspected to affect another variable.

  • Response Variable: The variable being tested or measured.

Example from Beer Consumption Study:

  • Explanatory Variable: Alcohol/beer consumption.

  • Response Variable: Blood alcohol level.

Observational Studies and Experiments

  • Observational Study: Researchers measure without interference to find associations.

  • Designed Experiment: Researchers manipulate conditions to find cause-and-effect relationships.

Types of Studies

  • Association vs Causation:

    • Observational studies suggest associations but do not prove causation.

    • Experiments can establish causal connections.

Sampling Principles and Strategies

  • Populations and Samples:

    • Population: A group of observational units of interest.

    • Sample: A subset of the population used for statistical analysis.

Principles of Sampling

  • Simple Random Sampling: Ensures every observational unit has an equal chance of being selected, minimizing bias.

  • Stratified Sampling: Population divided into strata; randomly select units from each stratum.

  • Cluster Sampling: Randomly select clusters and include all units within selected clusters.

Confounding Variables

  • Confounding Variable: Related to both explanatory and response variables; affects the ability to identify effects.

  • Example: Difficulty of Sudoku as confounded in a study measuring completion time.

Experiments

  • Principles of Experimental Design:

    • Controlling: Minimize the effect of outside variables.

    • Randomization: Randomly assign subjects to treatment groups.

    • Replication: Collect large samples to ensure results are reliable.

    • Blocking: Group subjects based on similar characteristics before randomization to control variables.

Blind Studies and Placebos

  • Blind Study: Participants are unaware of their treatment assignment.

  • Placebo Effect: Patients experience improvement from belief in treatment efficacy.

  • Double-Blind Study: Both researchers and participants are unaware of treatment assignments.