## Glossary of Homo Ratman Testies

Created by dudeofthedayno1

### Deck Info

#### Description

#### Tags

#### Recent Users

#### Other Decks By This User

- 1. How much area of the normal curve falls between –1.96sd and +1.96sd?

A. 10%

B. 64%

C. 95%

D. 100% - C. 95%

- 2. What is an index?

A. Asking for exact numbers, like hours of TV viewed

B. It is a type of sampling used to even out strata

C. A multiple item measure used to assess magnitude of knowledge

D. The part of the book that defines jargon - C. A multiple item measure used to assess magnitude of knowledge

- 3. After you have tested a new scale what do you do?

A. Use the test results to estimate population values

B. Fix any problems and test it again

C. Figure out the sampling error

D. Change the scale to direct magnitude estimates - B. Fix any problems and test it again

- 4. What is a scale?

A. How high the level of confidence is

B. The graphic layout of sampling error

C. How many questions can be in a questionnaire before people get tired of it

D. A multiple item measure usually used to assess attitude, opinion or i - D. A multiple item measure usually used to assess attitude, opinion or image

- 5. Why are some scale items positive and some negative?

A. To prevent response bias

B. So the average is always the middle of the scale

C. So people don’t get tired of filling out the scale

D. So you can have a negative scale score - A. To prevent response bias

- 6. In order to put verbatim responses into categories, what process should you follow?

A. Figure out categories before the study begins

B. Have at least two coders put responses into categories and check for disagreement between the coders

C. Always - B. Have at least two coders put responses into categories and check for disagreement between the coders

- 7. Multi-stage sampling is …

A. Sampling groups from one frame then sampling from those groups, such as sampling 5 zip codes from a city, then sampling addresses from those 5 zip-codes

B. Not random

C. Selected from cities in time sequences

D. Use - A. Sampling groups from one frame then sampling from those groups, such as sampling 5 zip codes from a city, then sampling addresses from those 5 zip-codes

- 8. What is the normal curve?

A. The shape of heads of Communication Faculty

B. The distribution of sampling errors for different sample sizes

C. The curve from which we get χ2, t-values, r-values, etc.

D. The probability distribution of equally lik - D. The probability distribution of equally likely events

- 9. Which of the following is true about the selection of sample sizes?

A. To cut sampling error in half you must double the sample size

B. To cut sampling error in half you must quadruple the sample size

C. To cut sampling error in half you must cut - B. To cut sampling error in half you must quadruple the sample size

- 1. How much area of the normal curve falls between –1.96sd and +1.96sd?

A. 10%

B. 64%

C. 95%

D. 100% - C. 95 %

- 2. What is an index?

A. Asking for exact numbers, like hours of TV viewed

B. It is a type of sampling used to even out strata

C. A multiple item measure used to assess magnitude of knowledge

D. None of the above - C. A multiple item measure used to assess magnitude of knowledge

- 3. The example for ratings showed that for ±1.96sd the SE was ±3.1 rating points. How confident are you of the true result being within ?3.1 rating points?

A. 64%

B. 75%

C. 95%

D. 100% - C. 95%

- The normal curve is…

A. The only probability distribution used by social scientists

B. Not the only curve use by social scientists. Others include “chi-square,” “F,” “t,” and “u”

C. Not the only curve use by social scientists. Other - B. Not the only curve use by social scientists. Others include “chi-square,” “F,” “t,” and “u”

- Why are some scale items positive and some negative?

A. To prevent response bias

B. So the average is always the middle of the scale

C. So people don’t get tired of filling out the scale

D. So you can have a negative scale score - A. To prevent response bias

- In order to put verbatim responses into categories, what process should you follow?

A. Figure out categories before the study begins

B. Have at least two coders put responses into categories and check for disagreement between the coders

C. Always ask - B. Have at least two coders put responses into categories and check for disagreement between the coders

- Multi-stage sampling is …

A. Sampling groups from one frame then sampling from those groups, such as sampling 5 zip codes from a city, then sampling addresses from those 5 zip-codes

B. Not random

C. Selected from cities in time sequences

D. Used f - A. Sampling groups from one frame then sampling from those groups, such as sampling 5 zip codes from a city, then sampling addresses from those 5 zip-codes

- Convenience samples are…

A. Random, so you use the SEmax and SE calculations

B. Used only for cross sectional surveys

C. Selected from more than three locations to be valid

D. Not random - D. Not random

- Which of the following is true about the selection of sample sizes?

A. To cut sampling error for N=100 in half you have to increase to N=200

B. To cut sampling error for N=100 in half you have to increase to N=300

C. To cut sampling error for N=100 i - C. To cut sampling error for N=100 in half you have to increase to N=400

- 10. The area under the normal curve between –1.96sd and + 1.96sd is important to social scientists. Why?

A. It corresponds to being 95% confident that estimates from sample-based studies are accurate

B. It is the area of doubt for estimates from sam - A. It corresponds to being 95% confident that estimates from sample-based studies are accurate

- 11. Which of the following is true about sample size?

A. As sample size increases from 100 to 200, sampling error falls from 9.8% to 6.9%

B. As sample size increases, sampling error decreases

C. Increasing the sample size from 100 to 400 cuts the sam - D. All of the above are true

- 12. Why is it important to have interviewer instructions?

A. To make sure bias is reduced by standardizing how the questions are asked

B. To make sure that telephone numbers match respondents, such as voters

C. So interviewers show up to work on tim - A. To make sure bias is reduced by standardizing how the questions are asked

- 13. What basic analysis can you do in a spreadsheet?

A. Percentage

B. Average

C. Median

D. All of the above - D. All of the above

- 14. Which of the following can affect internal validity?

A. Regression

B. Testing

C. History

D. All of the above - D. All of the above

- 15. What is contact rate?

A. Number of interviews completed divided by the total number of units in the sample

B. Number of interviews completed divided by the number of legitimate units in the sample

C. Number of interviews completed divided by the - D. None of the above

- 16. Why use a sample?

A. So your estimates will be 100%

B. It is more time and cost effective than a census for large populations

C. You don’t have to worry about testing effects

D. All of the above - B. It is more time and cost effective than a census for large populations

- 17. If you use volunteers what do you need to bear in mind?

A. That they are representative of the overall population

B. That their answers cannot be generalized to a population

C. That you must check to see if you have a balance of males and femal - B. That their answers cannot be generalized to a population

- 18. Significance tells you…

A. The probability that there is a relation between two variables

B. Whether your sampling error for descriptive data is accurate enough

C. If you have a strong relation

D. How much explained variance there is - A. The probability that there is a relation between two variables

- The area under the normal curve between –1.96sd and + 1.96sd is important to social scientists. Why?

A. It is the area of uncertainty that censuses represent the population corresponds sample-based studies are accurate

B. It is the area of doubt for - D. It is the basis for saying there is a high probability that our sample represents the population

- Which of the following is true about sample size?

A. As sample size increases from 100 to 200, sampling error falls from 9.8% to 6.9%

B. As sample size increases, sampling error will eventually decrease to zero

C. Increasing the sample size beyond 40 - A. As sample size increases from 100 to 200, sampling error falls from 9.8% to 6.9%

- Why is it important to have interviewer instructions?

A. To make sure bias is reduced by standardizing how the questions are asked

B. To make sure that telephone numbers match respondents, such as voters

C. So interviewers show up to work on time

D. - A. To make sure bias is reduced by standardizing how the questions are asked

- What basic analysis can you do in a spreadsheet?

A. Percentage

B. Average

C. Median

D. All of the above - D. All of the above

- Cluster sampling uses…

A. Groups of questions rather than an entire questionnaire

B. Groups of telephone numbers, but not locations or addresses

C. Groups of the population for the frame rather than individual members

D. Multiple research designs - C. Groups of the population for the frame rather than individual members

- What is contact rate?

A. Number of interviews completed divided by the total number of units in the sample

B. Number of interviews completed divided by the number of legitimate units in the sample

C. Number of interviews completed divided by the elig - D. None of the above

- Why use a sample?

A. So you can be confident that your results are close to the true value in the population

B. It is more time and cost effective than a census for large populations

C. It accurately reflects opinions in the population

D. All of the - D. All of the above

- If you use volunteers what do you need to bear in mind?

A. That they are representative of the overall population

B. That their answers cannot be generalized to a population

C. That you must check to see if you have a balance of males and females

D. - B. That their answers cannot be generalized to a population

- Significance tells you…

A. If you have a strong relation

B. Whether your sampling error for descriptive data is accurate enough

C. The probability that there is a relation between two variables

D. How much explained variance there is - C. The probability that there is a relation between two variables

- 19. What significance test is used if one variable is interval level of measurement and the other variable is ratio level?

A. t-test

B. χ2

C. t-test for correlation

D. r2 - C. t-test for correlation

- 20. The F-test for multiple means is used for…

A. Significance tests for means by subgroups where one variable has more than two categories

B. Significance tests for means by subgroups where one variable has only two categories

C. Significance test - A. Significance tests for means by subgroups where one variable has more than two categories

- 21. If 47% of people in a sample of 350 voters said they support the president’s economic policies, what can you conclude?

A. A majority supports the president

B. A minority supports the president

C. The study needs to be done over because the esti - D. You can’t tell whether a majority or a minority supports the president

- 22. If you use N=720, what is the SEmax?

A. ±1.3%

B. ±3.7%

C. ±4.8%

D. You can’t do SEmax until after the results are in - B. ±3.7%

- 23. What is the SEmax if all 1450 students at Whitman College get a questionnaire at registration?

A. ±10.1%

B. ±10.2%

C. ±9.5%

D. 0.0% - D. 0.0%

- 24. The graph that shows the relation between sample size and sampling error demonstrates that…

A. Bigger samples are always better because they have lower sampling error and lower cost

B. Sample sizes of about 300 are the best combination of cost a - B. Sample sizes of about 300 are the best combination of cost and accuracy

- 25. Stratified random samples are useful for…

A. Small populations

B. Short interviews

C. Doing personal interviews

D. Making sure the sample distribution closely matches the population distribution of important demographics - D. Making sure the sample distribution closely matches the population distribution of important demographics

- 26. Which of the following is random?

A. Rate people on a scale of 1 to 10

B. Having people pick a number from 1 to 10

C. Putting your finger in the middle of a phone book without looking

D. None of the above - D. None of the above

- 27. Why use a timeline?

A. To make sure you have selected the appropriate research design

B. To make sure the study is well organized

C. To make sure the sample is random

D. To make estimates more precise for a fixed sample size - B. To make sure the study is well organized

- What significance test is used if one variable is interval level of measurement and the other variable is ratio level?

A. t-test

B. χ-squared

C. t-test for correlation

D. r-squared - C. t-test for correlation

- The F-test for multiple means is used for…

A. Significance tests for means by subgroups where one variable has more than two categories

B. Significance tests for means by subgroups where one variable has only two categories

C. Significance tests fo - A. Significance tests for means by subgroups where one variable has more than two categories

- If 47% of people in a sample of 350 voters said they support the president’s economic policies, what can you conclude?

A. A majority supports the president

B. A minority supports the president

C. The study needs to be done over because the estimate - D. You can’t tell whether a majority or a minority supports the president

- If you use N=720, what is the SEmax?

A. ±1.3%

B. ±3.7%

C. ±4.8%

D. You can’t do SEmax until after the results are in - B. ±3.7%

- What is the SEmax if all 1450 students at Whitman College get a questionnaire at registration?

A. ±10.1%

B. ±10.2%

C. ±9.5%

D. 0.0% - D. 0.0%

- The graph that shows the relation between sample size and sampling error demonstrates that…

A. Bigger samples are always better because they have lower sampling error and lower cost

B. Sample sizes of about 300 are the best combination of cost and a - B. Sample sizes of about 300 are the best combination of cost and accuracy

- Stratified random samples are useful for…

A. Small populations

B. Having a even distribution of males & females in the sample

C. Making sure short interviews contain as many questions as possible

D. Doing personal interviews - B. Having a even distribution of males & females in the sample

- Which of the following is random?

A. Rate people on a scale of 1 to 10

B. Having people pick a number from 1 to 10

C. Putting your finger in the middle of a page of random numbers

D. None of the above - C. Putting your finger in the middle of a page of random numbers

- Why use a timeline?

A. To make sure you have selected the appropriate research design

B. To make sure the study is well organized

C. To make sure the sample is random

D. To make estimates more precise for a fixed sample size - B. To make sure the study is well organized

- 28. Most scales are assumed to be at least which level of measurement?

A. Nominal

B. Ordinal

C. Interval

D. Ratio - C. Interval

- 29. Why do you need to monitor a research project you have commissioned?

A. Honest people make mistakes

B. To make sure the results, the estimates, are what you want

C. To prevent competitors from doing research

D. To allow local police to know you - A. Honest people make mistakes

- 30. What should be pretested?

A. Questionnaires

B. Lighting and audio in a focus group room

C. Procedures for an experiment

D. All of the above - D. All of the above

- 31. If you are willing to accept ±1.5% sampling error how big should the sample size be?

A. 4267

B. 6400

C. 640

D. 652 - A. 4267

- 32. A timeline should include…

A. When the report should be done

B. When pre-tests should be done

C. When data collection should occur

D. All of the above - D. All of the above

- 33. In order to show whether or not there is racial profiling by traffic officers, what would be the best evidence?

A. The percent of minorities who are stopped

B. The percent of minorities who are stopped compared with the percent of others who are s - C. The percent of minorities who are stopped compared with the percent of minorities who commit traffic offenses

- 34. If you have the following research question what analysis will you do? RQ1: Who is more likely to buy a car in the next 12 months, older people or younger people? “Likelihood to buy a car” is measured as Yes, will buy; No, will not. Age is measure
- B. Relational analysis

- 35. If you have the following research question what analysis will you do? RQ1: Who is more likely to buy a car in the next 12 months, older people or younger people? “Likelihood to buy a car” is measured as Yes, will buy; No, will not. Age is measure
- B. Means by subgroups

- 36. If your completion rate is expected to be 45%, how many phone numbers should you select to end up with N=475?

A. 1056

B. 9474

C. 758

D. None of the above will get 475 completed - A. 1056

- Most scales are assumed to be at least which level of measurement?

A. Ratio

B. Interval

C. Ordinal

D. Nominal - B. Interval

- Why do you need to monitor a research project you have commissioned?

A. Honest people make mistakes

B. There are dishonest researchers

C. People do a better job if they know they are monitored

D. All of the above - D. All of the above

- What should be pretested?

A. Questionnaires

B. Lighting and audio in a focus group room

C. Procedures for an experiment

D. All of the above - D. all of the above

- If you are willing to accept ±4.5% sampling error how big should the sample size be?

A. 4752

B. 453

C. 475

D. 494 - C. 475

- Interviewer instructions and other standardizations are used for…

A. Both qualitative studies and quantitative studies

B. Only qualitative studies

C. Only quantitative studies

D. Only studies that use random samples - A. Both qualitative studies and quantitative studies

- In order to show whether or not there is racial profiling by traffic officers, what would be the best evidence?

A. The percent of minorities who are stopped

B. The percent of minorities who are stopped compared with the percent of others who are stopp - C. The percent of minorities who are stopped compared with the percent of minorities who commit traffic offenses

- If you have the following research question what analysis will you do? RQ1: Who is more likely to buy a car in the next 12 months, older people or younger people? “Likelihood to buy a car” is measured as Yes, will buy; No, will not. Age is measured as
- B. Relational analysis

- If your completion rate is expected to be 56%, how many phone numbers should you select to end up with N=365?

A. 1056

B. 652

C. 671

D. None of the above will get 475 completed - B. 652

- 37. Why is distribution important?

A. You can have different averages for the same group

B. Your strategies may be different depending on the average even though the distribution is the same

C. Your strategies may be different for different groups ev - C. Your strategies may be different for different groups even though the averages for the groups are the same

- 38. Based on a sample of 1500 U.S. voters, 52% of Democrats say they would vote for Hillary Clinton. What does that mean?

A. A majority of voters plan to vote for Hillary

B. A majority of Democrats plan to vote for Hillary

C. A minority of Democrats - D. None of the above

- 39. What are potential problems with RDD?

A. Too many unlisted numbers will be in the sample

B. You will get unusable numbers, like disconnected numbers

C. You can’t get all telephone numbers in a frame

D. All of the above - B. You will get unusable numbers, like disconnected numbers

- 40. What does it mean to have a negative relation between number of items recalled from an ad and hours of TV viewed on the average day?

A. As number of items recalled from an ad increases hours of TV viewed increases

B. As number of items recalled fr - B. As number of items recalled from an ad increases hours of TV viewed decreases

- 41. Why should you use probes?

A. To reduce the number of open-ended comments you get

B. To help understand why an open-ended comment is made

C. To reduce the size of the questionnaire

D. To allow direct magnitude estimates - B. To help understand why an open-ended comment is made

- 42. If you get a t-test result of p=.004 what does that mean?

A. There is no relation because it is too close to zero

B. There is a positive relation

C. There is a high probability that the tested relation exists in the population

D. There is a low - C. There is a high probability that the tested relation exists in the population

- 43. Why is there explained variance?

A. Because you need to know how strong a relation is

B. So you know which category contains the highest percentage

C. Because averages are not always correct

D. So you know if your scale is a real number or not - A. Because you need to know how strong a relation is

- 44. What are the three basic types of analysis?

A. Descriptive research questions, analytical research questions and causal research questions

B. Descriptions of the sample, descriptive analysis, and relational analysis

C. Percentages, frequencies a - B. Descriptions of the sample, descriptive analysis, and relational analysis

- 45. Which explained variance estimate would you use for the relation between race (Hispanic, white, etc.) and company image (scale score)?

A. η2

B. C

C. r2

D. Any of the above will work - A. η2

- 46. What does p
- 47. If you wanted to make sure there was proportional distribution of races in a sample compared to race distribution in the population, which sampling method would you use?

A. Simple random

B. Systematic random

C. Snowball

D. Stratified

- 48. Wanting to contact residences, you had an original sample of 400 addresses. There were 300 completed residential interviews, 25 refusals, 30 no answers, and 45 businesses and other junk. What was the response rate?

A. 86.2%

B. 75.1%

C. 84.5%

D. - C. 84.5%

- 49. In the table of correlations, above, which is the strongest relation?

A. .32

B. -.56

C. .03

D. Can’t tell from this table - A. .32

- 50. What should be pretested?

A. The questionnaire

B. Procedures like how an experiment will be conducted

C. The lighting and other equipment, if used

D. All of the above - D. All of the above

- Why is distribution important?

A. Your strategies may be different for different groups even though the averages for the groups are the same

B. Your strategies may be different depending on the average even though the distribution is the same

C. You - A. Your strategies may be different for different groups even though the averages for the groups are the same

- Based on a sample of 1500 U.S. voters, 54% of voters say they would vote for Hillary Clinton. What does that mean?

A. A majority of voters plan to vote for Hillary

B. A majority of Democrats plan to vote for Hillary

C. A minority of Democrats plan to - A. A majority of voters plan to vote for Hillary

- What are potential problems with RDD?

A. Unlisted numbers will not be in the sample

B. You will only get residential numbers, so you can’t get business phone numbers

C. You can’t get all possible telephone numbers in a frame

D. None of the above - D. None of the above

- What does it mean to have a negative relation between number of items recalled from an ad and hours of TV viewed on the average day?

A. As number of items recalled from an ad increases hours of TV viewed increases

B. As number of items recalled from a - B. As number of items recalled from an ad increases hours of TV viewed decreases

- Why should you use probes?

A. To reduce the number of open-ended comments you get

B. To allow direct magnitude estimates

C. To reduce the size of the questionnaire

D. To help understand why an open-ended comment is made - D. To help understand why an open-ended comment is made

- If you get a t-test result of p=.900 what does that mean?

A. There is a relation because it is close to 1.0

B. There is a positive relation

C. There is a high probability that the tested relation exists in the population

D. There is a low probabilit - D. There is a low probability that the tested relation exists in the population

- What is a frequency?

A. The number of people who gave the same response

B. The percentage of people who gave the same response

C. The number of times a question is used in a survey

D. The number of times the same category is used in a questionnaire - A. The number of people who gave the same response

- What are the three basic types of analysis?

A. Descriptive research questions, analytical research questions and causal research questions

B. Descriptions of the sample, descriptive analysis, and relational analysis

C. Percentages, frequencies and co - B. Descriptions of the sample, descriptive analysis, and relational analysis

- Which explained variance estimate would you use for the relation between race (Hispanic, white, etc.) and company image (scale score)?

A. χ-squared

B. C

C. r-squared

D. None of the above will work - D. None of the above will work

- Richard Harket, VP of Coleman Research in Radio In Search of Excellence, NAB, said “Research enables a radio station to speak to a few hundred listeners with the confidence that the opinions of these reflect the opinions of all listeners in equal propor
- C. Sample

- If you wanted to make sure there was proportional distribution of races in a sample compared to race distribution in the population, which sampling method would you use?

A. Simple random

B. Systematic random

C. Snowball

D. Stratified - D. Stratified

- Wanting to contact residences, you had an original sample of 300 addresses. There were 200 completed residential interviews, 25 refusals, 30 no answers, and 45 businesses and other junk. What was the response rate?

A. 84.5%

B. 78.4%

C. 66.7%

D. None - B. 78.4%

- In the table of correlations, above, which is the strongest relation?

A. .32

B. -.56

C. .63

D. Can’t tell from this table - C. .63

- Why would you compute maximum sampling error?

A. To see if the sample size is big enough to produce the accuracy you need

B. To see if the sample size is big enough to produce a sampling error above 95%

C. To make sure the sample is random

D. To fin - A. To see if the sample size is big enough to produce the accuracy you need

- How much area of the normal curve falls between –1sd and +1sd?
- 68%

- what does confidence level tell us?
- How confident we are that the sample represents the population

- If one poll says Obama leads Clinton 49% to 47% and another says Clinton leads Obama 50% to 46%, how do you know which one is correct?
- they could both be correct

- the normal curve is...
- Not the only curve use by social scientists. Others include “chi-square,” “F,” “t,” and “u”

- which of the following is true about the selection of sample sizes
- To cut sampling error in half you must quadruple the sample size

- In order to put verbatim responses into categories, what process should you follow?
- Have at least two coders put responses into categories and check for disagreement between the coders

- how do you standardize?
- Have a list of “rules” for interviewers so all interview procedures are the same

- significance tells you...
- How probable it is that you found a real relation

- The area under the normal curve between –1.96sd and + 1.96sd is important to social scientists. Why?
- It is the basis for saying there is a high probability that our sample represents the population

- Which of the following is true about sample size?
- all the above are true

- why use a sample?
- It is more time and cost effective than a census for large populations

- What relational analysis can you do if one variable is nominal and the other is ordinal?
- crosstabulation

- If you use a volunteer sample…
- you cannot do significance tests

- what is refusal rate?
- Number of interviews refused divided by the number of legitimate units in the sample

- If you use N=380, what is the SEmax?
- ±5.0%

- If 46% of people in a sample of 850 U. S. voters said they support the president’s economic policies, what can you conclude?
- a minority supports the president

- If there are 17695 WSU-Pullman students and 395 Communication majors, and you sample 93 WSU-Pullman students to fill it an online questionnaire, what is the SEmax?
- ±10.2%

- What significance test is used if one variable is interval level of measurement and the other variable is ratio level?
- t-test for correlation

- Which of the following is a population?
- All ads on Spokane radio in the last month

- what is "random"?
- The equal chance for each member in a population to be selected into a sample

- If you use N=720, and 56% of the sample is female what is the SE for 56%?
- ±3.6%

- What is the SEmax if all 2500 employees at Spokane Bearing Inc. get a questionnaire about internal communication?
- ±0.0%

- The graph showing the relation between sample size and sampling error demonstrates that…
- Sample sizes of about 300 are the best combination of cost and accuracy

- nonrandom samples are...
- Are not representative

- Which of the following is random?
- Putting your finger in the middle of a page of random numbers

- why use a timeline?
- To make sure the study is well organized

- If the election is close and you must have no more than ±1.5% sampling error how big should the sample size be?
- 4267

- Why do you need to monitor a research project you have commissioned?
- all the above

- What should be pretested?
- all the above

- If you are willing to accept ±4.5% sampling error how big should the sample size be?
- 475

- Interviewer instructions and other standardizations are used for…
- Both qualitative studies and quantitative studies

- In order to show that your campaign caused sales to increase what do you need to do?
- None of the above will show that your campaign caused sales to increase

- If you have the following research question what analysis will you do? RQ1: What percent of males are likely to buy a car in the next 12 months?
- descriptive analysis

- If you have the following research question what analysis will you do? RQ1 What percent of males are likely to buy a car in the next 12 months?
- none of the above

- If your completion rate for an e-mail crosssectional survey is expected to be 46%, how many phone numbers should you select to end up with N=300?
- none of the above will get 300 completed

- Why is distribution important?
- Your strategies may be different for different groups even though the averages for the groups are the same

- What is proprietary research?
- Research that you pay for and that you want to keep secret from competitors

- What are potential problems with RDD if you want to call residences?
- You get business phone numbers mixed in with residential numbers

- What does it mean to have a positive relation between number of items recalled from an ad and hours of TV viewed on the average day?
- As number of items recalled from an ad increases hours of TV viewed increases

- Why should you use probes?
- To help understand why an open-ended comment is made

- If you get a t-test result of p=.06 what does that mean?
- There is not a high probability that the tested relation exists in the population

- what is frequency?
- The number of people who gave the same response

- What are the three basic types of analysis?
- Descriptions of the sample, descriptive analysis, and relational analysis

- Which explained variance estimate would you use for the relation between income (dollars) and company image (scale score)?
- r-squared

- 1. Why is –1.96sd to +1.96sd important to social scientists?

A. It shows that there is 95% explained variance for a percentage

B. It shows that we’re 95% sure results from a sample are accurate estimates within sampling error

C. It is used to cal - B. It shows that we’re 95% sure results from a sample are accurate estimates within sampling error

- 2. What does confidence level tell us?

A. How sure we are that the results from a sample have no error

B. The sampling error

C. The probability that the sample represents the population

D. It’s the same as confidence interval - C. The probability that the sample represents the population

- 3. One of the polls using N=650 on 11/03/08 showed Obama getting 49% of the vote. He actually got 52%. Was the poll wrong?

A. Yes, the poll was wrong because the poll showed less than 50%

B. Yes, the poll was wrong because the sample was too big

C. - D. No, the poll was not wrong

- 4. How can you bias a scale?

A. Failing to have a mix of positive and negative items

B. Using only language you have created

C. Failing to pre-test the scale

D. All of the above - D. All of the above

- 5. Which of the following is true about the selection of sample sizes?

A. To cut sampling error in half you must double the sample size

B. To cut sampling error in half you must quadruple the sample size

C. To cut sampling error in half you must cut - B. To cut sampling error in half you must quadruple the sample size

- 6. If you get a verbatim response like “billing blood suckers” what should you do?

A. Start planning a PR campaign to counter the negative effect

B. Consider a follow-up future study to see if others have similar comments

C. Don’t worry about v - B. Consider a follow-up future study to see if others have similar comments

- 7. How do you standardize?

A. Never ask income in real dollars because too many people refuse

B. Have a list of “rules” for interviewers so all interview procedures are the same

C. Select only one analysis technique (e.g., crosstabulation) before - B. Have a list of “rules” for interviewers so all interview procedures are the same

- 8. Significance tells you…

A. How well your variable explains a change in another variable

B. How probable it is that you found a real relation

C. Whether one relation explains more than another relation

D. All of the above - B. How probable it is that you found a real relation

- 9. How do you figure sampling error for a convenience sample?

A. Use the SE formula

B. Only use the SEmax formula

C. You do it the same as any other sample

D. None of the above - D. None of the above

- What is refusal rate?

A. Number of interviews refused divided by the total number of units in the sample

B. Number of interviews refused divided by N

C. Number of interviews refused divided by the number of interviews completed

D. None of the above - D. None of the above

- Is it possible to sample content such as competitors’ web pages?

A. Yes

B. Yes, but only if you can identify the people associated with the web pages

C. No

D. No, you can only do a census of content - A. Yes

- Based on results from a sample of 650 in your target population, 57.0% have high trust in your company and 51.5% have high trust in your competitor. Does a majority have high trust in your company?

A. Yes

B. No

C. You can’t tell whether a majority - A. Yes

- Assume you are a legislator and you are given a research report that shows 89% of rapists consume pornography at least once per week. What does that show?

A. Pornography causes rape

B. There is a relation between pornography and rape

C. Pornography i - D. None of the above

- If you get a t-test result of p=.04 what does that mean?

A. There is no relation because the correlation is close to 0.0

B. There is a positive relation

C. There is a high probability that the tested relation exists in the population

D. There is not - C. There is a high probability that the tested relation exists in the population

- Which of the following is part of the Respondent’s Rights in the “Focus Group Bill of Rights?”

A. To be able to change any procedures at any time in the study

B. To be told whether there will be video or audio taping

C. To receive a discussion - B. To be told whether there will be video or audio taping

- 10. Which of the following is true about sample size?

A. As sample size increases from 100 to 200, sampling error falls from 9.8% to 4.9%

B. As sample size increases, sampling error will eventually decrease to zero

C. Increasing the sample size beyon - D. None of the above are true

- 11. Which of the following is true about sample size?

A. As sample size increases from 400 to 1600, sampling error is cut in half

B. As sample size increases, sampling error decreases

C. Sample sizes of about 300 are a good compromise between accurac - D. All of the above are true

- 12. What is a semantic differential?

A. A technique for making sure your scale is 95% accurate

B. A very precise way to scale, often using 0 to 100%

C. A scaling technique that uses extreme positions (polar adjectives)

D. All of the above - C. A scaling technique that uses extreme positions (polar adjectives)

- 13. What relational analysis can you do if one variable is a ratio level of measurement and the other is interval?

A. t-test

B. Crosstabulation

C. Means by subgroups

D. None of the above - D. None of the above

- 14. The PR Practitioner Roles Scale example in the book has eight items, four for Communication Manager Role and four for Technician Role. You should…

A. Calculate a total scale score on the eight items

B. Calculate separate scale scores, one for Co - B. Calculate separate scale scores, one for Communication Manager and one for Technician

- 15. What is refusal rate?

A. Number of interviews refused divided by the total number of units in the sample

B. Number of interviews refused divided by N

C. Number of interviews refused divided by the number of interviews completed

D. None of the ab - D. None of the above

- 16. If you use N=720, what is the SEmax?

A. ±2.5%

B. ±3.4%

C. ±3.7%

D. ±4.2% - C. ±3.7%

- 17. If 48% of people in a sample of 2400 U. S. voters said they support the new president’s economic policies, what can you conclude?

A. A majority supports the new president

B. A minority supports the new president

C. You can’t tell without know - B. A minority supports the new president

- 18. If there are 17695 WSU-Pullman students and 395 Communication majors, and you sample all 395 Communication majors who fill out an online questionnaire, and 13.4% say they voted in both the primary and the general election, what can you conclude?

A. - D. None of the above

- 19. Is it possible to sample content such as competitors’ web pages?

A. Yes

B. Yes, but only if you can identify the people associated with the web pages

C. No

D. No, you can only do a census of content - A. Yes

- 20. Why would you do a survey instead of interviewing all of your employees even if that population is only 100?

A. Sampling error is too big for a census of 100

B. You can’t calculate sampling error for small samples

C. You don’t want to sensit - C. You don’t want to sensitize all employees to the topics in the survey

- 21. Why is it important to have interviewer instructions?

A. To make sure bias is reduced by standardizing how the questions are asked

B. To make sure that telephone numbers match respondents, such as voters

C. So interviewers show up to work on tim - A. To make sure bias is reduced by standardizing how the questions are asked

- 22. If you use N=450, and 26% of the sample is over 45 years old what is the SE for 26%?

A. ±3.9%

B. ±4.1%

C. ±4.3%

D. ±2.6% - B. ±4.1%

- 23. Which of the following is a frame?

A. Males, 18-35 years old

B. County clerk voter registration list

C. A sample of 100 television stations

D. All of the above - B. County clerk voter registration list

- 24. Bigger samples are always better is…

A. True because sampling error always goes down as sample size goes up

B. Not true because extremely large samples cost more without a corresponding extreme reduction in sampling error

C. True because you ca - B. Not true because extremely large samples cost more without a corresponding extreme reduction in sampling error

- 25. If you have a frame of 3000 professors at WSU and you want a sample of 200 using systematic-random sampling, what is the skip interval?

A. 15

B. 200

C. ±6.9%

D. ±7.4% - A. 15

- 26. What is a snowball sample?

A. No chance in hell of being in the sample

B. Using people you initially interview to recommend others to interview

C. Adding sample members randomly to make sure you reach the desired sample size

D. Asking the sample - B. Using people you initially interview to recommend others to interview

- –1.96sd to + 1.96sd on the normal curve is important to social scientists. Why?

A. The area between –1.96sd to + 1.96sd is the probability (.95) that random samples represent the population

B. Between –1.96sd to + 1.96sd is the confidence interv - D. All of the above

- Which explained variance estimate would you use for the relation between income (dollars in the previous year) and ad recall (number of items remembered from an ad)?

A. χ²

B. C

C. r²

D. None of the above will work - C. r²

- What makes it likely that your sample represents the population?

A. The sample demographics are close to known population demographics

B. The frame was good (e.g., using random numbers instead of a telephone book)

C. Each member of the frame had an e - D. All of the above

- If you use N=1380, what is the SEmax?

A. ±2.6%

B. ±5.0%

C. ±5.8%

D. ±6.5% - A. ±2.6%

- Based on a sample of 600 diaries, if your rating is 16 (a rating is a percent of the TV viewing households) and your competitor’s rating is 10, what can you conclude?

A. You are ahead of your competitor

B. You are behind your competitor

C. You can - A. You are ahead of your competitor

- Why is it OK to put your finger in the middle of a random number table (without looking) to select a random number?

A. It’s not OK because it’s not random

B. Each number had an equal chance of being where you put your finger

C. It’s OK only if - B. Each number had an equal chance of being where you put your finger

- 28. If you have a very competitive media market and you want no more than ±1.0 rating point sampling error (a rating point=1%) how big should the sample size be?

A. 96

B. 967

C. 9600

D. 12300 - C. 9600

- 29. Why do you need to monitor a research project you have commissioned?

A. Honest people make mistakes

B. There are dishonest researchers

C. People do a better job if they know they are monitored

D. All of the above - D. All of the above

- 30. What should be pretested?

A. Questionnaires

B. Lighting and audio in a focus group room

C. Procedures for an experiment

D. All of the above - D. All of the above

- 31. Based on results from a sample of 300 in your target population, 22.0% have high trust in your company and 13.5% have high trust in your competitor. Does your target have higher trust in you vs. your competitor?

A. Yes

B. No

C. Can’t tell witho - B. No

- 32. Interviewer instructions and other standardizations are used for…

A. Both qualitative studies and quantitative studies

B. Only qualitative studies

C. Only quantitative studies

D. Only studies that use random samples - A. Both qualitative studies and quantitative studies

- 33. What evidence should you use to help show that your campaign caused sales to increase?

A. Time order, your campaign preceded sales increase

B. Use a census to show that your campaign was the only variable that increased sales

C. Figure sampling e - A. Time order, your campaign preceded sales increase

- 34. If you have the following research question what analysis will you do? RQ1: Is the percent of males likely to buy a car in the next 12 months greater than the percent of females likely to buy a car in the next 12 months?

A. Descriptive analysis

B. - B. Relational analysis

- 35. If you have the following research question what analysis will you do? RQ1: Is the percent of males likely to buy a car in the next 12 months greater than the percent of females likely to buy a car in the next 12 months?

A. Crosstabulation

B. r-sq - A. Crosstabulation

- 36. If your completion rate for an e-mail crosssectional survey is expected to be 54%, how many phone numbers should you select to end up with N=300?

A. 547

B. 556

C. 553

D. None of the above will get 300 completed - B. 556

- Significance tells you…

A. If you have a strong relation

B. Whether your sampling error for descriptive data is accurate enough

C. The probability that there is a relation between two variables

D. How to be creative - C. The probability that there is a relation between two variables

- Explained variance tells you…

A. If you have a strong relation

B. Whether your sampling error for descriptive data is accurate enough

C. The probability that there is a relation between two variables

D. How to be creative - A. If you have a strong relation

- If you are interested in the relation between two Likert-type scales which analysis do you do?

A. Descriptive

B. Means by subgroups

C. Correlation

D. Crosstabulation - C. Correlation

- Why use a sample?

A. So you can get exact statistics for a population

B. Even small samples (e.g., N=20) have low sampling error (under 10%)

C. You don’t have to worry about testing effects

D. None of the above - D. None of the above

- Nonrandom samples…

A. Are representative

B. Are useful for estimating population numbers (e.g., average number of technologies owned)

C. Are used in surveys where accuracy is important

D. None of the above - D. None of the above

- How can you deal with verbatim responses?

A. You can just list them

B. You should always put them into categories for quantitative analysis

C. Probe by using closed-ended responses (such as “good, fair, poor”)

D. None of the above - A. You can just list them

- What is the SEmax if a sample of 1450 students at Whitman College get a questionnaire at registration?

A. ±4.1%

B. ±3.2%

C. ±2.6%

D. 0.0% - C. ±2.6%

- 37. Which of the following is part of the Client’s Rights in the “Focus Group Bill of Rights?”

A. To be able to change any procedures at any time in the study

B. To reduce payment proportional to the number of people who do not arrive at the foc - C. To receive a discussion guide well in advance of the groups

- 38. What is verification?

A. Agreeing to the timeline with the consultant or the research director

B. Checking references of a consultant to verify their credentials

C. Pre-testing the questionnaire

D. Contacting a small sample of those allegedly - D. Contacting a small sample of those allegedly interviewed to make sure interviews were conducted

- 39. What are potential problems with RDD if you want to call residences?

A. Unlisted numbers will not be in the sample

B. You get business phone numbers mixed in with residential numbers

C. You can’t get all possible telephone numbers in a frame

D - B. You get business phone numbers mixed in with residential numbers

- 40. What does it mean to have no relation between number of items recalled from an ad and hours of TV viewed on the average day?

A. As number of items recalled from an ad increases hours of TV viewed increases

B. As number of items recalled from an ad - C. As number of items recalled from an ad increases hours of TV viewed stays the same

- 41. Why should you use probes?

A. To reduce the number of open-ended comments you get

B. To allow direct magnitude estimates

C. To reduce the size of the questionnaire

D. To help understand why an open-ended comment is made - D. To help understand why an open-ended comment is made

- 42. If you get a t-test result of p=.04 what does that mean?

A. There is no relation because the correlation is close to 0.0

B. There is a positive relation

C. There is a high probability that the tested relation exists in the population

D. There is - C. There is a high probability that the tested relation exists in the population

- 43. If your sample had 44% males but the population had 51% males is your sample OK?

A. Yes, because the difference is small

B. No, because the difference is too big

C. You can only tell by comparing the difference to the sampling error

D. It is ne - C. You can only tell by comparing the difference to the sampling error

- 45. Which explained variance estimate would you use for the relation between income (categories of low, medium and high) and ad recall (yes, saw the ad vs no, did not see the ad)?

A. χ-squared

B. C

C. r-squared

D. None of the above will work - B. C

- 46. Why do a significance test?

A. To show if there is a high probability that a relation exists

B. To determine which relations are strongest

C. To find out how important a relation is for your campaign

D. All of the above - A. To show if there is a high probability that a relation exists

- 47. What does the “Note on Cause-and-Effect” in the book and the lecture say about significance and explained variance?

A. If both are strong then there is cause and effect

B. If you have significance and strong relations it is a good start in the - B. If you have significance and strong relations it is a good start in the right direction

- 48. Why would you compute maximum sampling error?

A. To see if the sample size is big enough to produce the accuracy you need

B. To see if the sample size is big enough to produce a sampling error above 95%

C. To make sure the sample is random

D. T - A. To see if the sample size is big enough to produce the accuracy you need

- 49. In the table of correlations, above, which is the strongest relation?

A. .82

B. -.56

C. .63

D. Can’t tell from this table - A. .82

- 50. What is the explained variance estimate for -.63 from the table above?

A. -.40

B. +.40

C. It is .31

D. It is -.31 - B. +.40

- If you must have no more than ±6% sampling error how big should the sample size be?

A. 467

B. 267

C. 458

D. 258 - B. 267

- If you have the following research question what analysis will you do? RQ1: Who is more likely to buy a car in the next 12 months, people between 18-34 years of age or those older than 34?

A. Descriptive analysis

B. Relational analysis

C. Ratio analy - B. Relational analysis

- After you have tested a new scale what do you do?

A. Use the test results to estimate population values

B. Fix any problems and test it again

C. Figure out the sampling error

D. Change the scale to direct magnitude estimates - B. Fix any problems and test it again

- Based on a sample of 1500 U.S. voters, 54% say they support the president’s stimulus package. What does that mean?

A. A majority of voters supports the president’s stimulus package

B. A majority of citizens supports the president’s stimulus pack - A. A majority of voters supports the president’s stimulus package

- If your completion rate is expected to be 65%, how many phone numbers should you select to end up with N=475?

A. 642

B. 894

C. 731

D. None of the above will get 475 completed - C. 731

- Which explained variance estimate would you use for the relation between race (Hispanic, white, etc.) and company image (scale score)?

A. η²

B. C

C. r²

D. Any of the above will work - A. η²

- If you use N=120, what is the SEmax?

A. ±8.9%

B. ±9.7%

C. ±7.4%

D. None of the above - A. ±8.9%

- Which of the following is true about sample size?

A. As sample size increases from 100 to 200, sampling error falls from 9.8% to 6.9%

B. As sample size increases, sampling error decreases

C. Increasing the sample size from 100 to 400 cuts the sampli - D. All of the above are true

- What is verification?

A. Pretesting questionnaires

B. Calling some of the phone number of completed interviews to make sure interviews were conducted

C. Making sure procedures in an experiment run smoothly

D. Double checking that sampling error calc - B. Calling some of the phone number of completed interviews to make sure interviews were conducted

- In the table of correlations, above, which is the strongest relation?

A. .32

B. -.56

C. .33

D. Can’t tell from this table - B. -.56

- What is the explained variance estimate for .33 from the table above?

A. .66

B. .99

C. It is .11

D. It is -.11 - C. It is .11