什么是实验？

An experiment is a systematic, planned activity carried out under controlled conditions to prove or disprove a hypothesis and uncover new knowledge. Researchers investigate the relationship between two variables by manipulating one variable and measuring changes in another. This approach allows scientists to establish cause-and-effect connections and draw statistically significant conclusions about phenomena.

The independent variable, also called the explanatory variable, is controlled and manipulated by the researcher. The dependent variable, or response variable, is measured and recorded to observe how it changes in response to the independent variable. In a mobile phone study, phone usage is independent while sleep hours and sleep latency are dependent variables that researchers measure.

Control groups serve as the standard of comparison, allowing researchers to isolate the effect of the independent variable. By comparing an experimental group receiving a treatment to a control group without it, researchers can determine whether observed changes result from the treatment or other factors. This comparison is fundamental to establishing cause-and-effect relationships through controls in experiments.

Lurking variables are additional factors beyond the explanatory variable that can cloud or confound study results. For example, in a vitamin E study, exercise, diet, and smoking habits could influence health outcomes independently of vitamin E consumption. Random assignment of subjects to treatment groups spreads lurking variables equally among groups, ensuring only the researcher's planned treatment differs between groups.

Random assignment distributes potential lurking variables equally across all treatment groups, ensuring the only systematic difference between groups is the researcher-imposed treatment. This isolation allows researchers to confidently attribute observed differences in the response variable to the explanatory variable alone, establishing valid cause-and-effect connections supported by randomized experiments.

Statistics is used to interpret experimental results and determine whether observed changes are statistically significant or due to chance. Hypothesis testing analyzes data from experimental and control groups to assess whether differences in the response variable are meaningful. This statistical analysis transforms raw measurements into evidence supporting or refuting the original hypothesis.

Observational studies compare groups based on existing characteristics without researcher manipulation or random assignment, leaving lurking variables uncontrolled. While observational studies can reveal associations, they cannot isolate the explanatory variable's effect. Only randomized experiments with controlled treatments and random assignment can definitively prove that one variable causes changes in another.