01 / 21
C E

Experimental Design
Controls, Replicates, and Confounds

Master the foundations of rigorous scientific experiments: learn to isolate variables, build proper controls, use replication for statistical power, and identify confounding variables.

AP Biology

What Is a Controlled Experiment?

A controlled experiment is a test in which only one factor is deliberately changed while all other conditions are kept constant. The goal is to determine a cause-and-effect relationship between that factor and the observed outcome.

The Three Pillars of Good Design

Controls Baseline groups that receive no treatment, used for comparison. Replicates Repeated trials or samples that increase reliability and precision. Confounds Uncontrolled variables that distort results and must be eliminated. RIGOROUS EXPERIMENTAL DESIGN Three Essential Components

Why Controls Are Essential

Without a control group, you cannot determine whether the observed effect was caused by the independent variable or by some other factor. Controls provide the reference point against which experimental results are compared.

Positive Control
A treatment known to produce the expected result. Confirms the experiment can detect an effect.
Negative Control
A treatment expected to produce no effect. Establishes the baseline and checks for contamination.

Replication: Building Statistical Power

Replication means performing the same experiment multiple times under identical conditions. More replicates reduce the impact of random variation and increase confidence that results are real.

\( n = 1 \)
 ? Low confidence
\( n = 3 \)
 Moderate
\( n = 10+ \)
 High confidence Increasing sample size

What Are Confounding Variables?

A confounding variable is an uncontrolled factor that correlates with both the independent and dependent variables, making it impossible to determine which variable actually caused the observed result.

Independent Dependent Causal? Confound Confound distorts the cause-effect link Without controlling confounds, results are ambiguous.

The Core Rule of Experimental Design

1
Single variable principle: Only the independent variable differs between experimental and control groups.
2
Control requirement: Both positive and negative controls validate that the system works and the baseline is clean.
3
Replication requirement: Multiple replicates (\( n \geq 3 \)) are needed to calculate meaningful statistics.

Positive vs Negative Controls

Positive Control Negative Control
Purpose Confirms the assay detects a known effect Confirms baseline (no effect expected)
Expected Result Known positive outcome No change / no signal
If It Fails Assay is broken: cannot trust results Contamination or systematic error present
Enzyme Example Known substrate + enzyme Buffer only, no enzyme

Biological vs Technical Replicates

Biological Replicates
Independent biological samples: different organisms, cell cultures, or patients. They capture biological variability and allow generalization.
Technical Replicates
Repeated measurements from the same biological sample. They capture measurement error and assess assay precision.

Step-by-Step Design Checklist

1
State a testable hypothesis: Define the predicted cause-effect relationship clearly.
2
Identify all variables: Name the independent, dependent, and all controlled variables.
3
Design control groups: Include at least one positive and one negative control.
4
Plan replicates: Use \( n \geq 3 \) biological replicates per condition.
5
Scan for confounds: Systematically list potential confounding variables and how to control each one.
6
Randomize and blind: Randomly assign subjects and blind the experimenter where possible.

Example 1: Enzyme Activity Assay

Hypothesis: Drug X inhibits the activity of enzyme lactase, reducing the rate of lactose breakdown.

Experimental Lactase + Lactose + Drug X Measure glucose production rate
\( n = 5 \)
 Negative Ctrl Lactase + Lactose + Buffer (no drug) Baseline activity (no inhibition)
\( n = 5 \)
 Positive Ctrl Lactase + Lactose + Known inhibitor Confirms assay detects inhibition
\( n = 5 \)
 Controlled: temperature, pH, enzyme concentration, incubation time, lactose concentration

Example 2: Identifying Confounds

Scenario: A student claims that sunlight increases bacteria growth because plates left near a window had more colonies than plates in a drawer. What's wrong?

Example 3: Full Design for Protein Expression

Question: Does growth factor Y increase expression of protein Z in human cell lines?

H: Growth Factor Y increases Protein Z Experimental Cells + Growth Factor Y
\( n = 3 \)
 Negative Ctrl Cells + Vehicle (buffer only)
\( n = 3 \)
 Positive Ctrl Cells + Known activator of Z
\( n = 3 \)
 Western blot: Measure Protein Z levels (3 technical replicates each) Controlled: cell type, passage number, seeding density, media, incubation time, temperature, antibody lot

Eliminating Bias: Randomization and Blinding

Randomization
Subjects are randomly assigned to experimental or control groups. This ensures pre-existing differences (genetic variation, health status) are evenly distributed and do not bias the outcome.
Blinding
The experimenter (and ideally the subject) does not know which group receives the treatment. This prevents unconscious bias during measurement and analysis. A double-blind design blinds both.

Common Mistakes in Experimental Design

✕ No Control Group
Observing an effect without a baseline comparison. You cannot tell if the treatment caused the change or if it would have occurred anyway.
✕ Pseudoreplication
Treating multiple measurements from a single sample as independent data points. Measuring one culture 10 times does not give you \( n = 10 \).
✕ Changing Two Variables
Altering both drug concentration and temperature simultaneously. You cannot determine which variable caused the effect.

Summary Comparison

Concept What It Does How to Implement
Negative Control Establishes no-effect baseline Omit the treatment; use vehicle only
Positive Control Validates that assay can detect a signal Use a known active agent
Biological Replicates Captures biological variability Use independent samples (\( n \geq 3 \))
Technical Replicates Assesses measurement precision Repeat measurement on same sample
Confound Control Prevents false conclusions Standardize, randomize, blind

Question 1 of 5

In an experiment testing whether Drug A lowers blood glucose, a group of mice receives a saline injection instead of the drug. This group is best described as:

Question 2 of 5

A researcher measures protein expression in one cell culture by running the Western blot four times. The researcher then reports \( n = 4 \). What error has been made?

Question 3 of 5

A student tests whether fertilizer increases plant height. The fertilized plants are kept in a sunny greenhouse while control plants are in a shaded room. What is the main problem?

Question 4 of 5

In an enzyme kinetics experiment, the positive control tube fails to show any product formation. What is the most appropriate conclusion?

Question 5 of 5

Which of the following best explains why randomization is important when assigning subjects to groups?