2_Hypothesis_testing_II

2_Hypothesis_testing_II - 9/1/10 Last lab Experimental...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 9/1/10 Last lab Experimental Design, Graphing, and Data Analysis BIOL 1005 Sec,ons 9 and 10 •  Observa,ons of 3 different organisms •  Models and generaliza,ons •  Formula,ng hypotheses (Null and Alterna,ve) This week: •  Experimental Design •  Graphing and Data Analysis Formula?ng Hypotheses •  Before we design an experiment to test a model we need to formulate a NULL and an ALTERNATIVE hypothesis. •  The null and alterna,ve hypotheses should cover ALL possible results that you may get at the end of your experiment. Formula?ng Hypotheses •  Null Hypothesis: –  States that there will be NO difference in the outcome of your treatment and your control groups •  Alterna,ve Hypothesis: –  States that there WILL be a difference between the outcome of your treatment and control groups Hypothesis vs. Predic?on •  Hypotheses will only state whether we will or will not see a difference between 2 groups. EXAMPLE: Fer?lizer ­Tomato Plant Experiment: Interested in tes,ng whether fer,lizer will have an effect on the number of fruits produced by a tomato plant •  Null Hypothesis: There will be no difference in the number of fruits produced by the plant grown with the fer,lizer compared to the plant grown without the fer,lizer. •  Alterna?ve Hypothesis: There will be a difference in the number of fruits produced by the plant grown with the fer,lizer compared to the plant grown without the fer,lizer. •  Predic?on: The plants grown with the fer,lizer will produce more fruit than the plants grown without fer,lizer. •  A PREDICTION states in what direc?on we will see that difference –  Predic,ons are more specific than hypotheses 1 9/1/10 Experimental Design Some Important Considera?ons: 1.  –  –  –  Experimental Design Some Important Considera?ons: 3.  –  –  –  Number of variables (e.g. treatment) to be manipulated: ONE!!! More than one variable (mul,variate) gets complicated The group in which the variable is manipulated is called the “Experimental Group” or “ Treatment Group” Replicates: repe,,on increases the reliability and accuracy of your results Es,mate variability of results Depends on the nature of the experiment, but more are be[er In lab ­ minimum of 3 replicates Think about HOW are you going to measure this variable 5.  Scale of measurements: What are you measuring? –  –  –  2.  –  Control Group: Standard to which we compare our results A[empt to equalize all the external condi,ons Be[er measure things in a small scale and then lump data into groups than to use a scale so big that you can miss important differences Reasonable ,me frame (e.g., Isopods, termites: minutes, Copepods: seconds) Metric units: m, cm, g, kg, etc. Experimental Design 5.  Criteria for suppor?ng/rejec?ng your hypothesis: –  How big of a difference you need to see? –  Descrip,ve STATISTICS to describe the popula,on of data points that you have collected  MEAN: Average of a series of data points  STANDARD DEVIATION: Describes how much the data spreads from the mean EXAMPLE: Fer?lizer ­ Tomato Plant Experimental Design: •  Two groups: –  Control Group: Tomato plants grown without fer,lizer –  Treatment Group: Tomato plants grown with fer,lizer •  Number of replicates: 3 plants in each group •  Unit of Measurement: Number of fruits per plant •  Table for data collec?on •  Criteria for suppor,ng/rejec,ng hypotheses: graphing and mean/std dev. Data Collec?on Example In Lab: Data Collection Replicate Number 1 No fertilizer (control group) 12 Fertilizer (treatment group) 14 Replicate Number Data Analysis Example No fertilizer (control group) 12 Fertilizer (treatment group) 14 2 15 16 3 16 18 This is an example of what your raw data will look like. You should record these data in the pages at the back of your lab manual 1 2 15 16 Calculate the average and standard deviation for the 3 replicates of each group. 3 Mean # of Fruits: Standard Deviation: 16 (12+15+16)/3= 14.3 18 (14+16+18)/3= 16.0 √ (12-14.3) 2 + (15-14.3) 2 +(16-14.3)2 (3-1) √ 12-14.3) 2 + (15-14.3) 2 +(16-14.3)2 (3-1) 2.1 2.0 Your final table should only include the mean and std. dev. of the replicates 2 9/1/10 Data Analysis: Table Example Table 1: Average number of fruits produced by tomato plants grown with fertilizer and without fertilizer. Every table should have a table caption numbered and placed ABOVE the table Use Excel or Word to generate tables Data Analysis: Graph Example Always make sure the axes are labeled with specific information and units. Use Excel to generate graphs Tomato Plant: No fertilizer (control group) Mean # of Fruits 14.3 ± 2.1 Tomato Plant: Fertilizer (treatment group) 16.0 ± 2.0 Always include error bars generated from the standard deviation calculated. Remember to be specific about what you were measuring Every time you cite a mean you should also include +/- the standard deviation. This tells you how widely spread the data were around the average. All tables should be clearly labeled so that there is no confusion what the treatment/ control group was. Every figure (e.g. graph) should have a descriptive figure caption placed BELOW the figure. Figure 1: The effect of fertilization on mean fruit production by tomato plants. Table vs. Graph Graphs show trends Tables show exact values Table 1. Number of total bats, female bats, and male bats captured in 2006 Bat species More details about graphing, tables and sta,s,cs Number of individuals captured 21 Number of female bats captured 10 Number of male bats captured 11 LET’S SEE THAT AGAIN… Fig. 1: Proportion of female reproductive and non-reproductive bats in two sites. Lophostoma silvicolum Platyrrhinus helleri Carollia perspicillata Artibeus obscurus 22 7 15 15 3 12 20 10 10 Making graphs 1. Choose the type of graph that best represents the data Histogram Graph 16 Number of individuals Making graphs 1. Choose the type of graph that best represents the data Line graph 30 Leaf lenght (mm) Sca[er Plot Bar graph 25 20 15 10 5 0 1 2 Monkeygrass 3 Turf 4 Collec?on number (day) 14 12 10 8 6 4 2 0 <1.5 1.5  ­ 1.75 1.75  ­ 2 Size classes  ­ Height (m) > 2 Frequency distributions showing the frequency of observations falling into a series of numerical categories plotted on the x axes Every individual data point is plotted directly on the graph Show correlation or strength of association between two variables When the dependent variable represents a continuous function of the independent variable e.g. Daily increase in leaf length between two different types of grasses When the independent variable presents qualitative categories e.g. Proportion of reproductive and non reproductive bats visiting two different types of habitat 3 9/1/10 Graph Details 2. Put informa?on on appropriate axis 3. Label both axes and include units Graph Details 4. Use easy to read and even increments on both axes 5. Plot the points (use EXCEL) 6. Do not connect the dots  ­ use a ‘best ­fit’ or ‘trendline’ X = independent variable (what you manipulated) Y = dependent variable (what you measured) Graph Details 7. Use real graph paper: Fit figure on paper Data Analysis Descrip?ve Sta?s?cs * Describes the popula,on of data points collected Popula?on vs. Sample: A sample is a small subset representa?ve of a popula?on Mean: average of a series of data points Fig 1. Growth of oaks over time Standard Devia?on: this value shows you how far from the mean the data spreads. * uses the same units as the mean * represented as Mean ± std dev. * Use the standard devia,on to plot error bars Data Analysis Once you have calculated the descrip?ve sta?s?cs (mean and standard devia?on) you can use these values to determine which of your hypotheses should be rejected Evaluate how similar two distribu,ons really are. Plot the mean and standard devia?on 1. Plot the mean and standard devia?on for you data Standard deviation Up 2.1 from the mean Down 2.1 from the mean St. dev = 2.1 4 9/1/10 Determine Overlap 2. Determine the amount of overlap between the error bars of the control and treatment Do the error bars overlap? - YES Is there a statistical difference in the number of fruits produced between plants in fertilized and unfertilized soil? - NO!!! Determine overlap 2b. Determine the amount of overlap between these values Do the error bars overlap between Maple and Oak? - NO Is there a statistical difference in the growth rate of Maple and Oak? -YES!!!! Which hypothesis do we reject? Let’s see our hypotheses again: •  Null Hypothesis: There will be no difference in the number of fruits produced by the plant grown with the fer,lizer compared to the plant grown without the fer,lizer. •  Alterna?ve Hypothesis: There will be a difference in the number of fruits produced by the plant grown with the fer,lizer compared to the plant grown without the fer,lizer. Which hypothesis do we reject? ASK THE QUESTION: Did we see a difference between the control and treatment group in terms of number of fruits produced?  ­NO! So, then, which hypothesis do we reject?  ­ The alterna,ve (because there was no difference between the control and treatment group) Can we accept a hypothesis? •  We do NOT accept hypotheses, rather we say : Can we accept a hypothesis? •  This is because… –  If you didn’t find support for the alterna,ve it could be because the measurement you used to evaluate it was not the correct one, but if you had used a different measurement or a different variable to evaluate the same hypothesis you may find different results –  Hence, accep,ng a hypothesis is TOO DEFINITIVE. –  If there are no significant differences (std. dev. bars overlap), results are CONSISTENT with the null hypothesis or that the results SUPPORT the null hypothesis. –  If you find significant differences you state your results SUPPORT, or are CONSISTENT with your alterna,ve hypothesis 5 9/1/10 What next? •  Go back to your model •  New hypotheses –  Always keep in mind other possible explana,ons For Lab •  Go over sec,ons in the student resource manual sec,on –  Bias, control, error, experimental design, graphs, hypotheses null and alterna,ve, metric system, orienta,on behavior, random, sampling, sta,s,cs, variables •  New experiments •  It never ends! •  Review oral report format document on Moodle •  Read a li[le about the biology of your chosen organisms Lab today In groups (same as last week) •  What was your generaliza,on from last week? Termites •  Culture tubes •  Ballpoint pens (different colors) •  White paper •  Different colored paper •  Different types of markers/pens •  Toothpicks •  Ethanol Copepods •  Culture tubes •  Red/blue/green acetate filters Isopods •  Reflector lamps •  Culture tubes •  Flashlights •  Petri dishes •  Lamps of different Clear and black with and wa[age for reflectors without gaps •  Drierite •  Nylon screen •  Reflectors •  Paper towels •  Water spray bo[les Before you start experiments… Show me the following: •  Hypothesis – null and alterna,ve –  Predic,ons if you have any •  Run experiments –  Data collec,on Up to one hour 30 mins •  Experimental Design –  Variable you are going to manipulate –  What is the treatment –  How are you going to measure the effect •  Time  ­ units •  Number of individuals? •  Data Analysis –  Graphs vs. table –  Sta,s,cs –  Conclusions –  How differen,ate between control and treatment –  How many replicates •  Table for data collec,on 30 mins •  Which hypothesis do your results support? 6 9/1/10 Week arer next •  Oral reports  ­ 8 mins each group –  Write relevant data/graphs/tables on your sec,on on the board –  Which hypothesis do your results support? •  Show your sta,s,cs •  Conclusions •  No class next week (Labor Day holiday) •  Quiz – Hypothesis, experimental design, graphing, data analysis, ques,on or 2 from upcoming lab to make sure you are prepared •  Read Chapter 2 Lab Manual – What is Life (pg. 9 – 12) •  Go over links on Moodle –  What is Life Links file –  Phoenix Mars Lander file –  Report Worksheet Word document –  Microscopy Tutorial •  Everybody in the group must par,cipate in the presenta,on •  Get a GROUP GRADE!!! Total one hour !!!REMEMBER!!! •  Take all materials back to the supplies table •  Clean hands with hand sani?zer ( table next to door) •  Clean your work sta,ons and push stools UNDER the bench 7 ...
View Full Document

This note was uploaded on 03/19/2011 for the course CMST 1150 taught by Professor Jackson during the Spring '07 term at LSU.

Ask a homework question - tutors are online