Experimental Design

Scientific method: purpose

A structured process for investigating questions, testing explanations with evidence, and communicating conclusions.

Six steps of the scientific method

1. Ask a question or identify a problem; 2. Do background research; 3. Form a testable hypothesis; 4. Test it with an experiment; 5. Analyze the data; 6. Draw and communicate a conclusion.

Testable hypothesis

A specific, measurable prediction about the relationship between variables that can be supported or rejected by evidence.

Null hypothesis (H0)

A hypothesis stating that there is no effect, difference, or relationship between the variables being tested.

Alternative hypothesis (H1)

A hypothesis stating that there is an effect, difference, or relationship between the variables being tested.

Why use a null hypothesis?

It provides a baseline claim that experimental data can be tested against.

Do experiments always need both null and alternative hypotheses?

In formal hypothesis testing, yes: the null states “no effect,” while the alternative states the predicted effect or difference.

Fish temperature experiment: null hypothesis

The temperature of the water has no effect on the metabolic rate of fish, measured by oxygen consumption.

Fish temperature experiment: alternative hypothesis

The temperature of the water affects the metabolic rate of fish, measured by oxygen consumption.

Independent variable

The factor the researcher deliberately changes or compares in an experiment.

Dependent variable

The outcome that is measured or observed; it may change in response to the independent variable.

Controlled variable (constant)

A factor kept the same for every group so it does not influence the results.

Potato osmosis experiment: independent variable

The salt concentration of the solution.

Potato osmosis experiment: dependent variable

The mass change of the potato slices after 24 hours.

Graphing independent variables

The independent variable is placed on the x-axis.

Graphing dependent variables

The dependent variable is placed on the y-axis.

Constants vs. controls

Constants are conditions kept the same across all groups. Controls are comparison groups used to show what happens without the experimental treatment or with a known treatment.

Negative control

A group that does not receive the experimental treatment and is expected to show no effect; it helps detect confounding variables, contamination, or background effects.

Positive control

A group that receives a treatment known to produce a result; it confirms that the procedure and measurement system can detect the expected effect.

When to use a negative control

Use one when a researcher needs a baseline for comparison or needs to check that an observed effect is not caused by contamination or another uncontrolled factor.

When to use a positive control

Use one when a researcher needs to verify that the experiment is capable of producing and detecting a known result.

Central tendency

A statistic that describes the center or typical value of a data set.

Mean

Add all values and divide by the number of values; use it for numerical data without strong outliers or extreme skew.

Median

The middle value when data are ordered from least to greatest; use it when data contain outliers or are skewed.

Mode

The value that occurs most often; use it to identify the most common value, especially for categories or repeated scores.

Outlier

A data value much higher or lower than the rest of the data; it can strongly affect the mean.

Measure of variability

A statistic describing how spread out data values are, such as range, variance, or standard deviation.

Standard deviation

A measure of how far data values typically fall from the mean.

Low standard deviation

Data points cluster closely around the mean, indicating low variability and more consistent measurements.

High standard deviation

Data points are spread far from the mean, indicating high variability and less consistent measurements.

Why does low standard deviation suggest reliable data?

It shows repeated measurements are similar to one another, so the results are more precise and consistent.

Reliability vs. validity

Reliability means results are consistent when repeated. Validity means an experiment measures what it is intended to measure.

Low variability on a graph

Data points are tightly clustered, or error bars are short.

High variability on a graph

Data points are widely spread, or error bars are long.

Standard error of the mean (SEM)

An estimate of how precisely a sample mean represents the true population mean.

Why do researchers use SEM?

They use it to show uncertainty around a sample mean and to compare the precision of mean estimates.

SEM error bars

Error bars drawn above and below a mean to show the standard error of that mean.

Overlapping SEM error bars

Overlap alone does not prove that two means are not significantly different; a statistical test is needed to determine significance.

Statistical significance

A result unlikely to have occurred by random chance according to a chosen statistical test and significance level.