1. 4.2 Designing a Good Experiment
1.1. Who Participants in Randomized Experiments?
1.1.1. Most participants who participate in randomized experiments are volunteers.
1.1.1.1. Often times the volunteers are recruited through newspaper ads.
1.2. Randomization: The Crucial Element
1.2.1. Randomization: random assignment to treatments or conditions.
1.2.2. Randomizing the Treatment
1.2.2.1. Keeps researchers from picking favorites and prevents bias in the study being conducted.
1.2.2.2. The important principle about randomization of studies is that it provides the same opportunity for all the individuals in the study to have the same probability of being picked.
1.2.3. Randomizing the Order of Treatments
1.2.3.1. In most studies all treatments are applied to every unit but the randomization can randomize the order in which they receive which treatment.
1.3. Control Groups, Placebos, & Blinding
1.3.1. Control group: treated identically to the experimental group but don't receive the treatment but may receive a placebo
1.3.1.1. Control groups are created so researchers can record data on the individuals who would not receive any form of treatment.
1.3.2. Placebos: looks like the real treatment but has no actual effect.
1.3.2.1. Placebo effect can be so great researchers give half of the participants in the study the real treatment and the other half the placebo. Patients do not know that they are receiving the placebo therefore their results in the study should not be biased toward either treatment.
1.3.3. If researchers know which individuals are receiving which treatments, their biased towards those patients may affect the outcome of the study. The best approach to receiving unbiased data is to have a double-blind study.
1.3.3.1. Single-blind: where the participants do not know which treatment they are receiving or if the participants knew what treatment they were receiving and the researchers were unaware.
1.3.3.2. Double-blind: both the experimenter or the subjects know if they are receiving the treatment or a placebo
1.3.3.2.1. Double-blind experiments are the preferred over single-blind but they are not always possible to conduct.
1.3.4. Double Dummy: giving every participant both treatments to ensure that the experiment is blind while only one of the treatments is actually active.
1.4. Pairing and Blocking
1.4.1. Researchers will match participants together based on similarities such as sex, age, weight or IQ, thinking that their results of the study will be similar.
1.4.1.1. Matched-Pair Designs: two matched individuals or the same individual to receive each of the two treatments
1.4.1.1.1. It is important to remember that the order of the treatments assigned to each individual is randomized.
1.4.1.1.2. The best way to conduct this experiment would still be to have the experiment be a double-blind so neither the researchers or patients knew in which order they were receiving the different treatments. This would keep the experiment from becoming biased.
1.4.2. Blocks: homogeneous groups.
1.4.3. Repeated-measures designs: participants are measured repeatedly under the differing conditions.
1.5. Design Terminology and Examples
1.5.1. Completely randomized design: treatments in an experiment are randomly assigned to experimental units without using matched pairs or blocks.
1.5.2. Matched-pair design: when matched pairs are used.
1.5.3. Randomized block design: when blocks are used.
2. 4.4 Difficulties & Disasters in Experiments & Observational Studies
2.1. Complications can occur when conducting studies. The following are some of the many complications that may occur:
2.2. Confounding Variables & the Implication of Causation in Observational Studies
2.2.1. Biggest mistake that researchers make in an observational study is stating the cause and effect relationship.
2.2.1.1. Rule for Concluding Cause & Effect: cause-and-effect relationships can be inferred from randomized experiments but not from observational studies.
2.2.2. Partial solution of an observational survey would be to take into consideration the different variables acquired in the study.
2.3. Extending Results Inappropriately
2.3.1. Fundamental Rule for Using Data for Inference: Available data can be used to make inferences about a much larger group if the data can be considered to be a representative with regard to the question(s) of interest.
2.3.2. Most observational and randomized studies use volunteers.
2.3.2.1. Since volunteers are used, it is up to the consumer to decide if the data collected is applicable to a larger population of people.
2.4. Interacting Variables
2.4.1. a 2nd explanatory variable interacts with the principal explanatory variable in its relationship with the response variable.
2.5. Hawthorne & Experimenter Effects
2.5.1. Hawthorne effect: those being watched perform differently than expected because they are aware they're being watched
2.5.2. Experimenter effects
2.5.2.1. Recording data erroneously to match the desired outcome.
2.5.2.2. Treating subjects differently on the basis of which condition they are receiving
2.5.2.3. Subtly making the subjects aware of the desired outcome
2.5.2.4. Most experimenter effects can be overcome by using double blind experiments
2.6. Ecological Validity & Generalizability
2.6.1. Lack of Ecological validity: the results don't accurately depict a real life sitation because it is a clinical trial
2.7. Using the Past as a Source of Data
2.7.1. These are often based on an unreliable account of someone's memory
2.7.2. Researchers may not think to measure in the fact that confounding variables in the past may no longer be relavent or similar to the confounding variables of today
3. 4.1 Speaking the Language of Research Studies
3.1. Cause-and-effect relationship: Does the changing one variable effect another?
3.1.1. How a study is conducted will affect the way we interpret the relationship of cause-and-effect.
3.2. Types of Research Studies
3.2.1. Observational Study: researcher observe or question the participants about opinions, behaviors, or outcomes
3.2.1.1. In observational studies, researchers are not asking participants to make drastic changes to their lives, they are merely observing their behaviors and lifestyles.
3.2.2. Experiment: researchers manipulating something and then measuring the effect
3.2.2.1. Randomized experiment: participants are randomly assigned to participate in a control or experimental group. These groups are known as treatments.
3.2.2.1.1. Random experiments attempt to have similar individuals in the experiments so when the different treatments are tested the difference in the results will come from the treatment given.
3.3. Observational Study or Randomized Experiment
3.3.1. Randomized experiment shows a stronger cause-effect relationship than observational studies
3.3.2. Ethical reasons force researchers to conduct more observational studies than randomized experiments.
3.3.3. Researchers must always weigh out the advantages and disadvantages for every strategy that they use in a study.
3.4. Who Is Measured: Units, Subjects, Participants
3.4.1. Experimental unit: the basic thing or person that is receiving the treatment or being tested
3.4.2. Subjects or participants: people who are experimental units
3.4.3. Participants in randomized experiments are usually volunteers and are also blind to the treatments that they will be receiving.
3.5. Explanatory and Response Variables
3.5.1. Explanatory (independent) variable: may cause explain a response variable
3.5.2. Response (dependent) variable: the outcome of a specific treatment
3.5.3. Researchers conduct these different studies because they want to discover more about the relationship of two or more variables.
3.6. Confounding Variables-Measured or Not
3.6.1. Confounding variable: a variable that both affects the response variable and also is related to the explanatory variable.
3.6.1.1. Confounding variable are more commonly present in observational studies.
3.6.1.1.1. Although they are more commonly present in observational studies, confounding variables can make the interpretation of the results problematic.
3.6.1.2. Confounding variables are less likely to affect the interpretation of randomized experiments.
3.6.1.2.1. Randomized experiments have been designed to help keep confounding variables under control.
3.6.2. Lurking variable: a potential confounding variable that is not measured and is not considered in the interpretation of the study.
4. 4.3 Designing a Good Observational Study
4.1. Types of Observational Studies
4.1.1. Retrospective study: data from the past
4.1.2. Prospective study: researchers follow participants into the future and record relevant events and variables
4.1.2.1. A prospective approach is usually a better procedure than a retrospective study because people often have difficulty remembering past events accurately
4.1.3. Case control study: "cases" who have a particular attribute or condition are compared to "controls" who do not.
4.2. Advantages of Case-Control Studies
4.2.1. Ethical Considerations
4.2.1.1. They do not suffer from ethical considerations that are inherent in the random assignment of potentially harmful or beneficial treatments
4.2.2. Efficiency
4.2.2.1. A new experiment doesn't need to be set up and executed. "Cases" and "controls" that have already happened are looked at and observed.
4.2.3. Reducing Potential Confounding Variables
4.2.3.1. Controls can be chosen to reduce potential confounding variables