Data Collection and Experimental Design in Statistics

Data Collection and Experimental Design

Introduction

This section introduces the foundational concepts of designing statistical studies, distinguishing between observational studies and experiments, and understanding various data collection and sampling techniques. Mastery of these topics is essential for conducting valid and reliable statistical analyses.

Designing a Statistical Study

Steps in Designing a Study

• Identify the variable(s) of interest and the population of the study.

• Develop a detailed plan for collecting data. If using a sample, ensure it is representative of the population.

• Collect the data.

• Describe the data using descriptive statistics techniques.

• Interpret the data and make decisions about the population using inferential statistics.

• Identify any possible errors.

Types of Statistical Studies

Observational Study

• A researcher observes and measures characteristics of interest of part of a population without influencing them.

• Example: Measuring the amount of time people spend on various activities (e.g., paid work, childcare, socializing).

Experiment

• A treatment is applied to part of a population (the treatment group), and responses are observed.

• Another part of the population may serve as a control group, receiving no treatment or a placebo (a harmless, fake treatment).

• All subjects in both groups are called experimental units.

• Example: Testing the effect of sucralose on glycemic and insulin responses in overweight subjects.

Distinguishing Study Types

• If a treatment is applied, it is an experiment.

• If no treatment is applied and only observation occurs, it is an observational study.

Data Collection Methods

Simulation

• Uses mathematical or physical models to reproduce conditions of a situation or process.

• Often involves computers and is useful for studying impractical or dangerous real-life situations.

• Example: Automobile crash simulations using dummies.

Survey

• An investigation of one or more characteristics of a population, typically by asking questions.

• Common methods: interview, Internet, phone, or mail.

• Survey design must avoid biased questions to ensure representativeness.

• Example: Surveying female physicians about career choice motivations.

Experimental Design Principles

Key Elements

• Control: Managing variables to minimize confounding effects.

• Randomization: Randomly assigning subjects to treatment groups.

• Replication: Repeating the experiment with a large group to validate results.

Confounding Variables

• Occur when the effects of different factors on a variable cannot be distinguished.

• Example: Increased business at a coffee shop after remodeling and a nearby mall opening—cannot determine which caused the increase.

Placebo Effect and Blinding

• Placebo effect: Subjects respond to a fake treatment as if it were real.

• Blinding: Subjects do not know if they receive treatment or placebo.

• Double-blind: Neither subjects nor experimenters know group assignments.

Randomization Techniques

• Completely Randomized Design: Subjects assigned to groups by random selection.

• Randomized Block Design: Subjects divided into blocks by similar characteristics, then randomly assigned within blocks.

• Matched-Pairs Design: Subjects paired by similarity; one receives treatment, the other does not.

Sample Size and Replication

• Sample size: Larger samples increase the reliability of results.

• Replication: Repeating the experiment with many subjects to confirm findings.

Sampling Techniques

Definitions

• Census: Count or measure of an entire population.

• Sampling: Count or measure of part of a population.

• Sampling error: Difference between sample results and population results.

Types of Samples

• Random Sample: Every member of the population has an equal chance of selection.

• Simple Random Sample: Every possible sample of the same size has an equal chance of selection.

• Stratified Sample: Population divided into groups (strata); random samples taken from each group.

• Cluster Sample: Population divided into clusters; all members from one or more clusters are selected.

• Systematic Sample: Starting point chosen at random; every kth member selected.

• Convenience Sample: Members easy to get are chosen; often leads to bias.

Examples of Sampling Techniques

• Stratified: Divide students by major, randomly sample from each major.

• Simple Random: Assign numbers to students, randomly select numbers.

• Convenience: Select students from your own class.

Table: Sampling Techniques Comparison

Sampling Technique	Description	Example
Simple Random	Every sample of size n has equal chance	Randomly select student numbers
Stratified	Divide into strata, sample from each	Sample students by major
Cluster	Divide into clusters, select all in some clusters	Sample all students in selected zip codes
Systematic	Select every kth member	Every 100th household
Convenience	Easy-to-get members	Students in your class

Key Formulas

• Sampling Error: $\text{Sampling Error} = \text{Sample Statistic} - \text{Population Parameter}$

Summary

• Proper design and sampling are crucial for valid statistical inference.

• Understanding study types, data collection methods, and sampling techniques helps avoid bias and confounding.

• Replication and adequate sample size increase the reliability of experimental results.