Presentation on theme: "By Julie-Anne Spatz and Adam MacLeod"— Presentation transcript:
1By Julie-Anne Spatz and Adam MacLeod Experimental DesignBy Julie-Anne Spatz and Adam MacLeod
2Experiment vs. Observational Study An experiment deliberately imposes a treatment on a group of units or subjects in the interest of observing the response.VS.An observational study involves collecting and analyzing data without changing existing conditions or imposing a treatment.
3The Experiment“This study was designed to examine the effects of carbohydrate-electrolyte ingestion on physical and mental function associated with the performance of intermittent high-intensity (IHI) exercise similar to many common competitive sporting events.”
4In Other Words…Researchers, RS Welsh, JM Davis, JR Burke, and HG Williams, examined athletes physical performance while drinking either Gatorade or water. We will use this experiment to explain general terms about experimental design.
5Units or SubjectsExperimental units: the individuals on which the experiment is doneSubjects: when the units are human beingsThe subjects used were 60 male soccer players.
6TreatmentsTreatment: a specific experimental condition applied to the unitsPlacebo: a dummy treatment, that can have no physical effectFactor: a controlled independent variable (called the explanatory variable), whose levels are set by the experimenterLevel: each treatments specific valueThe factors in this experiment include the amounts of Gatorade and water given to each soccer player every fifteen minutes throughout a soccer game.Example of levels:Gatorade: 6 ounces, 8 ounces, 10 ouncesWater: 6 ounces, 8 ounces, 10 ounces
7VariablesControl Group: the subjects that are not subjected to the treatmentExperimental Group: the subjects that are subjected to the treatmentControl Group: people received only water (30 people)Experimental Group: people received only Gatorade (30 people)
8RandomizationRandomization: the use of chance to divide experimental units into groupsTo conduct a completely randomized experiment:1. use a Table of Random Digits2. use the calculator to generate random numbers
10Calculator Technique Stat Edit highlight L1 Math PRB 5:randInt( Type in randInt(lower bound, upper bound, amount of integers needed)
11Stratified Random Sampling vs. Multistage Sampling Stratified Random Sample: an experimenter would first divide the population into groups of similar individuals called a strata, then choose a separate SRS in each stratum and combine these SRSs to form the full sampleMultistage Sampling: is a sample where the elements are chosen in more than one stageEx. 1: If the experimenters wanted to do a stratified random sample, they could do this by putting soccer players in to different stratas by region of the United States and then conduct the experiment.Ex. 2: If the experimenters wanted to do a multistage sample, they could do this by choosing the soccer players at the state level, then the city level, then the county level, and then by league.
12Randomization Technique Used The experimenter used a random number table to assign each soccer player a number 1 thru 60. Then, the people with odd numbers were put into the control group and the people with even numbers were put into the experimental group.
13Single Blind Experiment vs. Double Blind Experiment Single Blind Experiment: the experimenters, but not the subjects know which treatment a subject receivedDouble Blind Experiment: the experimenters nor the subjects know which treatment a subject receivedThis experiment was a single blind experiment because the experimenter knew what groups the subjects were put in, but the subjects were not aware. To ensure that the subjects could not tell the difference between the Gatorade and water, the experimenters put a type of sweetener into the Gatorade. The sweetener only affected the flavor, but not the amount of electrolytes and carbohydrates.If the experiment were to be a double blind experiment, the experimenter would have had to of had a separate person assign which groups would receive the Gatorade and water, without the experimenters knowledge. Then the researcher could take over conducting the experiment.
14Matched Pairs Matched Pairs: experiment that compares two treatments Ex. 1: If this experimenter were to use matched pairs, every soccer player would drink water throughout one game, then Gatorade throughout another, and the experimenter would note which one yielded higher performance.Ex. 2: An experiment is conducted to compare the taste of a new spaghetti sauce with the taste of a successful sauce. Each of a number of subjects tastes both sauces in random order and notes which one tastes better.
15Block DesignBlock: a group of experimental units or subjects that are similar in ways that are expected to affect the response to the treatmentsBlock Design: the random assignment of units to treatments is carried out separately within each block.Ex. 1: If the experimenter thought gender was an issue, the experimenter would separate the men from the women, and test the effects of both the Gatorade and the water on the two groups.Ex. 2: If the experimenter thought that age was an issue, the experimenter would separate the people into four age groups; one of people aged 1-15, one of people aged 16-30, one of people aged 31-45, and one of people aged and test each group with the Gatorade and water.Ex. 3: If the experimenter was interested in physical condition, the experimenter would separate the subjects into two groups of strong physical condition and weak physical condition, and then test the Gatorade and water on each group.
16Block Design By Gender Treatment 1 Gatorade Group 1 30 Men Treatment 2 WaterRandom AllocationCompare ResultsTreatment 1GatoradeGroup 230 WomenTreatment 2Water
17Cautions of Experimental Design Hidden Bias: if the units or subjects are dealt with any different ways, hidden bias can arise (unequal conditions introduce bias)Ex. If the researcher were to hand a soccer player Gatorade and smile and hand the other soccer player water with a frownLack of Realism: units or subjects may not believe in the experiment being conducted, making the results of the experiment less reliableEx. If the soccer players already believed that Gatorade produced stronger athletic performance than water
18Cautions of Experimental Design Continued Undercoverage: not including the whole populationEx. If the researcher only included soccer players from the East coast.Non-response: Some people may choose not to respond to a questionnaire or phone survey, decreasing the number of responsesEx. If some soccer players turned down the opportunity to participate in the experiment
19Confounding Variables Confounding Variable: is a "hidden" variable that affects the variables in question but is not known, and distorts the resulting dataIn this experiment, some confounding variables include diet, athletic ability, age, weather conditions, gender, etc.