Science 480 Research Methods in Science

Study Guide :: Unit 1

The Nature of Science

Science is a unique way of knowing and describing our natural world, which includes ourselves. Science is actually a process, referred to in the literature as “the scientific method,” rather than an established body of knowledge (a common misunderstanding). While science does make predictions about outcomes of experiments or observations, the discipline is unique in that both correct and incorrect predictions contribute to the progress of science by increasing our knowledge.

A unique property of science is its self‑correcting nature. A scientific hypothesis, theory, or model will be accepted if it is sufficient to explain our observations and the results of our experiments. If, however, we observe a contradiction to our predictions, we may have to reject the hypothesis or modify the theory or model. The open‑endedness of scientific research could be (and is sometimes) seen as a weakness, but is in fact its greatest strength. Failed predictions (we prefer the term “hypotheses”) are a scientist’s daily bread because they allow us to exclude the proposed explanation and begin working on an alternative hypothesis.

This approach explains why science books must be rewritten so frequently. It is very rare that theories are rejected completely—in many cases they are only modified or amended. Newton’s laws can still explain many physical observations, even though Einstein’s theories of relativity have transcended Newtonian physics. Likewise, chromosomal and classical Mendelian genetics still claim a rightful place in biology although we know much more about DNA structure and function after the discoveries by Watson and Crick.

The scientific method can be compared to walking through a maze: frequently your experiments lead to a dead end; you then retrace your path and try another one that may or may not lead out of the maze, and so on. Another—imperfect—comparison is the procedures in a courtroom: rather than having a predetermined outcome of guilt or innocence about the case, a fair judge allows the parties to provide additional evidence to confirm or disprove previous allegations in court.

Hypotheses are testable statements that have the potential to be falsified. We will now walk you through an experiment that is an example of hypothesis‑driven research.

To test your hypothesis you would first create a stock solution of copper sulfate (CuSO₄) at the indicated concentration, and fill several containers (which you call replicates) with it. Into these you add a certain number of the aquatic insect you have chosen as your test organism. You then introduce so‑called controls, which are containers of plain water to use for comparison with the experimental containers of CuSO₄, and add the test organisms. You ensure that both experimental containers and controls are kept under exactly the same conditions (e.g., light, temperature). Several resulting scenarios are possible:

• the insects in copper sulfate die immediately, those in the control containers survive
• all insects survive
• all insects die
• some but not all of the insects survive both in the controls and in the CuSO₄ solution

All but the first of the possible scenarios falsify your hypothesis, indicating that that you would have to reject it. Let us assume you observe the first one—all insects in CuSO₄ die, those in water survive. Does this mean that you can accept the hypothesis and move on?

If you reread your hypothesis you will realize that you have taken only the first step: it refers to “all aquatic insects.” Copper sulfate killed the first insect species you tested, but now you have to choose a different insect. You may do this test for several other species and if you get the same results you will become more confident that your hypothesis can be accepted. Practically it would be impossible to test all aquatic insect species, so you may accept the hypothesis after testing 10 to 50 species. Alternatively you could re‑word your hypothesis to include only the tested species. Doing so should also show you how important clear and precise language is in science.

1.1: Types of research

In the previous section we explored the classical scientific approach (the scientific method), which is hypothesis driven. It is important to realize, however, that not all research is hypothesis driven. The textbook readings provide some good examples of the different approaches.

Learning Outcomes

After completing this section, you should be able to

• differentiate among the types of research: hypothesis driven research, curiosity driven research (called “experimental research” in the textbook), observational research, and applied research.

Reading

Study pages 1‑15 of the textbook.

Notes on Chapter 1

1. 1.1 Course goals: the SCIE 480 course outcomes are outlined in this Study Guide.
2. Ignore the “Assignments” at the end of the chapter.

Learning Activity

In the example above (aquatic insects in CuSO₄), select one of the other possible outcomes. How would you interpret these results and how would you change your experiment?

Solutions

Terms to Understand

hypothesis, falsified hypothesis (= rejected hypothesis)

A1‑I Graded Discussion (Part I of Assignment 1)

Define and discuss the difference between a hypothesis and a theory (in the scientific sense). (3% of the total course grade).

Step 1: Sign up for a Discussion Group

Join the discussion through the A1 - Part I: Group Sign Up page. Each group will have maximum of 12 student members. Only you, your group members, and your academic expert will be able to access your postings.

Step 2: Discuss

(minimum of three postings)

Using the web or other resources, define in your own words the difference between a hypothesis and a theory (in the scientific sense). A short paragraph is sufficient. Be aware that the word “theory” has a different meaning in common usage.

Sample student discussion: Erna posts her definition of the difference hypotheses and theories. Hongxin suggests an improvement. Elaine includes a few examples. Heather offers thoughtful questions, some of which Jingfen answers. Rafael points out that Erna’s definition of “hypothesis” does not apply for physics, and so on.

While every contribution counts, it is crucial to read what has already been posted (unless you are the first contributor). The idea is to get a real discussion going based on previous contributions. Academic experts may weigh in occasionally, or ask questions. A minimum of three contributions is expected from each student, but there is no upper limit. Students postings should demonstrate that they have studied the problem, used their own wording (no plagiarism!) and interacted with their fellow students.

Students are expected to use professional language in all postings. Personal attacks or degrading comments will result in a mark of 0. “I don’t agree with your opinion” is fine; “your definition is nuts” is unacceptable.

1.2: How to find literature resources from the library and other sources

Learning Outcomes

After completing this section, you should be able to

• find literature by using online resources in general and by accessing the numerous resources available at AU.

Here are some important considerations for your literature search that apply for any search engine or database.

• Google Scholar (https://scholar.google.ca/) is an excellent resource to start your search for any literature.
• Importance of search terms: be mindful of search terms and use various synonyms for your search. Search engines are not intuitive and you may miss a lot of information if you restrict yourself to just one search term.
• Be mindful of spelling variants: “behaviour” is spelled “behavior” in the U.S.
• You can also use asterisks “*” after the word stem, which allows the search engine to find variants, e.g., “environment*” will find all kinds of endings attached to “environment….”
• Avoid simple searches, rather use the “advanced” option, this will open a host of useful search options.
• Improve the efficiency of your search engine by using “AND,” “OR” and “NOT” (a Boolean search).
• Use quotation marks for crucial terms: If you look for articles on environmental impact assessment, type “environmental impact assessment.” Leaving the quotation mark out would give you thousands of irrelevant articles containing “assessment”, “environmental”, etc. without the term you are looking for.
• After you have exhausted Google Scholar, a regular google search often finds numerous other resources.

1.3: How to read scientific papers

If you want to know the latest developments in a particular field of research, you have no choice but to scan the scientific journals in it. Reading books and web‑based resources will familiarize you with a field, but cannot capture the latest research results (for very long). Scientific journals, which do publish recent discoveries, are not created equal, however. The differences in quality among them (and individual articles) are vast—in the ability of the authors to explain their research, in the clarity of the illustrations, and other (measurable) characteristics. A key quality indicator that is widely agreed upon is initial peer review, that is, whether prospective articles for the journal are sent out for critical review to researchers in the author’s field. Authors are then required to take reviewers’ comments into account when revising their papers before resubmission. This system, despite its flaws, ensures that research articles undergo a thorough quality control before being accepted. Journals without a mandatory system of peer review are uniformly considered inferior. You can find information about a particular journal’s peer review process on its website, under a heading such as “advice for authors.”

Reading scientific papers may not be easy for beginners, but once you become familiar with the jargon and learn a few basic techniques, you will find it easier with each one you read. Start with articles in popular scientific journals, such as New Scientist or Scientific American, which are generally based on a more technical paper from an academic journal. After reading the popular version, you can then move on to the original, which was written for a more scholarly audience.

Learning Outcomes

After completing this section, you should be able to develop a habit of reading scientific papers in your field of interest, from popular scientific journals to more technical specialist journals.

Reading

Study pages 173‑178 of the textbook.

Learning Activities

Scan some of the recent back issues of Scientific American from AU Library using the skills you just learned. Select two articles in your field of interest and read them. Then, look in the references for the original sources, search for these articles in the library, and read them.

1.4: Research hypothesis and null hypothesis, Type I and Type II errors

Researchers spend a considerable amount of planning for their research: formulating a research hypothesis, finding feasible ways to test their hypothesis, and determining the correct statistical analysis for the results obtained. This process, which often involves interdisciplinary teams of researchers, is referred to as research design.

The following sections provide an overview of the main elements of research design. It is always the null hypothesis which is tested. It is generally the status quo, the prevailing opinion, or the ‘no change’ hypothesis. The conclusion of a test of hypotheses is always

EITHER

Do not reject the null hypothesis

(meaning that there is not enough evidence to indicate, at a ___ level of significance, that the null hypothesis is false)

OR

Reject the null hypothesis

(meaning that there is sufficient evidence to indicate, at a ___ level of significance, that the null hypothesis is false, which, in turn, implies that there is sufficient evidence to declare the alternative hypothesis may, in all probability, be true.)

The outcomes of experiments are not always clear‑cut and usually require a statistical treatment. Occasionally a researcher may erroneously accept the research hypothesis, although the treatment has no effect. This error would be called a false positive or Type I error.

On the other hand, the researcher may have an outcome with a false negative if he or she sees no effect of the treatment, but in reality an effect has occurred and been overlooked. This situation can be referred to as a Type II error.

Typically, both Type I and Type II errors may result from mistakes in the statistical analyses and/or poor research design. However, even when applying the correct statistical test, a tiny percentage of Type I and Type II errors may occur.

Learning Outcomes

After completing this section, you should be able to

• understand the difference between a null hypothesis and a research hypothesis (= alternative hypothesis).
• describe the difference between Type I and Type II errors.

Terms to Understand

Independent variable, dependent variable (= control variable), treatment.

The term treatment is often used to describe manipulative actions by the researcher (e.g., the addition of fertilizer to a population of plants is a treatment). Common scientific jargon also uses the term for the product, so in our example, the group of plants that received the fertilizer is also called “treatment.” Treatments are only valid if a control is part of the experiment in which the treatment conditions are absent (absence of fertilizer).

Reading

Study pages 18‑30 of the textbook.

Notes on Chapter 2: The three examples used in the textbook are very instructive. Reading through them carefully will help you understand basic principles of research design. In the final paragraph on p. 18 the author refers to cause and effect for the independent and the dependent variable. This is correct but sometimes the two parameters under investigation are correlated without one causing (influencing) the other. The two would be called correlated, in which case the terms dependent and independent variable should be avoided.

Exercise

In the example above, adding nutrients to an aquarium, what if you believe the treatment has no effect? Would you have to formulate a different hypothesis? If yes, would this be legitimate and would it make a difference for the research project?

Solution

1.5: Random and systematic errors, accuracy and precision, bias

Random and systematic errors may occur when taking measurements during an experiment.

Random errors result from the limitation of any research instrument. They are easy to correct by taking the mean (averaging) of a whole series of repeated measurements.

One of the basic principles of research design and practice is that a sufficient number of measurements must be taken.

Systematic errors occur when a research instrument is faulty and does not allow accurate measurements. This can be corrected by calibrating the research instrument, but, unlike for random errors, the researcher may not be aware of systematic errors.

Note the fundamental difference between random/systematic and Type I/ Type II errors. The former may occur at the experimental stage whereas the latter may arise during the final analysis of a research project when a decision is made whether or not the research hypothesis is accepted.

Accuracy and precision are important concepts for research practitioners to understand: Accuracy indicates how close a measured value is to the actual value (e.g., an accurate ruler would measure the length of a 20.0 cm piece of paper as 20.0 cm, not 19.9 cm or 20.1 cm).

Precision, on the other hand, refers to how detailed (how many decimals) an instrument can measure values. Precision is also indicated by how close a repeated series of measurement values are to each other. Quality scientific instruments always specify their level of precision (e.g., to the third decimal). Frequently, they show output with more decimals, in which case you would round the result indicated down to the third decimal.

An accurate instrument may not be very precise and a precise instrument, if not calibrated well, may not be accurate. If you wish, you can find a number of videos explaining the difference between the two words on YouTube (e.g., https://www.youtube.com/watch?v=8Cl5CeiT7hU).

Finally, research must be free of bias. A researcher’s preconceptions about possible outcomes of a research project should not influence its research design in any way that would make a particular outcome more likely.

Learning Outcomes

After completing this section, you should be able to

• differentiate between random and systematic errors as well as between accuracy and precision.

Terms to Understand

Random errors, systematic errors, accuracy, precision

Reading

Study pages 30‑40 of the textbook.

In addition, please read https://norvig.com/experiment-design.html before beginning Assignment 1‑II.

Exercise

Would you match accuracy with random errors or with systematic errors?

Solution

1.6: Research involving humans and animals

The textbook describes basic aspects of research on humans and animals from a U.S. perspective. For Canadian viewpoints, see the Canadian Council on Animal Care in science (CCAC) Guidelines on the ethical use and care of animals and the Tri‑council Policy Statement on Ethical Conduct for Research Involving Humans.

Learning Outcomes

After completing this section, you should be able to

• follow ethical principles if your research involves human and animals.

Reading

Study pages 43‑48 of the textbook.

Assignment 1

Complete and submit Assignment 1.