DDD- Chapter 2 Flashcards
Not a truly representative sample
This misuse occurs when the sample that a statistician chooses isn’t truly representative of the entire population* he or she will draw a conclusion about. If there are important differences between the sample and the larger population, the conclusion made from the same may not represent the larger population.
Response bias
This misuse occurs when the respondents to a survey say what they believe the questioner wants to hear. This bias can occur as a result of the wording of a question.
Conscious bias
This misuse occurs when the surveyor is actively seeking a certain response to support his or her theory or cause. Bias can occur when the researcher manipulates the phrasing of questions in order to elicit the desired response.
Missing data and refusals
This misuse occurs when a certain part of the sample gets lost, or subjects refuse to contribute to the overall data collection. This can distort the survey data significantly as you could lose demographic segments of the population under study and consequently could arrive at a false conclusion.
Small sample sizes
This misuse occurs when a sample size is too small to draw inferences from.
Association and causality
This misuse occurs when a researcher notices a relationship between two variables and assumes that one variable is the cause of the other. In reality, these variables might both be caused by a separate variable. In this case, they would merely be correlated, which means they show up together. Or there might be no relationship at all.
Training and test data
This misuse occurs when the same data that’s used to form a hypothesis is then used to test that hypothesis. This misuse most often occurs when a small population is being studied, and different samples have a lot of crossover.
Unfounded assumptions
This misuse occurs when an assumption is made that has not been proven.
Faulty operationalization
Operationalization refers to the development of specific research procedures that allow for observation and measurement of abstract concepts. For example, if a researcher wants to study how new parents feel about their financial security, he/she can operationalize this objective by determining a testable hypothesis; developing a mechanism for collecting observations (for example, a survey); identifying a representative sample (200 randomly selected parents of newborns in suburban towns); asking relevant questions; and interpreting the data received. Each of these research decisions (hypothesis, method, sampling, questions, interpretation) is an example of operationalization.
A key aspect of operationalization is defining variables and attributes that adequately represent the concept of the study. For example, to assess attitudes about financial security, researchers might ask respondents to characterize their feelings regarding their financial security as “anxious” “confident” and so forth. These responses can be coded as “1” “2” “3” etc., for analysis. Or they could ask respondents to specify how much money they have in a college savings plan. In both examples, these are measurable dimensions that serve as a proxy for the non-measurable concept the researcher is studying. If the researcher’s reasoning behind any aspect of operationalization is faulty, it can result in misleading or irrelevant findings. For example, data showing high amounts in college savings plans may not necessarily correspond to a strong feeling of financial security. The reasoning making this connection is flawed, and leads to inaccurate conclusions.
Lack of blinding
A lack of blinding can cause bias to occur. Blinding is when researchers place barriers between themselves and subjects in order to ensure that the researchers do not influence subjects’ behavior during the experiment. Without blinding, subjectivity can be introduced into the results.
True or False?
A misuse of assumptions occurs when someone tests something that is accepted to be true but has not yet been proven.
False
Correct. This is a false statement. A misuse of an assumption occurs when it is used to make conclusions.
True or False?
Oscar surveys thirty households in Ohio about their preferences for pasta and, based on the results, concludes that a majority of Midwesterners prefer spaghetti over egg noodles. He is correct in making this inference.
False
Correct. This is a false statement. This sample size is too small with only thirty households.
Where is the flaw in the following experiment?
George is working for a nonprofit, trying to determine the unemployment rate in his city. George surveys people in different parts of the city between 3:00 P.M. and 5:00 P.M. every weekday to gain results. George is able to get about 100 people a day to respond to his survey over a two-month period. George determines that the unemployment rate in his city is 15 percent.
a. association vs. causality
b. small sample size
c. response bias
d. missing data
d. missing data
Correct. The correct answer is D. While going to the different parts of the city, George does not take into account the many people who are at their jobs from 3:00 P.M. to 5:00 P.M. This has likely caused George’s measured unemployment rate to be higher than the true unemployment rate.
Where is the flaw in the following experiment?
As part of their succession planning strategy session, the board of directors for a company with a long history of providing insurance to local families is brainstorming ideal qualities for a CEO candidate. One member suggests that they look for a candidate that closely resembles Bill Winters, the CEO that was in place in the late 60s when the company successfully scaled despite economic hardship in the region. They remember Winters as smart, effective, and authoritative. Based on the reasoning that what worked during a shaky economic period in the 60s will work in present day, they decide that they will groom Carson Sandel as the next CEO because his characteristics are most similar to Bill Winters’s.
a. unfounded assumptions
b. lack of blinding
c. association vs. causality
d. conscious bias
a. unfounded assumptions
The correct answer is A. The unfounded assumption is that what worked in the 1960s will also work now. The company’s market has probably greatly changed in the more than 50 years since the CEOs that were successful in the 1960s, and the company has probably changed as well.
