Scientific method

You make observations and predictions about the world every day, and you unconsciously test those predictions. When you do this, you are thinking scientifically.
Everyday example: at an airport, you notice some lines move faster than others. You predict that a line with lots of children will move more slowly than one with mostly business travelers.
You choose a line (business travelers) and the person next to you chooses the other line, then you watch to see if your prediction was correct. You reach the front first, so you conclude that children slow lines. Next time, you consult this prior knowledge. If your results differ, you consider that another factor might be at work.
In this informal way, you are using a version of the scientific method in daily life, just as scientists do in experiments.

Scientists use a similar logic when designing experiments. They follow a cycle of observation, question, hypothesis, prediction, testing, and interpretation.
Key steps often described:
- Make observations about the world.
- Ask a question based on those observations.
- Consult prior knowledge to inform the question and potential explanations.
- Form a hypothesis or tentative explanation.
- Write a prediction derived from that hypothesis.
- Test the hypothesis with an experiment.
- Analyze results and draw conclusions.
- Communicate the methods, results, and conclusions for peer review.
- Build prior knowledge that informs future work.

Researchers ask: How do brewing conditions affect the bitterness of coffee?
Prior knowledge about coffee taste identifies several affecting factors: water temperature, brewing time, and grind size.
Hypothesis (tentative explanation):
- If water temperature affects the rate of coffee extraction, then very hot water will make the coffee taste bitter.
Prediction: Based on the hypothesis, coffee brewed with very hot water will taste more bitter than coffee brewed with other temperatures.

Independent variable (manipulated): water temperature. Levels (three groups): very hot, normal, cool. The three levels are called treatments.
Control group: normal water temperature, representing the typical machine temperature.
Dependent variable (measured): bitterness of the coffee, scored by tasters on a scale from 0 to 5 (lower is less bitter, higher is more bitter).
Standardized (controlled) variables: brewing time and grind size are held constant across all treatments to avoid confounding effects.
Sample size: 100 volunteers taste the coffee.
Experimental setup: each taster experiences all three types of coffee (three types) and provides a bitterness score for each.
Data collection method: scores for each treatment are collected from all tasters, then summarized.

For each treatment, calculate the average bitterness score:
- Let $n$ be the number of tasters per treatment (here, 100).
- The average bitterness for a treatment is\bar{x} = \frac{1}{n} \sum{i=1}^{n} xi,where $x_i$ is the bitterness score given by taster $i$ for that treatment.
Example results reported:
- Cold water: average bitterness = \bar{x}_{cold} = 1.2
- Normal temperature water: average bitterness = \bar{x}_{normal} = 2.1
- Very hot water: average bitterness = \bar{x}_{hot} = 4.0\text{ (approximately, as reported)}

Graph: plot average bitterness on the y-axis and water temperature on the x-axis (independent variable on the horizontal axis).
Bars represent the average bitterness for each temperature; the top of each bar aligns with the corresponding average score.
Error bars: each bar has an error bar at the top, representing variation in the scores among the 100 tasters.
- A larger error bar means tasters disagreed more about the bitterness at that temperature.
- A smaller error bar means tasters agreed more on the bitterness for that temperature.
Interpretation of error bars: smaller error bars indicate greater confidence in the estimated average for that treatment.

The hypothesis predicts that water temperature affects extraction and can produce bitter coffee.
Results summary: coffee brewed with very hot water produced significantly more bitter coffee than the other treatments.
Statistical significance notion: the researchers use a statistical calculation to determine whether the observed differences are unlikely to be due to chance.
In this study, an asterisk or similar marker is placed above the bar for very hot water to denote statistical significance relative to other treatments. Normal-temperature coffee was not significantly more bitter than cold coffee.
Important nuance: while very hot water showed a clear significant effect, the comparison between normal and cold water did not show a significant difference.

The independent variable (water temperature) is manipulated, and the dependent variable (bitterness) is measured across replicates (tasters).
Comparison across treatments (cold, normal, hot) allows the investigators to determine the effect of temperature on bitterness.
They examine both the mean scores and the variability (error bars) to assess the reliability of the results.
By analyzing averages and variability, they assess whether observed differences are likely due to the manipulated variable rather than random variation.

Researchers write a manuscript detailing:
- Methods: how the experiment was conducted, including how temperature levels were implemented and how bitterness was measured.
- Results: the calculated averages, variability, and statistical outcomes.
- Conclusions: interpretation of what the results mean for the hypothesis.
Peer review: a process where other experts in the field evaluate the study to ensure quality, reliability, and rigor before publication.
Once published, the study becomes part of prior knowledge that informs later research and experiments by others.

Many people use the scientific method in daily life by standardizing variables, recording data, and drawing conclusions from experiences.
Understanding how to interpret graphs and data helps people become informed citizens, capable of evaluating arguments that rely on data.
The ability to interpret error bars and significance helps distinguish meaningful patterns from random variation.
Media literacy: since media sources often present data, knowing how to read graphs and understand methods allows for better critical evaluation of arguments.

Observations: careful noting of phenomena in the real world.
Question: a query arising from observations that can be tested.
Prior knowledge: existing information that informs hypotheses and predictions.
Hypothesis: a tentative, testable explanation or educated guess.
Prediction: a specific outcome expected if the hypothesis is true.
Independent variable: the variable that the experimenter deliberately changes or manipulates (e.g., water temperature).
Dependent variable: the variable measured to assess the effect of the independent variable (e.g., bitterness score).
Treatments: the different levels of the independent variable (e.g., very hot, normal, cool water).
Control group: a baseline condition used for comparison (e.g., normal water temperature).
Standardized (controlled) variables: other factors kept constant to prevent confounding effects (e.g., brewing time, grind size).
Sample size: number of experimental units or participants (e.g., 100 tasters).
Mean (average): the central value of a set