Experimental Design and Hypothesizing - Study Notes

Hypothesis and the Scientific Mindset

• A hypothesis is your tentative answer to a research question and makes predictions that could be tested. It is not just an educated guess when described as a single statement like “prettier” or “smells better” (these are not testable).

• Every hypothesis has two essential parameters:

  • It must be testable.

  • It must have two possible outcomes (i.e., be falsifiable).

• The scientific process is not just a linear set of steps; it is a circular, iterative way of thinking:

  • Make observations

  • Generate questions

  • Make observations

  • Formulate new questions

  • Develop hypotheses

  • Collect data

  • Compare with predictions

  • Revise or reformulate hypotheses if data don’t match.

• Hypotheses must be testable and falsifiable. The idea of falsifiability is best understood in the context of null and alternative hypotheses.

  • Null hypothesis ($H_0$) represents the assumption that observed differences are due to random differences (random variation).

  • Alternative (experimental) hypothesis ($H<em>a$ or $H</em>1$) represents the opposite claim to be tested against $H_0$.

• In science, conclusions are never final proofs; conclusions are based on current evidence and can be revised with new data or technology.

• Example of thinking about data in terms of normal expectations:

  • Bell curve concept: If test scores in a class follow a typical distribution, most scores cluster around the mean with fewer extreme high/low scores. If the distribution skews dramatically (e.g., many high A’s and few B’s/C’s/D’s/F’s), that suggests results beyond random variation.

  • In lab settings, scientists typically write paired hypotheses: $H<em>0$ and $H</em>a$. In classroom labs, you may not always be asked to formalize $H_0$, but you should understand what each term means.

Null and Alternative Hypotheses

• Null hypothesis ($H_0$): No statistical difference between the test groups.

• Alternative hypothesis ($H<em>a$): The difference is in the direction or exists (opposite of $H</em>0$).

• Example 1 (shoe sizes):

  • Let $\mu_a$ = average shoe size for males aged 10–15

  • Let $\mu_b$ = average shoe size for females aged 10–15

  • $H<em>0: \mu</em>a = \mu_b$

  • Ha: \mua > \mu_b

  • Possible outcomes: either \mua > \mub or $\mu<em>a \le \mu</em>b$. The chosen $H_a$ determines the statistical test direction.

  • The writing of the hypotheses (e.g., “greater than” vs “not greater than”) determines the type of statistical analysis used.

• Example 2 (chickens and food):

  • Null: All chickens eat the same average amount of food. $H<em>0: \bar{F}</em>{\text{purple}} = \bar{F}_{\text{brown}}$

  • Alternative: Purple chickens eat more on average than brown chickens. Ha: \bar{F}{\text{purple}} > \bar{F}_{\text{brown}}

  • If the experiment shows a difference, the null is rejected in favor of the alternative.

• Important points:

  • If data show a difference, you reject the null hypothesis; you focus on the alternative hypothesis.

  • Nothing in science is ever proven; conclusions are supported by current evidence and can be revised.

  • Practice prompt (to reinforce): write some pairs of hypotheses based on observations (e.g., purple vs yellow flowers and bee visitation).

Relationship: Theories vs. Laws

• Theories and laws have two separate purposes and are not hierarchical such that one becomes the other:

  • Theories explain phenomena or mechanisms.

  • Laws describe observed relationships or regularities in nature, often with mathematical form.

• They are both well-supported by evidence but serve different roles; neither is more “valid” than the other, and one does not turn into the other.

• Examples mentioned:

  • Darwin’s theory of evolution by natural selection explains how fitness influences survival and evolution across populations.

  • Boyle’s law describes the relationship between pressure and volume of a gas (as pressure increases, volume decreases; there is a mathematical relationship that can be used to calculate behavior under given parameters).

• Clarifications:

  • Science cannot address supernatural phenomena.

  • Hypotheses must be testable and falsifiable; results must be repeatable.

• Case study: the vaccination-autism controversy

  • A real-world example where data were falsified by a researcher, leading to incorrect conclusions and public health impacts.

  • There is no evidence supporting a link between vaccinations and autism; data falsification undermines scientific conclusions.

• Human interpretation issues:

  • Observations and conclusions can be misinterpreted due to cognitive biases or incomplete information.

  • Examples include misinterpretations of limited data from media or sensational reports.

• Stomach ulcers: historical ideas vs. new evidence

  • Earlier beliefs blamed stress or spicy foods; later evidence implicated a bacterium as a cause, and antibiotics could cure ulcers.

  • Acceptance of new explanations can be slow when they contradict established beliefs.

Limitations of Science and the Role of Technology

• Science cannot address supernatural explanations.

• Observations and conclusions are limited to what can be observed and tested in the natural world.

• Technology as application of science:

  • Vaccinations, germ theory, pasteurization, food safety, medications, etc., show how biological advances translate into practical tools.

  • Edward Jenner’s smallpox vaccine is an example of early vaccine development, sometimes before a formal germ theory was fully established.

  • The development of microscopes enabled the discovery of bacteria and the identification of pathogens.

• Models in science:

  • Models help test and predict phenomena when real-world testing is impractical.

  • Example: a model of blood flow in the heart helps predict heart function.

Experimental Design: Variables and Data

• Independent variable (IV): also called the manipulated or experimental variable; what you change deliberately.

• Dependent variable (DV): also called the responding variable; what you measure in response to the IV.

• Controls: all other factors kept constant to isolate the effect of the IV.

• Key rules for experiments:

  • Change only one variable at a time in a given hypothesis/test.

  • A single hypothesis typically addresses one IV and one DV.

  • You can construct multiple hypotheses to test different parts or aspects of an experiment, but each should be tested one at a time.

• Practical example:

  • If you change the amount of sunlight (IV), you should not also change the amount of water or temperature in the same test, as those would confound results.

  • Observe how plant growth (DV) responds to sunlight (IV).

• Recap of core concepts:

  • A hypothesis must be testable and have two possible outcomes.

  • Theories explain; laws describe.

  • Independent/Dependent variables and controlled variables define the structure of an experiment.

  • Only a single variable is tested at a time in a well-designed experiment.

  • Observations and conclusions are limited to natural phenomena and repeatable evidence.

Putting It All Together: How to Apply These Concepts

• Be able to formulate a scientific hypothesis that is testable and falsifiable.

• Identify the independent and dependent variables in any given investigation.

• Distinguish between a theory (explanation) and a law (descriptive relationship).

• Recognize the role of technology and models in advancing scientific inquiry.

• Practice writing paired hypotheses ($H<em>0$ and $H</em>a$) for simple, controllable scenarios and interpreting potential outcomes.

• Remember that science advances by revising conclusions in light of new evidence, not by proving a hypothesis definitively.

Quick Practice Prompts (you can try now)

• Prompt 1: Purple vs yellow flowers; bees attraction observed. Write a null and alternative hypothesis about bee visitation rates.

• Prompt 2: A simple plant growth experiment varying only sunlight. Define one IV, one DV, and two or more control variables.

• Prompt 3: State a theoretical scenario where you would need to distinguish between a theory and a law, and describe what each would explain or describe.

• Prompt 4: Consider a historical medical claim. Explain how you would design a study to test whether the claim is supported, including how you would handle potential misinterpretation or data falsification.

Summary Takeaways

• Hypotheses are testable, falsifiable statements about relationships between variables.

• Null hypotheses posit no difference; alternative hypotheses posit a difference or a directional effect.

• Theories explain phenomena; laws describe relationships; both are supported by extensive evidence and are not hierarchical.

• Science has limits and is complemented by technology and models to extend inquiry.

• Clear identification of IV, DV, and controls is essential for valid experimental design.

• Scientific conclusions are provisional and subject to revision with new evidence or improved methods.