Presentation is loading. Please wait.

Presentation is loading. Please wait.

Chapter 1 Introduction to the Scientific Method Can Science Cure the Common Cold?

Similar presentations


Presentation on theme: "Chapter 1 Introduction to the Scientific Method Can Science Cure the Common Cold?"— Presentation transcript:

1 Chapter 1 Introduction to the Scientific Method Can Science Cure the Common Cold?

2 Science refers to a body of knowledge
Science is not a giant collection of facts to be memorized. It important to learn about the process of science called the scientific method. The scientific method allows the solving of problems and answer questions. Observations Proposing ideas Testing the ideas Discarding or modifying ideas based on results

3 The Nature of Hypotheses
1.1 The Process of Science The Nature of Hypotheses Hypothesis: proposed explanation for a set of observations Hypotheses needs to be: Testable – it must be possible to examine the hypothesis through observations Falsifiable – it must be able to potentially be proven false

4 Where do hypotheses come from?
Both logical and creative influences are used to develop a hypothesis

5 Scientific Theory Powerful, broad explanation of a large set of observations Based on well supported hypotheses Supported by research from several different independent sources As close to a “law” as you are going to get in Biology Examples: Cell Theory, Evolutionary Theory

6 The Logic of Hypothesis Tests
Inductive reasoning: combining a series of specific observations into a generalization to create a hypothesis Use specifics to generalize For example: Dogs have hair. Dogs are mammals. Cats have hair. Cats are mammals. Therefore, all mammals must have lots of hair. Is that necessarily true? Can give a starting point to researching more

7 The Logic of Hypothesis Tests
To test the hypothesis use deductive reasoning: This involves using a general principle to predict an expected observation using if/then statements For example, If vitamin C decreases the risk of catching a cold, then people who take in additional Vitamin C will get less colds.

8 The Logic of Hypothesis Tests
The process looks something like this:

9 If people who take vitamin C suffer fewer colds than those who do not. . . If people who take vitamin C suffer the same number of colds or more than those who do not. . . Why should scientists consider alternative hypotheses even if their hypothesis is supported by their research? Conclude that prediction is true Conclude that prediction is false Do not reject Reject the the hypothesis hypothesis Conduct additional tests Consider alternative hypotheses Figure 1.3 (continued)

10 The Logic of Hypothesis Tests
A hypothesis that fails our test is rejected and considered disproven. A hypothesis that passes is supported, but not proven. Why not? An alternative hypothesis might be the real explanation.

11 1.2 Hypothesis Testing The most powerful way to test hypotheses: do experiments!!

12 The Experimental Method
Experiments are designed to collect new and unknown data or information to test specific hypotheses. Variables: factors that can change in value under different conditions Independent variables can be manipulated by the scientist In ideal experimental conditions, you only want to manipulate 1 variable at a time – why? Dependent variables cannot be changed by the researcher

13 Controlled Experiments
Controlled experiment: tests the effect of a single variable in a conclusive fashion Control: a subject who is not exposed to the experimental treatment but has all other variables the same Use to make sure all the other parameters of the experimental system are behaving If many patients, including the control, die from a bad procedure, can you really say the drug would have killed them, or would have kept them alive?

14 Controlled Experiments
Differences seen between the experimental group and control group can be attributed to the experimental treatment. Random Assignment An effective way of assigning individuals to groups for testing Prevents “cherry picking” of subjects to go into one group or another by the investigators May flush out unexpected variables in an experiment

15 Controlled Experiments
Example: Echinacea tea experiment to relieve cold symptoms: Hypothesis: drinking Echinacea tea relieves cold symptoms Experimental group drinks Echinacea tea 5-6 times daily. (dried ground Echinacea tea leaves + water + cup) Control group drinks “sham” Echinacea tea 5-6 times daily (placebo). (dried ground non-Echinacea tea leaves + water + cup) Both groups rated the effectiveness of their treatment on relieving cold symptoms.

16 Controlled Experiments
People who received echinacea tea felt that it was 33% more effective at reducing symptoms.

17 Minimizing Bias in Experimental Design
If human subjects know whether they have received the real treatment or a placebo, they may be biased. Blind experiment: subjects don’t know what kind of treatment they have received Double blind experiment: both the person administering the treatments and the subjects do not know who is in each group until after the experiment is over Removes bias of data analysis by “motivated” investigator – scientists who know what group got a treatment may treat that data “better” than control data

18 Minimizing Bias in Experimental Design

19 1.2 Hypothesis Testing Using Correlation to Test Hypotheses
Sometimes direct experimentation is not possible Can you cut open a person every day to see if a tumor is shrinking from a treatment? It is not always possible or ethical to experiment on humans. Nazis experimented on people; ethics enforced because of this! Using existing data, is there a correlation between variables? Correlations are next best thing when experiments cannot be done

20 Summary of how we Test Hypotheses Using Experimentation
The “gold standard” for experimentation Double-blind, placebo controlled and randomized experiments Model systems can be used in experiments when it appears to dangerous or unethical to test on humans examples: mice, rats, dogs and pigs A correlation can be used to test hypotheses when controlled experiments on humans is impossible to perform

21 Using Correlation to Test Hypotheses
Using existing data, is there a correlation between variables? Hypothesis: stress makes people more susceptible to catching a cold Is there a correlation between stress and the number of colds people have caught?

22 Using Correlation to Test Hypotheses
Results of such a study: the number of colds increases as stress levels increase. Caution! Correlation does not imply causation.

23 Using Correlation to Test Hypotheses
The correlation might be due to other reasons.

24 1.3 Understanding Statistics
The Benefit of Statistics Do you believe what someone says, or do you believe what the number says??? “How much do you weigh?” Statistics is the use of mathematics to back up various claims about the observable world

25 1.3 Understanding Statistics
Overview: What Statistical Tests Can Tell Us We can extend the results from small samples to an entire population. Difference between two samples: real or due to chance?

26 Overview: What Statistical Tests Can Tell Us
Statistics in science is used to evaluate and compare data. We can extend the results from small samples to an entire population using statistical tests. Statistically significant: results of difference between groups is due to random chance and not an error in experimenting

27 The Null Hypothesis We use statistics to see if something is unusual or different from what we would expect. Philosophically, we must compare a result to a hypothesis that something is NOT unusual, that it CAN occur, no matter how strange it may look. This “normal” hypothesis is the null hypothesis. We can hypothesize that a result is different or unusual by REFUTING the null hypothesis with a degree of confidence (usually 95 or 99%). The exact wording of the null hypothesis is dependent on the situation.

28 Statistical Framing Let’s try a few basic experiments to explain the need to use statistics. Assumptions are shown in blue, and “results” are shown in red. Q: Guess how many jellybeans are in this jar? If you were to ask 400 people to guess, you would get 400 guesses. If the 400 people were guessing in the same way, (the null hypothesis is that no one guesses in a different way) the answers will aggregate around a particular value (the mean), which SHOULD come close to the real value that we can NEVER know by guessing. We could only know from counting (which is a different way to “guess”!). If we tallied each guess, and plotted the guess numbers by the number of people guessing the same number (and grouping the guesses by 5’s or 10’s, a process called binning), the shape of the resulting plot (called a histogram) SHOULD fit a bell curve. This bell curve represents a normal distribution. (don’t count!) how many guessed jellybean guess

29 Statistical Testing Q: Someone guessed 240 jellybeans. Is that a reasonable guess? Note that we need a “pool” of guesses to make this evaluation. Stated statistically, does the guess satisfy the null hypothesis (the null hypothesis is that the guess is not unusual)? It does not look like a reasonable guess; stated statistically, it may violate the null hypothesis. It can be demonstrated statistically that the guess is worse than over 99% of all guesses. If we define a reasonable guess as within 95% of guesses (this is related to p < 0.05 as a threshold for significance) then this guess is NOT reasonable. It is too far from “normal” guesses to satisfy the null hypothesis, and so it violates the null hypothesis. We are over 99% sure it is not a normal guess, and are justified (b/c > 95%) in wondering why the guess was so poor…. how many guessed jellybean guess

30 Statistical Significance Testing
Let’s try a few basic experiments to explain the need to statistics. Assumptions are shown in blue, and “results” are shown in red. Q: Are the number of jellybeans in both jars the same? There is ALWAYS a null hypothesis, even if it seems outlandish. What is the null hypothesis? Most people would guess there are more in the right-hand jar, implying that the null hypothesis is wrong. If they made guesses in both jars, and you made a histogram of both pools of guesses ON THE SAME GRAPH, it would be obvious that most of the values of the right-hand jar would fall well outside of the guesses of the left-hand jar. Statistics can indicate that the histograms overlap less than 5% (p < 0.05), thus one is more than 95% certain that the two values are different. how many guessed jellybean guess

31 The Problem of Sampling Error
Sampling error: the effect of variation in a measurement We can calculate the probability that a different result is simply due to normal variations in measurement. Statistically significant: an observed difference is probably not due to sampling error Opinion polling before a recent election indicated candidate A was favored by 47% of likely voters, and candidate B was favored by 51% of voters. There was a error of 3%. Why was this poll a statistical tie?

32 The Problem of Sampling Error
Confidence interval: the range of values from a sample that has a 95% probability of containing the true population mean (average). We are 95% sure that the value we obtain is close enough to the real value to be “identical” to it; in other words; it is close enough Much population variation = large confidence interval Small population variation = small confidence interval

33 The Problem of Sampling Error
The one circled on the left is better – see how the red lines don’t meet? You can say with confidence that there is a difference

34 What Statistical Tests Cannot Tell Us
If an experiment was designed and carried out properly If observer error occurred, only can evaluate the probability of sampling error May not be of any biological significance

35 1.4 Evaluating Scientific Information
Primary Sources Researchers can submit a paper about their results to a professional journal (primary source). Primary Sources undergo peer review: evaluation of submitted papers by other experts Primary Sources tend to be highly technical, and easily understood only by professionals Secondary sources: books, news reports, the internet, and advertisements

36 Information from Anecdotes
Anecdotal evidence is based on one person’s experience, not on experimental data. Example: a testimonial from a celebrity

37 Science in the News

38 Science in the News Needs to be explained so that non-scientists can understand it! Secondary sources may be missing critical information or may report the information incorrectly. (shortcuts to make it understood or make it a better story) Consider the source of media reports. Be careful with the internet since anyone can post information. (Check multiple sources, that are written differently!) Be very cautious about claims made in paid advertisements. (conflict of interest; financial motivation to make sure their product looks good!)

39 Understanding Science from Secondary Sources
Use your understanding of the process of science to evaluate science stories. News media generally highlight only those science stories that seem newsworthy.

40 Now let’s review with some questions:

41 A(n) ________ is a proposed explanation for a single observation.
scientific method hypothesis scientific theory experiment Answer: B A hypothesis is an explanation for one or a few observations; a theory explains many observations and rests on many tested hypotheses. An experiment is one way of testing a hypothesis, as part of the scientific method.

42 Which of the following is a scientific hypothesis?
Jazz is better music than rap. Garden fairies make tomatoes grow better. Hunting species to extinction is wrong. Increasing the amount of protein in a cow’s diet increases her milk yield. Answer: D Only answer D leads to a testable prediction. A is an aesthetic judgment, B is a supernatural explanation, and C is a moral judgment.

43 Which of the following is correct?
A hypothesis can be wrong. A hypothesis is not always testable. A hypothesis can prove a person’s values. A hypothesis should be formed before making any observations. Answer: A A hypothesis must be falsifiable, and sometimes it will in fact be wrong. It should always be testable. We can’t scientifically test moral values. Hypotheses should be suggested from our observations.

44 In an experiment, Dr. Smith feeds different amounts of protein to dairy cows and measures the differences in their milk yields. What is the independent variable? time milk yield amount of protein placebo Answer: C The researcher can alter the amount of protein and it is the independent variable; the milk yield is the dependent variable. Time isn’t being measured and there is no placebo described in this experimental design.

45 What does this graph show?
There is no difference between placebo and zinc lozenges. Taking placebo lozenges causes people to have more types of cold symptoms. Zinc lozenges lead to people catching fewer colds. People taking zinc lozenges show cold symptoms for a shorter period of time. Answer: D The Y axis measures how many days individuals have cold symptoms. The shorter bar for zinc lozenges indicates that people taking them have cold symptoms for a shorter period of time than with placebo lozenges.

46 The confidence interval is the range of values that has a _____ probability of containing the true population mean. 5% 30% 75% 95% Answer: D The actual population mean will have a 95% probability of being within this range.

47 A statistical test evaluates the chance of __________.
observer error. sampling error. alternative mechanisms. need for controls. Answer: B A statistical test measures the probability of chance errors in sampling. They cannot tell us about poor performance from the experimenter, possible other explanations, or the need for better experimental design.

48 Which of these statements about primary and secondary sources are correct?
Primary sources are written by researchers; secondary sources are written by book authors, news reporters, and advertisers. Primary sources aren’t biased; secondary sources are biased. Primary sources are not peer reviewed; secondary sources are peer reviewed. Primary sources are less reliable; secondary sources are more reliable. Answer: A Secondary sources are written by various types of authors based on the primary sources written by researchers. Sometimes, primary sources can be biased (researchers are human). Primary sources are peer reviewed, and therefore more reliable than secondary sources.

49 What is the best way to prevent the common cold?
Take vitamin C. Wash your hands. Take zinc lozenges. Get a cold vaccine. Answer: B The best way to prevent the common cold is frequent hand washing. Research shows that vitamin C intake has no effect on susceptibility and zinc lozenges only affect the duration of a cold. There is currently no cold vaccine.


Download ppt "Chapter 1 Introduction to the Scientific Method Can Science Cure the Common Cold?"

Similar presentations


Ads by Google