Experimental Question – Your turn to come up with an experiment! The experimental method is key to the process of science, and we want you to hone your skills in asking questions and attempting to answer them. This assignment is designed to get you to design experiments that you could actually conduct (but you won't actually conduct them), and report them as a scientist would. Remember that the goal of a simple experiment is to test a hypothesis by setting up conditions in which the hypothesis makes specific predictions, then determining whether or not the predictions made by the hypothesis work. At its simplest, an experiment should manipulate a single factor (for example, temperature) and then measure the effect of changes in that variable on a second factor (for example, the rate at which a dog pants). More complicated experiments are certainly possible, but often require fairly sophisticated statistical analyses to interpret. Other variables that you think might affect the second factor should be kept constant in all of your treatments. You will turn in a genuine scientific questions of the sort that you could investigate experimentally. This is important. They must be experimental, and they must be genuinely yours, not borrowed from other individuals, former students, the internet, or anywhere else. They should also be testable by you. You won’t actually be conducting the experiments, but they should be of the sort that you could do if you had the time. Observations are the key to developing good questions, and are likely to come to you at any time (e.g. walking to the parking lot, driving to and from campus, working at home, etc.). Therefore, be especially attentive to your observations throughout the quarter. The Write up: Include underlined words in your paper as section separators. Introduction – The main idea of the introduction is to let the reader know the question you are asking, your hypothesis and predictions. In your introduction you must: Identify the question you are asking. Explain why you are asking the question – this should include observations you’ve made, past experience etc. that led you to ask the question Explicitly state your hypothesis: Remember that a hypothesis is a potential explanation of why something happens or is the way it is, and should lead logically to a prediction. Hypotheses may be right or wrong, but they must lead to testable predictions. Explicitly state your prediction(s): Your hypothesis can be used to develop a prediction about what might happen given a particular situation. Be very clear about both the situation and what you expect to happen if your hypothesis is correct. Be sure that this prediction follows logically from your hypothesis and is NOT just repeating your initial observation. If you predict something you’ve already observed you are using circular reasoning – you cannot falsify any hypothesis this way. You must apply your hypothesis to a new situation, and use your hypothesis to predict what should occur in the new situation. Although they are often equated by non-scientists, hypotheses and predictions are NOT the same thing (and should never be taught as such). If you find that your hypothesis and prediction mean the same thing, it is time to revise – make sure your hypothesis is a potential explanation. Methods – the main idea in the methods section is that someone who reads it should be able to duplicate your experiment precisely. If they can’t, it is not a goods methods section. In your methods you must: Provide a brief description of each treatment, and how you would collect data. Be precise, and always include metric units of measurement. Avoid vague statements like "there will be 2-3 groups with varying salt concentrations" -- decide on a number of groups, and state which concentrations you would use! (“there will be 2 groups, one with a 0% salt concentration and one with a 5% salt concentration…”), Also, make your life easier by limiting your experiments to simple ones in which there are only 2 important variables: An independent variable and a dependent variable. Experiments with multiple independent and/or dependent variables are done all the time, but are much more complicated. Explicitly identify both the independent variable (what you will manipulate) and the dependent variable (what you will measure). Include descriptions of how you will make your measurements and manipulations. Be sure to include the metric units you will use to measure variables that need measurements. Explicitly identify the sample size you would ideally use, in terms of number of samples per treatment. If you use less than 20 samples per group, be sure to explain why you are using a small number; if there isn’t a good reason that large numbers can’t be collected, use larger numbers. Explicitly identify factors that should be held constant. Identify any factors that might affect the outcome of your experiment, and so should not vary between treatments. Explicitly identify which of your treatments (if any) represents a control group. Remember that there are two basic kinds of control groups, plus experiments with no control group: 1) “No treatment” controls: The control group does not receive the independent variable. For example in medical experiments, the experimental group receives the medicine being tested, and the control group doesn’t. Instead, they get a placebo. 2) “Natural” controls: The control group receives a natural amount of the independent variable. For example, in an experiment to determine the effect of water on plants, giving the plants no water is trivial, so your control might match the amount of water to the amount of rainfall over the length of the experiment – i.e. what the plant would have received under natural conditions. 3) No control: some experiments do not have a control because it is impossible. For example, in Galileo’s ball rolling experiment, there would be no such thing as a ball with “no mass” (remember that mass was his independent variable) or a “natural mass” for his balls. If your experiment has no control, be sure to state this explicitly and explain why a control is not possible. Expected results – Usually you would just have a results section, where you’d put the results of your experiment. We won’t actually be doing the experiments, so we won’t have real results. Instead we will have expected results: Provide a graph showing what you predict your results will look like assuming that your hypothesis was supported by the “data”. Be sure to include axis labels, units, a caption etc. as outlined in graphing labs. Be sure to use the type of graph that is appropriate for your experiment (i.e. comparison of means or regression) State in separate text (not in the caption) whether the graph demonstrates statistically significant results or not – remember that your conclusion should match the prediction you made if your hypothesis is correct. For regression questions, be sure to include regression equation and the p-value. DO NOT include your regression analysis or output tables – just the equation and p-value. You will turn in one experimental question during the quarter, but you will have two submissions (see lab schedule for due dates). The 1st submission will be worth 20 points. The 2nd submission is an opportunity for you to fix your errors and learn from your mistakes. It is worth 30 points, so pay attention to your instructor’s comments. You must turn back your original first submission with your second submission or you will receive an automatic 10% deduction on your 2nd submission. Your question may be either a comparison of means or a regression. Important details: 1. This assignment allows only experimental research questions, not observational. An experiment involves manipulating some variable (independent variable) and measuring its effect on a response (dependent variable). A demonstration of a principle is NOT an experiment. Making observations or measurements without manipulating a variable is NOT an experiment (though the questions may certainly be interesting). If you fail to conduct an experiment you will receive a 20% deduction on top of any other problems in your paper. Focus your questions on cause and effect relationships: how do changes in one variable effect a 2nd variable? 2. All questions must be something that YOU could really investigate – if you had the time and equipment. 3. The metric system must be used for all measurements – no English units! 4. Many types of questions are NOT appropriate for this assignment. Below are examples of what not to include. Do not use questions involving “how does something work?” For example, “How does a microwave work?” would not be appropriate for this assignment because it is not an experimental question that you could test. Do not use questions to which most adults already know the answer (e.g. Will it take longer for mold to appear on bread if it is kept in the freezer or on the shelf?). The answers to these are common knowledge; therefore they would not be appropriate for this assignment. You will not lose points for an incorrect hypothesis, but if you ask a trivial question you will receive a 20% deduction, on top of any other deductions from problems in your experiment. The following are specific examples of such trivial questions, but not an exhaustive list. No questions allowed that relate to mold or food spoilage. No questions about the time to boil or freeze water with different things added. No questions regarding the effect of plant-growth substances (e.g. fertilizer, manure) or plant-inhibition substances (e.g. round-up) on plants. No questions allowed that relate to microbial growth (no petri plates of bacteria). No questions about extending battery life. No questions regarding the effect of any kind of sound (music, talk, shout, whisper) on plant growth. No using questions already covered in our lab, or given as examples. No questions allowed using human subjects without prior permission of instructor. Take your cue from medical investigators: nearly anything you can do to a human you can do to a rat (unless it involves verbal responses!). As above, human experiments without permission will receive a 20% deduction, on top of other deductions. No Product Comparisons. Different products invariably differ in many ways, so a simple experiment is generally not possible by simply comparing products. As above, 20% deduction for product comparisons. Example Experimental question. Introduction: Venus flytraps eat flies. Flies should not land on Venus flytraps, since it often results in their untimely demise, yet they do so anyway. These observations led me to ask the question: why do flies land on Venus flytraps? My hypothesis is that flies land on Venus flytraps because they are attracted to the meat colored area (a dull red) at the center of the trap. If this is true, then I predict that flies will spend more time in a given area if it is red than they will if it is another color. Methods: Two identical aquaria (25cm x 30 cm x 30cm high) with lids will hold flies during testing. In the experimental aquarium, the entire bottom will be covered with black paper with a 10cm x 10cm paper square that is as close to the color of the Venus flytrap as possible placed on top of the black paper and in the center of the aquarium. In the control aquarium, the entire bottom will be covered in black paper, with a 10cm x 10cm black paper square placed on top and in the center. Thus, the independent variable is the color of the paper square: either red or black. The two aquaria will be set into the same room, with identical lighting and temperature. Flies will be introduced individually to each aquarium and observed for 10 minutes. The dependent variable is the amount of time they spend on the center square of paper; a stopwatch will be used to measure the cumulative time in seconds. 100 flies will be used in each treatment. Expected Results: The expected results of this study are provided in figure 1, which shows a hypothetical data set demonstrating that statistically significantly more time was spent on a red central square than on a black central square. Figure 1. The total time spent on a central square of paper when it is red and when it is black for 100 flies. The central horizontal lines are the means, and the error bars represent 95% confidence intervals.