Question 1

The field of statistics concerned about finding and correcting problems within regression models is called:
A) regression troubleshooting
B) regression diagnostics
C) regression medicine
D) regression health

Accepted Answer

The field concerned with finding and correcting issues within regression models is called regression diagnostics. Option A is not a commonly used term, option C and D are not appropriate choices as the term 'medicine' and 'health' are not used in statistics terminology.

Question 2

Which of the following can offer signs that there is a problematic outlier in your data?
A) the R-squared value
B) the statistical significance
C) the standard deviation
D) all of the above

Accepted Answer

All of the options listed can offer signs of a problematic outlier in data. An outlier can significantly affect the R-squared value by distorting the overall fit of a regression model. It can also impact the statistical significance of results by influencing the mean and variance in unpredictable ways. Additionally, an outlier can greatly increase the standard deviation, indicating a wider spread of data points around the mean than would be expected in a normal distribution.

Question 3

You run some examples using your regression results, and they just don't make sense given what you know about the world (i.e. they say that as education rises, income goes down). You are putting your results through the:
A) eyeball test
B) sniff test
C) logic test
D) sense test

Accepted Answer

The question describes a scenario where the results of a regression analysis are counter-intuitive and do not align with the researcher's prior knowledge about the world. In this case, the best option is to subject the results to a "sniff test" to see if they "smell right." This involves checking the results for plausibility and coherence with prior knowledge. The other options, such as the "eyeball test" and "sense test," also involve checking the results for validity but may not be as appropriate in this specific scenario. The "logic test" may not be as relevant since the regression analysis is likely already based on a logical model.

Question 4

If two of your independent variables are highly correlated with one another, you might have a problem with:
A) nonlinearity
B) causality
C) multicollinearity
D) explained variation

Accepted Answer

Multicollinearity occurs when two or more independent variables in a regression model are highly correlated with one another. This can cause problems such as unstable coefficients and difficulty in interpreting the significance of individual variables.

Question 5

Concerned about multicollinearity between two of your independent variables, you look at the Tolerance statistic, and it is .98. You should be:
A) not concerned at all
B) a little concerned
C) somewhat concerned
D) very concerned

Accepted Answer

Tolerance values close to 1 indicate that multicollinearity is not a concern. A value of .98 suggests that the independent variables are not highly correlated, hence there is no need to be concerned about multicollarity.

Question 6

Concerned about multicollinearity between two of your independent variables, you look at the Tolerance statistic, and it is .25. You should be:
A) not concerned at all
B) a little concerned
C) somewhat concerned
D) very concerned

Accepted Answer

A Tolerance statistic of .25 indicates that 75% of the variance in one of the independent variables is shared with the other variables, suggesting a high degree of multicollinearity. Therefore, one should be very concerned about multicollinearity in the model.

Question 7

You have identified multicollinearity as a problem in your regression model. Which of the following is NOT a suggestion the textbook makes with regard to what to do?
A) remove one of the involved independent variables from the model
B) combined the involved variables into an index
C) create an interaction effect with the two involved variables
D) all of the above are suggestions the textbook makes.

Accepted Answer

In fact, the textbook suggests that creating an interaction effect may exacerbate the issue of multicollinearity rather than solve it. Therefore, it is not a recommended solution to the problem. The other two suggestions (removing one of the involved independent variables or combining the involved variables into an index) are both potential solutions to address multicollinearity in a regression model.

Question 8

We want to use a satisfaction variable (measured: very satisfied, somewhat satisfied, not very satisfied, not at all satisfied) as our dependent variable. Which technique might we want to use?
A) regular linear regression
B) logistic regression
C) ordered logistic regression
D) multinominal logistic regression

Accepted Answer

Since the satisfaction variable is measured on an ordered scale, we should use an ordered logistic regression. This model can account for the ordinal nature of the categories and allows for estimating the probabilities of being in each satisfaction category while controlling for other variables.

Question 9

We want to use a work status variable (measured: working fulltime, working part-time, unemployed, retired) as our dependent variable. Which technique might we want to use?
A) regular linear regression
B) logistic regression
C) ordered logistic regression
D) multinominal logistic regression

Accepted Answer

Multinomial logistic regression is appropriate for a dependent variable with more than two categories that do not have a natural order, such as the work status variable described.

Question 10

Tobit regression is most similar to which kind of regression?
A) regular linear regression
B) logistic regression
C) nonlinear regression
D) standardized regression

Accepted Answer

Tobit regression is a type of regression that is used when the dependent variable is censored. This means that the value of the dependent variable can only be observed within a certain range. Logistic regression is also used in cases where the dependent variable is binary or categorical, which is similar to the concept of censored variables. Therefore, Tobit regression is most similar to logistic regression.

Question 11

For which of the following variables might someone suggest using negative binomial regression?
A) number of marriages
B) age at first marriage
C) length of marriage
D) cost of wedding

Accepted Answer

Negative binomial regression is typically used for count data that has overdispersion, meaning that the variance is greater than the mean. The number of marriages is a count variable that could potentially exhibit overdispersion, therefore negative binomial regression would be an appropriate choice. The other variables listed (age at first marriage, length of marriage, and cost of wedding) are not counts and do not typically exhibit overdispersion, so negative binomial regression would not be the best choice for analyzing those variables.

Question 12

You have data from students, parents, teachers, and school principals. You might want to use which technique?
A) path analysis
B) poisson regression
C) multi-level modeling
D) negative binomial regression

Accepted Answer

Multi-level modeling would be the best choice as it allows for the analysis of data that is nested within different levels (e.g. students within classrooms, classrooms within schools), which is common when analyzing data from multiple sources within a school setting. This technique would allow for the examination of both within-group and between-group variability in the data. Path analysis may also be useful in analyzing relationships between variables, but may not account for the nested structure of the data. Poisson regression and negative binomial regression are more appropriate for count data, which may not be the primary type of data collected from these sources.

Question 13

The technique of taking many, many samples from within your sample, and calculating statistics from these samples, is called:
A) multisampling
B) multi-level modeling
C) repeated measures
D) bootstrapping

Accepted Answer

The technique described is called bootstrapping. Bootstrapping is a resampling technique where multiple samples are taken with replacement from the original sample, and statistics are calculated from each of these resampled datasets. It is useful for estimating the standard error of a statistic, constructing confidence intervals, and testing hypotheses.

Question 14

Which of these advanced techniques is similar to path analysis?
A) structural equation modeling
B) multi-level modeling
C) hazard modeling
D) runway modeling

Accepted Answer

Structural equation modeling, like path analysis, is used to examine relationships between variables and test theoretical models. It allows for the testing of multiple hypotheses simultaneously and can include both observed and latent variables. Multi-level modeling is used to analyze data with hierarchical structures and is not directly similar to path analysis. Hazard modeling is used to analyze time-to-event data and is also not directly similar to path analysis. Runway modeling is not a method commonly used in statistical analysis.

Question 15

If your research question involves time, such as figuring out when, if ever, something will happen, you might want to use which advanced technique?
A) structural equation modeling
B) multi-level modeling
C) negative binomial regression
D) hazard modeling

Accepted Answer

Hazard modeling is the best choice for research questions involving time, as it allows for the analysis of time-to-event data and can estimate the probability of an event occurring at a specific time. Structural equation modeling, multi-level modeling, and negative binomial regression are not specifically designed for time-based research questions.

Question 16

Your friend shows you some SPSS output for a regression mode she ran, and you notice that the amount of total variation seems extremely high, and even though none of the slopes are statistically significant, the R-squared value is .25. What might be going on here?

Accepted Answer

The answer of Your friend shows you some SPSS output...

Question 17

Your friend shows you some SPSS output for a regression model he ran, and you happen to notice a Tolerance statistic of .30. Tell your friend what problem he might have and describe the problem.

Accepted Answer

The answer of Your friend shows you some SPSS output...

Question 18

You want to study zoos: each separate animal, the zookeepers in charge of each type of animal, and the head of the zoos. Explain what advanced technique you might want to use.

Accepted Answer

The answer of You want to study zoos: each separate...

Question 19

Describe one way structural equation modeling is similar to path analysis, then describe one way it is different.

Accepted Answer

The answer of Describe one way structural equation modeling is...

Question 20

You want to study why and when students drop out of high school. Explain what advanced technique you might want to use.

Accepted Answer

The answer of You want to study why and when...

Quiz 16: Problems and Prospects: Regression Diagnostics, Advanced Techniques, and Where to Go Now

The field of statistics concerned about finding and correcting problems within regression models is called:

Which of the following can offer signs that there is a problematic outlier in your data?

You run some examples using your regression results, and they just don't make sense given what you know about the world (i.e. they say that as education rises, income goes down) . You are putting your results through the:

If two of your independent variables are highly correlated with one another, you might have a problem with:

Concerned about multicollinearity between two of your independent variables, you look at the Tolerance statistic, and it is .98. You should be:

Concerned about multicollinearity between two of your independent variables, you look at the Tolerance statistic, and it is .25. You should be:

You have identified multicollinearity as a problem in your regression model. Which of the following is NOT a suggestion the textbook makes with regard to what to do?

We want to use a satisfaction variable (measured: very satisfied, somewhat satisfied, not very satisfied, not at all satisfied) as our dependent variable. Which technique might we want to use?

We want to use a work status variable (measured: working fulltime, working part-time, unemployed, retired) as our dependent variable. Which technique might we want to use?

Tobit regression is most similar to which kind of regression?

For which of the following variables might someone suggest using negative binomial regression?

You have data from students, parents, teachers, and school principals. You might want to use which technique?

The technique of taking many, many samples from within your sample, and calculating statistics from these samples, is called:

Which of these advanced techniques is similar to path analysis?

If your research question involves time, such as figuring out when, if ever, something will happen, you might want to use which advanced technique?

Your friend shows you some SPSS output for a regression mode she ran, and you notice that the amount of total variation seems extremely high, and even though none of the slopes are statistically significant, the R-squared value is .25. What might be going on here?

Your friend shows you some SPSS output for a regression model he ran, and you happen to notice a Tolerance statistic of .30. Tell your friend what problem he might have and describe the problem.

You want to study zoos: each separate animal, the zookeepers in charge of each type of animal, and the head of the zoos. Explain what advanced technique you might want to use.

Describe one way structural equation modeling is similar to path analysis, then describe one way it is different.

You want to study why and when students drop out of high school. Explain what advanced technique you might want to use.