Forum questions 3

Describe four things that can be done to keep the behavior of the researcher from threatening the validity of the study.

1. double-blind studies are performed with the subjects and the researcher blind to the conditions of the experiment to prevent bias. 2. Automation is the practice of standardizing the instructions to participants and the processes for collecting the data. 3.Randomization of test subjects also controls the bias. 4. having multiple researchers and different people analyze the data also increases the validity of the experiment.

Give an example of a 2 x 4 factorial design in an experiment. Explain why it is factorial and why it is 2 x 4.

A 2x4 factorial design may test the effectiveness of four different dosage levels of a type of sleeping pill given at two different times of day. This is factorial because it allows for examination between two different factors to be compared with each other at different measurable levels. It is 2x4 because there are 2 times of day and 4 dosages of sleeping pills.

What is a placebo? Why is it needed? Explain.

A placebo is a fake, for example a fake drug which has no impact. They ensure a difference in the real drug and fake drug and reduces bias from researchers and participants that may affect the outcome.

Give an example of a single blind study. Why was the single blind study used?

A single-blind study would be one where the subjects do not know what treatment they are receiving but the researcher knows which group they are in. For example, testing out a new and old drug and two different groups, but the people don't know if they are using the old or new drug. This is to make sure the participants are unbiased or do not lie.

Give an example of a situation where a stratified design would be desirable.

A stratified design involves the division of a large population into smaller groups, also known as strata. Those strata's are made based on shared attributes. One example of a stratified design would be determining the income across Tennessee. Researchers would have to look at the population and divide based on shared attributes such as type of job, cities, or other common attributes.

When would a subjective measure be preferable to an objective measure?

A subjective measure would be preferable to an objective measure when you are not able to objectively measure something, such as pain or severity of symptoms, in which case you might ask the patient to subjectively measure their pain or severity of symptoms on a scale of one to ten.

What are advantages and disadvantages of the repeated measurement design?

Advantages: holds more statistical power, requires fewer subjects, measure the effect over time with the same sample, less likely to record mistakes because of multiple repetitions. The disadvantages are that it can take a great amount of time, resources, and money to keep repeating the same experiment.

"The behavior of the researcher may influence the independent variable, thus threatening the validity of the study." Give an example identifying a possible behavior that is a threat to validity.

An example of a threat to the validity of a study is if an organization is funding it. If a pharmaceutical company is funding a researcher to conduct a study on their drugs the researcher may alter the independent variables in order to reach a conclusion that would keep the company content. This would alter the results and lower the validity of the study.

Give an example of a nest treatment design. Explain what makes it nested.

An example would be if researchers gathered two groups of students and tested for caffeine effects on alertness. One group would receive coffee with caffeine and the other group would receive coffee without caffeine. This is a nested treatment design because each group has a single purpose and the analysis is within the subject group. Interactions between the two are not determined.

What does it mean when it says that the results were confounded?

Confounding results are factors that affected the dependent variables aside from the ones the researchers were interested in.

How could we control for edge-effect? Give a specific example.

Edge effects are changes in a population structure near the boundary between habitats. They can be controlled by coordinating the adjacent habitats. Their length, orientation, features, and directionality can be kept track of individually. Spatial optimization can reduce the amount of edges in a plot. For example, if different plots of forest are being examined for growth, each edge between the plots needs to be observed individually so that factors such as sun exposure and wind at an edge do not become a characteristic of the plot as a whole.

What do you learn by each of the following: exact replication, systemic replication, and conceptual replication?

Exact replication: by repeating an experiment you can find mistakes, and see if results are different. Systematic replication: by doing a study with small variance such a change in sample size. Conceptual replication: Testing the same question, but in a different way to see if there was bias or it was not the best way to test.

Distinguish between factorial, nested, and gradient designs.

Factorial: Interactions between different subjects are tested. Nested: The same types of subjects are analyzed with no interaction tested. Gradient: Used when direct manipulation is less possible. Variables treated as continuous.

Define love. Give an operational definition of love - one that can be used to make scientific observations

Love is a feeling of strong affection. An operational definition of love would be holding hands/kissing. This definition includes actions, which can be observed scientifically, unlike the feelings that are typically used to define love, which cannot be observed.

Distinguish between objective and subjective measures. Give an example of each.

Objective measurement is measured consistently and cannot be influenced by opinion or perspective. Example: measuring someone's weight. Subjective measurement is measuring what people say, such as answers to open-ended questions or ranking something based on feelings. Example: a doctor asking a patient to measure their pain on a scale of 1 to 10.

What are pseudoreplicates? Give an example.

Pseudoreplication is artificially increasing your sample size and treating the data as independent observations. For example, you can test soil levels from a plot of growing cabbages but take all your samples from different spots of the plot. You are artificially increasing your sample size instead of finding different plots of growing cabbages.

What is randomization in an experiment and why is it important?

Randomization involves randomly distributing subjects into different testing groups in an unbiased manner. It is important because it controls for an unknown bias which may lead to bad results.

What is the difference between sample size and the number of replicates in an experiment?

Sample size refers to the number of experimental units used in an experiment. Replicates are additional experimental units assigned to the same treatment group. This allows for random variation to be assessed in the experiment.

What is the difference between statistical power and effect size?

Statistical power is how likely you are to detect a significant difference through a statistical test. Effect size is the importance/meaning of the difference.

What do you learn from subsampling?

Subsampling is useful because it allows for replication to occur if needed. Subsampling ensures consistency in your experiment when you subsample as well as testing the precision of the instruments used.

What methods are used to reduce error variance? Be specific.

The most practical way to decrease error variance is to increase the sample size and select randomly. The more subjects and observations you have, the less error variance your experiment will suffer. Also, one must maintain or increase the level of constriction. Make sure to use precise instruments during experimentation.

Give an example of a double blind study. Why was the double blind study used as compared to a single blind study?

When investigating a new drug, neither assistants interacting with patients and administering the drug nor the patients would know which drug (real or placebo) they are receiving. The double blind experiment prevents researchers and participants from knowing which treatment they are administering/receiving, which makes it less likely for researchers to accidentally reveal subtle hints that will influence the outcome and less likely for patients' beliefs to influence the outcome.

What happens when the sample size is too small?

When sample size is too small, there is the risk of more standard error ( outliers) in comparison to the population which would mean that the sample does not accurately represent the population.

Can the sample size be too large? What happens if the sample size is too large?

Yes, the sample size can be too large. Large sample sizes can lead to an excessive amount of time spent on an experiment and may also generalize the experiment so that the question at hand may not be answered.