AP Biology The Nature of Science and Scientific Inquiry First things first… The role of discussion (Besides the obvious importance of being a capable thinker if you’re going to succeed in science) All AP questions ask you to interpret data or a model in light of what you know. Some AP questions do not ask you to “remember” anything at all; they give you a novel problem, and ask you to logic your way through it Full participation in discussions - no matter how difficult or how silly the questions look - is crucial to developing the intellectual capabilities this course (and the AP exam) demand. Discuss Let’s start with a straightforward one: Is there a distinction between “nature” or “the natural world” and “science?” If so, what? Discuss We distinguish science from math, from history, from poetry, etc. We call them separate fields of study, but what are their natures and what are their boundaries… What actually makes something “science?” What characteristics must something have in order for us to call it science? Discuss There are different methods we have for trying to understand the natural world (such as?) Science has been called the “most powerful” of those methods. What does “powerful” mean in this context? Once you feel satisfied with your answer to that, try to tackle putting into words what about science gives it that power. Scientific Disciplines Scientific disciplines are interrelated and interdependent, and occur in levels of “fundamental” (not cognitive) complexity Physics (fundamental forces of material world) Chemistry (nature and behavior of matter) Space and Planetary Sciences (non-living macroscopic phenomena) Biological Sciences (living things) Increasing Complexity Scientific Disciplines Biology Biology has many subdisciplines, and different authors ascribe it different themes Read Ch. 1 for a nicely illustrated traditional list of themes - particularly note “emergent properties,” it’s probably least familiar to you BUT the AP Biology curriculum is organized around just four themes… Big Ideas The Four “Big Ideas” Evolution Cellular Processes Genetics and Information Transfer Interactions Every topic that we study is connected to all four, and you need to get accustomed to noticing those connections as they come up Orientation to the AP Curriculum Document - “Enduring Understandings” Scientific Inquiry Major labs in AP Biology all feature inquiry Inquiry: “The diverse ways in which scientists study the natural world and propose explanations based on the evidence derived from their work. Scientific inquiry also refers to the activities through which students develop knowledge and understanding of scientific ideas, as well as an understanding of how scientists study the natural world.” Scientific Inquiry What this means for you: Creativity Collaboration Work Frustration Feelings of intimidation Feeling lost or directionless Independence …Improved scientific reasoning skills Scientific Inquiry What is the scientific method? Just kidding… sort of. There is NOT one scientific method. It’s an umbrella term for a variety of different methods that are scientific because they’re logical, naturalistic, and evidence-based (remember PLORNT). Types of Scientific Studies Controlled experiments Scientist-generated set up, the kind of experiment you’re more familiar with. Natural experiments Picking your independent, dependent, control variables, then going out and finding a situation that already occurred/already exists with those variables in place. Field studies Emphasis on inference from structured observation rather than establishment of variables and controls. Types of Scientific Studies Thought Experiments Evaluates a hypothesis by thinking through to its consequences. Einstein’s are famous. Mathematical Evaluation Using math theorems to work out underlying phenomena. Almost exclusive to physics. Can be considered a form of modeling. Modeling Using physical models, as in chemistry, or computer models, like weather models, to address questions. The models are generated based on real-world data, but the study you conduct doesn’t involve real-world data itself. Scientific Method The “scientific method” is flexible and creative as part of its power, but science is not a free-for-all either! A study must still be logical, evidence-based, carefully organized etc. regardless of its form. Traditional Scientific Method An observation is a description of information gathered with one of your five senses. It is important not to conflate observation with inference. Inference = ideas, assumptions, conclusions. Why is it important that observations be free of inference? Data Your observations can yield two types of data: Quantitative = data that can be measured. Numerical. (Ex.: number of objects, length, duration, mass, etc.) Qualitative = data that is non-numerical, observed but not measured. (Ex.: color, health, location, etc.) It’s possible to turn qualitative data into quantitative data and vice versa. For instance, ranking a reaction speed on a scale of 0-5 rather than “very slow, slow, medium…” Lab Notebook Write in pen Make a title page with your name, the course (“AP Biology Laboratory Notebook”), instructor name (“Instructor: Ms. Rebecca Stang”), the school & school year (2014-2015) (Either leave a page for Table of Contents, or plan to mark each lab with a sticky tab) On the next page, title it with the name of the lab at the top (Animal Behavior Lab, this time) As you proceed, always clearly title new sections. For instance, today, you will be taking “Observations.” Later, “Brainstorming Questions,” “Final Question,” “Hypotheses,” etc. Always DATE when each new bit of writing begins. You can do this in the margin or with the section title. NEVER tear out pages, NEVER erase or scribble anything out! If you need to make a change, draw a single line, so that the original text is still visible. For an optional professional touch, number each page, and date any cross-outs Observations In your lab notebooks, make detailed observations of these animals’ behaviors. You may feel free to manipulate them, place them in different environments, etc., but do NOT: Start running an off-the-cuff experiment Let them be harmed Detail! Avoid inference! Movement Animal movements can be kinesis or taxis. A kinesis is a simple change in activity or turning rate in response to a stimulus. It is non-directional. For instance, when humidity increases, wood lice spend less time stationary. But they don’t move towards or away from a human or moist area. Movement A taxis is a more or less automatic, oriented movement toward or away from a stimulus. Examples of taxis in animals include: Phototaxis = movement toward/away from light Phonotaxis = …sound Chemotaxis = …a chemical Anemotaxis = …wind Trophotaxis = …food Geotaxis = …earth or gravity Magnetotaxis = …a magnetic direction Klinotaxis = …a slope Rheotaxis = …water currents Discussion “British” birds “African” birds Blackcaps generally breed in SW Germany and winter in Africa, but some winter in Britain. Take both kinds of bird, put them in Germany, do a “peck test” to determine flight direction. What kind of movement is most likely being demonstrated here? Scientific Questions Not all questions are scientific, and not all scientific questions are conducive to a good study. A question must be: Centered on phenomena (objects, organisms, events) in the natural world Connects to scientific concepts rather than opinions, feelings, beliefs Possible to investigate through experiments and/or observations Leads to gathering evidence and using data to explain how the natural world works Scientific Questions Following these guidelines, meanwhile, isn’t necessary for the question to be defined as scientific, but will lead to a more productive study: It’s something you’re interested in finding out! You don’t already know the answer Shouldn’t be a “yes or no” answer Has a clear focus Is grounded in existing scientific understanding Is of a scope that matches the materials and setting available Can lead to further questions once all data is gathered Hypothesizing Hypothesis: A proposed explanation for a natural phenomenon. Hypotheses are small-scale models of nature based on prior knowledge. They are not guesses. They may be made before a test, or they may be the result of a test. There is an important difference (gets down to the philosophy of science) between hypotheses and predictions. Hypothesizing Two kinds of hypotheses: hypothesis: The general or “default” condition, the hypothesis that there is no relationship between the variables, that the treatment does not have any effect, etc. Alternate hypotheses: That there is a relationship, effect, etc. Null There may be multiple alternate hypotheses, but only one null Alternates may be “combined” Your “favorite” hypothesis may be either null or alternate, but be aware of both in order to be able to coherently explain your experiment Discussion So you have hypotheses, you have a strong experimental protocol that will collect well-structured data… but how will you know which hypothesis was probably right? How do you know whether or not the data support it? Predictions Prediction: A statement of the data that will result if a given hypothesis is correct, and this result would not be obtained if the hypothesis is incorrect. Predictions “coming true” are how you support or refute hypotheses. Predictions can be directly verified/observed, hypotheses (explanations) can’t. A well-written prediction will clearly set hypotheses apart from each other. Theory vs. Law You won’t be generating theories or laws, but you’ll be working with them. What’s the difference between a hypothesis, a theory, and a law? How are these terms different as used in science vs. as used in layman’s terms? Variables Review: Independent variable Dependent variable Control variable Control group Fair test A fair test of your hypothesis is one that avoids confounding variables - variables that damage the internal validity of your study. The easiest way to do this is often to ensure that there’s only one independent variable, but that’s not true of every study! A fair test also minimizes the chance of errors while maximizing the statistical significance of your results, while still being logistically feasible What can you do to improve the statistical significance of your data? You may wish to conduct a pilot study – a “pre-test” to confirm the protocol works Statistical Analyses What statistics CAN do: Quantify your results Clarify your results Provide an additional representation of your results Provide additional evidence What statistics CANNOT do: Evaluate or interpret your results Answer your question Statistical Analyses and Data Representation Basic operations: mean, median, mode, range, rate Use them whenever it’s appropriate, and don’t use them when it’s not Does it help illustrate your point? Is it not necessary to back up your point? If you conduct an operation and it REFUTES the point you were planning to make, not including it is dishonest, and a real scientist could get in big trouble for that! Discussion: Explain to your partner how to calculate/determine each of these five. Statistical Analyses A particular problem that statistics can help you to address is the significance of your results. How reliable is your sampling? (Standard Deviation & Standard Error) How certain can you be that your data swing that way because something drove it to? How do you know your results aren’t random? (Chi-Square Analysis) Heads up: Sigma aka “Funky E” This symbol: …means “the sum of” For instance, what is 3, 5, 6? Standard Deviation (calculation not on AP exam) Standard deviation is a measure of how diverse your values are. That’s not generally very helpful at the AP level, but you need it for the next calculation. The formula: Let’s say we measure 6 wingspans in centimeters: 2,2,2,5,8,12. Standard Deviation What does this mean? For AP Bio labs, not much… The greater your standard deviation (especially as compared to your mean), the greater your variation in data. The more standard deviations a figure is away from your mean, the more unusual it is compared to the rest of your data. Values within 4.12 of the mean (5.16) in our example are considered normal for this particular data set. Standard Error (calculation not on AP exam) Standard error indicates the average difference between the data mean you obtained from your limited number of trials, and the calculated data mean in the “real world.” “How certain am I that my sample is representative? If I’d done more trials, what could my mean turn out to be instead?” Standard Error (calculation not on AP exam) Simple equation: standard deviation divided by the square root of the sample size. (SE = s / √n) Standard deviation in our wingspan study was ___, and we sampled 6 birds. Standard error:. Our mean wingspan (5.16 cm) was within ___ cm of what we’d mathematically anticipate to be the real-world wingspan. Real-world mean wingspan is likely to be somewhere between __ cm and ___ cm. That’s a pretty large standard error; our mean varies from the expected by about 25%! Maybe we can’t necessarily be very confident that 5.16 is a representative mean wingspan… Notice that this equation shows you, mathematically, that a bigger sample size = less standard error! Reporting Standard error can be useful to report in some labs, but not always. Reporting standard error in writing, include the sample size, mean, and standard error: “The thing being studied (n=sample size) averaged mean +/- standard error.” “Wingspan length (n=6) averaged 5.16 +/- 1.68 cm.” Reporting (fair game on AP exam) Standard error, when calculated, should be represented in any relevant graphs using “confidence intervals (CI)” or “error bars.” How could you show SE +/- 1.0 on this graph? Mean Height (cm) Mean Hatchling Height by Diet 5 With sugar Without sugar Standard Error You’ll notice newspapers abuse this statistic CONSTANTLY. “Candidate A is crushing Candidate B in the polls with a lead of 4%! 52% of respondents plan to vote for Candidate A vs. only 48% voting for Candidate B (margin of error +/3%)” ……Sigh :/ Chi-Squared Test (IS tested!) The chi-squared ( ) test, or Pearson’s chi-squared test, evaluates the likelihood that variation in your results was due to chance. It can’t tell you whether the variation was because your independent variable caused it, but it can be used as evidence to rule out a null hypothesis. Warning: this is just the symbol for this test, it does not actually mean x squared! Chi-Squared Test Sigma, “the sum of” “Observed,” the data you actually collected “Expected,” the data point you would get if the null hypothesis is correct Chi-Squared Test “How do I know what to expect?” It varies… Examples: Suppose you want to know which of four bottles flies prefer. If the null is true, they have no preference. In which case, you would expect them to spend 25% of their time in each bottle. Chi-Squared Test “How do I know what to expect?” It varies… Examples: Suppose you want to know which of two flower colors, blue vs white, is more advantageous in an environment. There are 300 flowers. Chi-Squared Test “How do I know what to expect?” It varies… Examples: Suppose you want to know which of two forests a species of finch prefers. One forest is 800 acres, the other is 200 acres. A finch’s movements are tracked for 100 hours. Chi-Squared Test Let’s calculate chi-square! :D Our question: the heads side of a coin seems to have more mass to its image. As a consequence, is a coin actually weighted towards heads when you flip it? What is the null hypothesis in this instance? Chi-Squared Test . When I conduct this test, there will be two “outcomes” I can get: heads, or tails. If I flip a coin 100 times, and the null hypothesis is correct, what are my expected values for each of those two outcomes? Chi-Squared Test If I flip the coin 100 times, and the null hypothesis is correct, it should come up heads 50 times and tails 50 times. E Heads 50 Tails 50 O Chi-Squared Test I do the test, and it comes up heads 68 times and tails 32 times. E Head 50 s Tails 50 O 68 32 Chi-squared analysis can help me determine whether that variation from my expectations is due to chance or due to something actually causing heads to be more frequent. i.e., it assesses whether my null hypothesis holds any water. You must have at least two possible outcomes in your experiment (heads and tails, here) for the test to work. Chi-square DOESN’T WORK if you don’t have enough data points/trials or if their values are too small. An oft-cited magic number is 30, but run as many trials as possible, measure as precisely as possible, and let the mathematical chips fall where they may. No E can be 0. On to the calculation! You also need to know: Degrees of freedom: The number of outcomes minus 1 In our coin example, we have two outcomes being tested, heads and tails. That gives us one degree of freedom (2-1 = 1). P-value: Basically, how certain you can be of your result. The industry standard p-value is .05, and if your chisquare works it, that amounts to “I am 95% positive that this result is non-random.” A p-value of .01 amounts to “I am 99% positive that this result is nonrandom.” p-value of .001 is 99.9% certainty. Use .05 in AP Bio. Chi-Squared Test Now that you have your chi-square, degrees of freedom, and p-value, you’re nearly done. You just need a chart of critical values. Find your degree of freedom and your p-value in the row and column headers. Read down and across to find your cell with your critical value. If your chi-squared value is GREATER than the critical value, your null hypothesis is REJECTED. You’ve supported your results as non-random. If your chi-squared value is LESS than or EQUAL to the critical value, your null hypothesis is SUPPORTED. Variation is likely random. Chi-Squared Test .05 .01 .001 1 3.841 6.635 10.828 2 5.991 9.210 13.816 3 7.815 11.345 16.266 4 9.488 13.277 18.467 5 11.070 15.086 20.515 Chi-Squared Test Our coin test gave us a chi-square of ___. Does that support or reject the null hypothesis? Does this mean that the coin is definitely rigged or definitely fair?? Chi-Squared Test Try this problem: You’re testing to see if fruit flies prefer different fruits: apples, oranges, grapefruits. You set up a triangular choice chamber with an apple in one corner, an orange in another, and a grapefruit in the third. Null hypothesis: there is no preference. Actual data: Of 147 fly visits that landed on fruit for at least 20 seconds, 48 flies did so on the apple, 87 flies did so on the orange, and 12 flies did so on the grapefruit. Is this variation due to chance? Chi-Squared Test Remember, your values must be sufficiently large to run chi-square! Avoid this problem in experimental design by using variables with the opportunity for high precision. How about taking fruit fly data like this instead? You’re testing to see if fruit flies prefer different fruits: apples, oranges, grapefruits. Null hypothesis: there is no preference. Actual data: You release 30 flies into a container with three fruits and clock how much time they spend on fruit. Altogether, they spend a total of 418 seconds of their time on apples, 220 seconds of their time on oranges, and 187 seconds of their time on grapefruits. Is this variation due to chance? Chi-Squared Test Let’s do the same problem, with a twist! data: Apple – 418 sec, Orange – 220 seconds, Grapefruit – 187 seconds But what if, in the choice chamber, there were TWO apples in the apple corner, and still only one orange and one grapefruit in their corners? Same Chi-Squared Test If you reject the null hypothesis, your results can be reported as “significant” or “statistically significant.” When writing them up, you need to include all of the following: degrees of freedom, p-value (written as “less than” the p value), number of subjects (N), chi squared value. For instance, I would write of our coin test: Coin flips were found to be non-random in a chi-squared test, (X2 (2, N=100) = 12.96, p<.05). From this, we can conclude that coin flips were significantly weighted towards heads. The contents of those parens were: ((X2 (degrees of freedom, N=number of study subjects) = chi-square result, p<p-value). Statistics Again: statistics like these don’t answer your question for you. Even if I’m more than 95% confident based on this single statistical evaluation that the coin flips are nonrandom, but it doesn’t mean the coin was rigged! Maybe it was the way I flipped it, or air currents, or the table shape, or something else. Or, maybe I’m wrong! The stats are like another data point, another piece of evidence. You have to engage your brain and interpret your statistics, no differently than how you must interpret raw data. And a crummy study design can give you great-looking statistics (or terrible ones). A scientist would look at your mere 100 coin flips and not place a lot of trust in that chi-square analysis.