Chapter 9-Producing data

Chapter 9: Producing Data

Lecture 1: Introduction to Statistical Data Collection Methods

Statistical data collection methods are essential for gathering, analyzing, and interpreting data in various fields, including social sciences, health, and market research. Understanding the distinctions between observational studies and experiments is crucial for valid data analysis.

9.1: Observation vs Experiment

There are two primary methods to collect statistical data:

1. Observational Study

• Definition: An observational study involves observing and measuring characteristics without modifying subjects or their conditions.

• Purpose: The main goal is to describe effects or situations as they naturally occur, providing descriptive insights without establishing causative relationships.

• Examples:

  • Conducting surveys to gather information about respondents’ preferences in presidential elections.

  • Sampling laboratory measurements from hospital patients without intervening in their treatment or behavior.

2. Experiments

• Definition: Experiments apply specific treatments to subjects and observe the resulting effects to determine causal relationships.

• Purpose: The aim is to establish whether a particular treatment (cause) directly influences an observed effect (outcome).

• Examples:

  • Vaccinating one group of laboratory animals while administering a placebo to another group for comparative analysis of vaccine efficacy.

Types of Observational Studies

1. Cross-sectional Study:

   • Definition: In a cross-sectional study, data observations are made at a single point in time.

   • Example: Collecting data on health conditions for all residents of Northside of Statesboro during the year 1990 to assess illness prevalence.

2. Retrospective Study:

   • Definition: This study collects data about a characteristic over a defined period in the past, allowing researchers to identify trends or patterns.

   • Example: Analyzing historical influenza cases over a five-year span, starting from 2014, to examine trends over time.

3. Prospective Study:

   • Definition: A prospective study collects data moving forward from a specific starting point in time, allowing for real-time analysis.

   • Examples:

     • Following a cohort of individuals with illnesses over five years beginning in 2014 to monitor health outcomes.

     • Conducting a long-term study of sick individuals from 1995 over the following years to determine health trajectories.

Example Exercise 9.1

Scenario: Archaeologists excavate and record objects used by an ancient settlement in Ethiopia, dated 20 BC.

• Type of Observational Study:

  • The data collection here can be considered a Retrospective Study since it involves analyzing historical data from the past.

9.2: Subjects, Factors, and Treatments in Experiments

Key Components of an Experiment:

1. Subjects (Individuals):

   • This term refers to the people, animals, or objects being studied in the experiment.

2. Factors (Explanatory Variables):

   • Factors are the conditions or variables that may cause changes in response.

   • Examples: Different study habits, health statuses, financial conditions, etc.

3. Response (Response Variable):

   • The specific effect that is observed or measured as a result of the treatments applied to the subjects.

4. Treatment:

   • Refers to the specific conditions or interventions delivered to subjects within the experimental design.

Examples of Treatments:

• Single Factor Experiment:

  • Title: Evaluating Study Methods

  • Factor: The type of study method (e.g., using a calculator vs. using traditional methods).

• Two Factor Experiment:

  • Title: Examining Study Habits and Health

  • Factors: The study method and the health status of the participants.

  • Treatments: Tailoring variations according to the use of calculators and having respective health conditions (sick vs. healthy).

Treatment Groups in Experiments

• Treatment Group:

  • This group receives the primary treatment or intervention being investigated (e.g., the group of subjects receiving a new vaccine).

• Control Group:

  • This group receives a placebo or no treatment, serving as a baseline for comparison against the treatment group.

• Response Variable:

  • This is the specific effect that is observed following the application of the treatment.

  • Example 9.1: An experiment assessing a vaccine’s effectiveness includes:

    • Subjects: Rats are often used for controlled studies due to their manageable size and biological similarities to humans.

    • Factor: The medication being tested (true vaccine versus a placebo).

    • Treatments: Involves giving one group the vaccine while the other group does not receive it for comparison.

    • Treatment Group: The group of rats that receives the vaccine, allowing researchers to analyze its effects.

    • Control Group: The group of rats that does not receive the vaccine, helping researchers understand the baseline health effects.

Chapter 9: Lecture 2: Setbacks in Experimental Design

Common Impediments to Results

• Several challenges may affect experimental outcomes, including:

  • Confounding Variables:

    • These are extraneous variables that can impact the response variable, complicating the interpretation of results.

    • Example: Certain medications might cause allergic reactions when combined with specific foods, resulting in unclear outcomes.

  • Lurking Variables:

    • These are unexpected variables that are not measured but can influence both the explanatory and response variables, often leading to misleading conclusions.

    • Example: In researching the impact of study hours on academic performance, factors such as sleep quality and financial stability may go unrecognized but significantly influence results.

Advantages of Experiments over Observational Studies

1. Control Over Variables:

   • Experiments allow researchers to systematically manage confounding and lurking variables by randomly assigning treatments to subjects.

2. Avoiding Correlation Misinterpretation:

   • Observational studies may conflate correlation with causation, while experiments provide more reliable data to distinguish cause-and-effect relationships.

3. Detailed Interaction Investigation:

   • Experiments allow researchers to study complex interactions between multiple factors, yielding clearer insights regarding their relationships.

Randomized Comparative Experiment

• This design involves multiple groups that compare different treatments, with subjects randomly assigned to reduce bias in results and increase reliability.

Characteristics of Good Experimental Design

1. Treatment and Control Groups:

   • Having both groups is vital for effective comparisons to determine the treatment's effectiveness.

2. Randomization:

   • This principle minimizes assignment bias during the subject selection process, improving the reliability of conclusions.

3. Replication:

   • Conducting experiments with a larger number of subjects helps validate findings and reduce variability in results.

4. Blinding:

   • This practice limits the influence of external variables on participant responses, which can include:

     • Double Blind: Neither the subjects nor the investigators are aware of treatment assignments, helping to prevent bias.

     • Single Blind: Only the subjects are unaware of their treatment assignment, limiting expectation pressures on results.

Principles of Statistical Experiment Design

• Control:

  • Establishing groups helps isolate specific treatment effects for clearer causal inference.

• Randomization:

  • Random assignment of subjects mitigates bias effects in results, fostering reliability.

• Replication:

  • Confirming responses across larger subject samples enhances confidence in the results.

Statistical Significance

• For differences in treatment responses to be meaningful, they must be large enough to preclude chance occurrence, indicating genuine significance.

  • Example: If 85% of newborns in a study are female, this finding would demonstrate statistical significance, as such a proportion is unlikely to occur by random chance alone.

Importance of Experimental Design

• Carefully structured randomized experiments yield valid conclusions about treatment effects and their underlying causes.

• Experimental research is more capable of establishing causation than observational studies, which primarily identify associations.

Example Exercise 9.2: Importance of Treatment and Control Groups

• Treatment and control groups are crucial for controlling bias, enabling direct comparisons of outcomes, and helping eliminate personal biases through effective blinding practices.

Types of Experimental Designs

1. Completely Randomized Design:

   • Involves random assignment of subjects to various treatment groups to minimize potential biases and ensure representativeness.

2. Matched Pairs Design:

   • Focuses on comparing two treatments by:

     • First Method: Pairing closely matched subjects that receive different treatments.

     • Second Method: Assigning one subject to receive both treatments in random order for comparison.

   • Example: An experiment assessing driver distraction, where subjects drive with and without cellphones on separate days, with random treatment assignments.

3. Block Design:

   • Involves grouping subjects with similar characteristics into blocks before random assignment, enhancing the comparison’s validity.

   • Example: Separating participants by gender before studying differences in advertising effectiveness.

Example Exercise 9.3: Study Design Involving Twins

• In studies assessing genetic factors, the Matched-Pairs Design is most suitable, cleaving subjects into pairs that closely resemble each other.

Chapter 9 Homework

• Page 246: Provide detailed explanations for questions: 9.19, 9.26, 9.27, 9.28.

Chapter 10: Data Ethics

Introduction

Ethical considerations for research involving human or animal subjects are critical for maintaining responsible research standards and practices.

Basic Data Ethics for Human Subjects

• Any study involving human participants must be reviewed by an Institutional Review Board (IRB) to ensure participant safety and welfare are prioritized.

Institutional Review Board (IRB)

• The IRB plays an essential role in protecting the rights of research participants, ensuring proper risk management, and obtaining informed consent.

Informed Consent

• It is imperative that participants are informed about the study’s purpose, potential risks involved, and how their data will be handled. Consent must be explicitly obtained, typically through signed documentation.

Confidentiality

• Protecting participants’ privacy is of the utmost importance, often achieved through methods such as data coding to handle sensitive information appropriately (e.g., data connecting to HIV tests).

• Researchers may be required to undertake privacy training to ensure effective confidentiality compliance.

Chapter 10 Homework

• Page 264: Provide detailed explanations for questions 10.9, 10.10, 10.14.