ACT Science 2: Research Summaries and Experimental Design

Experiment

A systematic, repeatable procedure used to test a hypothesis, investigate a phenomenon, or determine the effect of changing a factor. Experiments use observations and logical analysis to draw conclusions about cause and effect.

What is the role of an experiment in the scientific method?

An experiment provides empirical evidence that can support, refine, or contradict a hypothesis or model. It helps distinguish among competing explanations by observing how a system responds under specified conditions.

Hypothesis

A testable proposed explanation or prediction about how a process or phenomenon works. A useful hypothesis identifies measurable outcomes that could support or contradict it.

Why can an experiment support a hypothesis but not conclusively prove it?

Experimental results are obtained under limited conditions and may be affected by measurement uncertainty or unrecognized variables. Repeated evidence can increase confidence in a hypothesis, while a reliable counterexample can contradict it.

Independent variable

The factor deliberately changed or manipulated by the experimenter. In a chemistry investigation, it might be reactant concentration, temperature, pressure, or the identity of a substance.

Dependent variable

The measured response that changes as a result of the independent variable. Examples include reaction rate, mass of product, gas volume, absorbance, or temperature change.

Controlled variable

A factor that is kept constant or accounted for so that it does not confound the relationship between the independent and dependent variables. Examples include sample mass, total volume, reaction time, or apparatus.

Confounding variable

A factor other than the independent variable that changes in a way that can affect the measured result. Confounding variables reduce the ability to attribute an observed effect to the factor being tested.

Control group or control sample

A comparison condition that is treated like the experimental condition except for the factor being investigated. It establishes a baseline against which the effect of the independent variable can be evaluated.

Why are controls important in a chemistry experiment?

Controls help identify changes caused by the independent variable rather than by the solvent, reagents, environment, or measurement procedure. They improve the reliability and interpretability of the results.

Positive control

A control known from prior evidence to produce the expected positive response. It verifies that the reagents, apparatus, and basic procedure are capable of detecting the phenomenon.

Negative control

A control expected to produce no measurable positive response. Its result estimates the background signal or response that should be distinguished from the experimental effect.

Replication

Performing the same measurement or treatment on multiple independent samples or trials. Replication helps reveal random variation, improves estimates of the average response, and identifies anomalous results.

Why should an anomalous replicate not automatically be discarded?

An outlier may reflect experimental error, but it may also reveal real variation or an unexpected phenomenon. It should be investigated using the procedure, observations, and measurement quality before deciding whether exclusion is justified.

What is the purpose of averaging replicate measurements?

Averaging can reduce the influence of random error and provide a more precise estimate of the measured quantity. It does not eliminate systematic error or compensate for a flawed procedure.

Random assignment

Assigning experimental units to treatment or control conditions by a random process. It tends to distribute uncontrolled characteristics among groups, reducing systematic differences and selection bias.

Why is random assignment less central in many physical chemistry experiments than in human studies?

Identical chemical samples can often be prepared and divided into equivalent portions, whereas human or biological subjects naturally differ in many characteristics. Random assignment is especially useful when individual differences could confound treatment comparisons.

Double-blind experiment

An experiment in which neither the subjects nor the researchers interacting with them know which subjects receive the treatment or control. This reduces effects caused by expectations or observer bias.

What is the difference between a controlled experiment and an observational study?

A controlled experiment deliberately manipulates an independent variable and measures the response under specified conditions. An observational study records variables without deliberately imposing the treatment, so it generally provides weaker evidence for causation.

Natural experiment

An investigation that uses naturally occurring variation in conditions rather than deliberately manipulating the independent variable. It may be useful when manipulation is impossible, impractical, unethical, or illegal, but uncontrolled confounding can limit causal conclusions.

Field experiment

An experiment conducted in a natural, real-world setting rather than a highly controlled laboratory. Field experiments may have greater realism or external validity, but they generally provide less control over contamination and confounding variables.

Laboratory experiment

An experiment performed under carefully controlled and often artificial conditions. Laboratory settings allow precise control of variables and replication, although results may not represent behavior in more complex natural environments.

External validity

The extent to which conclusions from an experiment apply to other settings, populations, or conditions. Natural and field experiments may have greater external validity, while laboratory experiments often provide stronger control.

Selection bias

A systematic difference between groups caused by how subjects or samples enter or are assigned to conditions. It can make an apparent treatment effect actually reflect preexisting differences between the groups.

Why do observational correlations not necessarily establish causation?

A correlation may result from a confounding variable, reverse causation, or coincidence rather than a direct causal relationship. Manipulation, control, and appropriate comparison groups provide stronger evidence for causation.

What does it mean for experimental groups to be probabilistically equivalent?

Before treatment, the groups are expected to have similar distributions of relevant characteristics, even though individual measurements may differ. Random assignment and sufficiently large sample sizes help produce this approximate equivalence.

Null hypothesis

The claim that the tested treatment or independent variable produces no meaningful effect or difference. Experimental data are evaluated to determine whether the null hypothesis is inconsistent with the observations.

How should a scientist interpret results that do not match the prediction?

The results may contradict the hypothesis, reveal an uncontrolled variable or procedural error, or show that the hypothesis needs refinement. The experiment should be examined and, when appropriate, repeated before drawing a final conclusion.

What is the difference between random error and systematic error in an experiment?

Random error causes unpredictable variation among measurements and can often be reduced through replication and averaging. Systematic error shifts measurements consistently in one direction and is not removed merely by taking more trials.

Why is repeatability important in experimental science?

A repeatable procedure allows others to check whether a result is reliable rather than accidental. Reproducible results strengthen confidence that the observed pattern reflects the investigated variables.

What is the purpose of measuring several different treatments in an experiment?

Different treatment levels allow the investigator to determine how the response depends on the independent variable rather than merely comparing one treatment with one control. For example, varying concentration can reveal a trend in reaction rate.

How does controlling variables improve a cause-and-effect conclusion?

If all relevant conditions are held constant except the independent variable, differences in the dependent variable are more plausibly attributed to that variable. Poor control leaves alternative explanations for the result.

Why might a natural experiment be less reliable than a controlled experiment?

The researcher cannot fully control or measure all variables in a natural setting. Unmeasured factors may vary along with the variable of interest and create an illusory correlation.

What is the main trade-off between laboratory and field experiments?

Laboratory experiments generally maximize control, precision, and ease of replication. Field experiments generally provide more realistic conditions but make it harder to isolate the effect of one variable.

How do experiments differ from informal everyday comparisons?

A scientific experiment uses a planned, repeatable procedure, identifies variables, controls relevant conditions, and analyzes results systematically. An informal comparison may generate an idea but often lacks sufficient control and measurement to support a strong causal conclusion.

What is a treatment in an experiment?

A treatment is a specific condition or level of the independent variable applied to an experimental unit. Experiments may compare two or more treatments to estimate how they affect the measured response.

What is an experimental unit?

The smallest independent subject, sample, or group to which a treatment is assigned in an experiment. Correctly identifying experimental units is important for applying replication and comparing treatment effects.