Deck 15: Multivariate and Advanced Quantitative Methods

Question 1

A third variable that is related to two other concepts/variables in a way that makes them appear to be related even in cases when they are not is called a(n)

A)inverse association.
B)exogenous variable.
C)interaction variable.
D)confounding variable.

Answer 1

D

Question 2

A body of research shows that people who have been divorced have a higher risk of divorce in any future marriage with another spouse.However,rather than divorce causing additional divorces,this seems to be more about other factors,such as how tolerant people are of divorce behavior.In this case,divorce attitudes would be a confounding factor because

A)more tolerant divorce attitudes are related both to first divorces and to later divorces.
B)people are almost never remarried to their ex-spouse,so the link is unlikely to be causal.
C)people's attitudes toward divorce may change after they've actually experienced it.
D)attitudes toward divorce may be related to first divorces,but not additional divorces.

Answer 2

A

Question 3

A team of researchers is studying the possible effect of parents taking child-rearing classes on children's behavioral well-being.When they test the relevant null hypothesis,they are assessing whether

A)sample data show that child behavior differed by whether parents took the classes.
B)parenting classes are actually the independent variable rather than the dependent variable.
C)the observed association between parenting classes and child behavior is the result of chance.
D)the observed association is actually explained by a confounding variable.

Answer 3

C

Question 4

If an observed difference is statistically significant,but isn't large enough to be justifiably used to affect social policies,this difference is understood to lack

A)randomization.
B)elaboration.
C)statistical power.
D)practical significance.

Question 5

If a study uses a p-value of 0.05 to determine statistical significance,the study's results only have a 1 in 20 chance of being due to random chance.If a researcher runs a large number of analyses,however,this increases the possibility that p-values

A)have no mathematical meaning with multiple analyses.
B)are at risk of being abused and selectively interpreted.
C)will be underestimated with smaller samples that require more analyses.
D)will be the same whether the sample is probabilistic or not.

Question 6

Arturo wants to use a probability sample to study the potential effects of health policies on infant mortality risk.To achieve this goal,the most important thing Arturo can do is to

A)use standardized questions in his survey.
B)get the largest sample he can acquire.
C)use randomization to select the sample.
D)only study populations with normally distributed mortality rates.

Question 7

Ben has used a random sample of 20,000 respondents to test the relationship between participating in a summer reading program and children's reading scores in the following academic year.He observes a modest difference in reading scores by program participation and is able to reject the null hypothesis with a p-value of 0.04.When Ben presents his findings,his colleagues question whether the difference is very meaningful.Their concerns are most likely related to the fact that

A)his sample might not be representative of the population of children in schools.
B)he did not consider possible moderating variables.
C)his sample was so large that he was almost certain to find statistical significance.
D)a p-value is a very unreliable test for assessing statistical significance.

Question 8

Bohai and Adal have administered a survey to a sample of college students to study the association between type of high school attended (public versus private)and social adjustment in college.They also considered the importance of students' socioeconomic background because that could affect the high school-college adjustment association.When all three variables are in the statistical model,the researchers observe a moderately larger adjustment score for students who attended public high schools,with a p-value of 0.03.These results tells Bohai and Adal that

A)there is only a small chance that the observed difference by high school is due to chance.
B)the null hypothesis can be rejected,but only for students who went to public high schools.
C)there is only a small chance that the sample is not representative of the college population.
D)socioeconomic background makes a 3% difference by high-school type.

Question 9

Janae has examined sample data on a possible association between age and use of online coupons,comparing those who are 65 years or older to those 64 years of age or younger.She observes a 9% difference in online coupon use by age,such that those under 65 in her sample use coupons more than those who are 65 or older.She has a p-value of 0.08,which tells her that

A)younger people do use online coupons more,but that the difference is small.
B)she can reject the null hypothesis regarding age and online coupon use.
C)there is really an 8% age difference rather than a 9% difference.
D)there is an 8% chance that the observed age difference is due to chance.

Question 10

Two separate studies examine the potential association between occupation type and voter participation.The first advertises online for participants and asks those who respond to tell the researchers in-depth accounts regarding their paid work and their voting behaviors for the last three years.The second study randomly selects 200 respondents and asks them to complete a standardized set of survey questions about their paid work and voting behaviors over the past five years.The first study observes a patterned relationship between work and voting,while the second study does not.The reason the second study's results are more valid is that the

A)first study is based on experimental data,which are less reliable.
B)second study is based on a probability sample.
C)second study is based on more years of data on voting behaviors.
D)first study does not allow us to infer causality,but the second does.

Question 11

Paloma has conducted a study on whether religious attendance is related to attitudes about legal gun control,using a nationally representative sample of 145 adults.She is surprised to find no association between the two variables.If there is actually a small correlation between religious attendance and gun control attitudes,one factor that is likely to have affected Paloma's findings is that

A)she only included adults in her sample instead of also including children and teens.
B)her sample size was not large enough to detect the difference by religious attendance.
C)religious attendance is too rare these days to be detected by national samples.
D)there is reciprocal causality that hides the basic association between the two variables.

Question 12

A quantitative research project examines the relationship between unemployment rates and voter turnout.In this study,the researchers consider whether current public policies play a role in the unemployment-voter turnout connection.We can see this study is one that uses multivariate methods because it includes

A)both macro-and micro-level variables.
B)at least three variables.
C)at least two dependent variables.
D)longitudinal data.

Question 13

Michelle is conducting research for her dissertation on whether birth order has an influence on how well high-school students perform on standardized tests.She has hypothesized that firstborn will perform better on average and has a long list of control variables she will test to make her study more informative.When Michelle tests for the main association between birth order and test scores,she finds a statistically significant relationship,but it is very small in magnitude.Her adviser suggests that she choose a different topic because

A)the finding does not have practical significance.
B)standardized test scores are so skewed that studies of them lack statistical power.
C)the null hypothesis cannot be rejected with sufficient confidence.
D)the importance of birth order is so varied across families.

Question 14

Which of the following would be an example of an inverse association?

A)Health and employment are related in a way that suggests reverse causality.
B)Fewer hours of sleep are associated with less weekly exercise.
C)More hours of weekly reading are associated with higher grades.
D)More education is associated with a lower likelihood of poverty.

Question 15

A researcher examines the possible effect of single-sex high-school education on men's college grades.He provides tables of his final results,which include several p-values that are less than 0.05,but he does not describe his many preliminary analyses,nor does he ever specify which high-school type he expects to predict higher grades among male college students.The scientific community is likely to have concerns over his interpretation of the p-values primarily because the researcher

A)does not hypothesize which high-school type should predict higher college grades.
B)does not also study the potential effect of single-sex schools for women.
C)was not transparent with the number and type of analyses he conducted.
D)does not study any measures of college outcomes other than grades.

Question 16

If certain values or responses on one variable tend to co-occur with certain values on another variable,we would interpret this as the existence of a(n)

A)association.
B)elaboration.
C)path diagram.
D)confounding.

Question 17

A study finds a statistically significant association between divorce status and family attitudes,such that divorced people have more unconventional attitudes about family life.Which of the following would provide evidence of the statistical significance of this association?

A)that divorce had an effect on family attitudes that was at least a certain magnitude
B)a sufficient level of confidence that the observed association is not due to chance
C)that the association would be found across many different types of populations
D)evidence that divorce predicts family attitudes,but that attitudes do not predict divorce

Question 18

The chance that at least some p-values will be lower than 0.05 increases when

A)the sample is nonprobabilistic.
B)the sample is smaller than 100 cases.
C)there is actually reverse causality.
D)the number of analyses gets higher.

Question 19

A city government has asked a research group to evaluate the effect of a new curfew ordinance on the number of monthly crimes.This curfew has reduced city revenue due to decreased commerce,and citizens are generally unhappy with it.While the curfew has already been instated,the city wants to use this study to decide whether to continue or discontinue the curfew.After six months,the study finds a statistically significant difference in the number of monthly crimes before and after the curfew was instated,but observes that the average number of monthly crimes over the six months only went down by 1.2 crimes with the curfew in place.When the researchers make their recommendation to the city based on this study,they should most likely recommend that the

A)study extend for one more year.
B)curfew be continued.
C)curfew be discontinued.
D)curfew be changed to one hour later.

Question 20

If a researcher hypothesizes that employment causes people to live longer,he or she may interpret an observed association between employment and number of years lived as evidence that supports his or her hypothesis.However,if adding healthiness to the model eliminates all of the original association between employment and number of years lived,this suggests that healthiness might predict both employment and longer life.This demonstrates that

A)reciprocal causality must always be considered for original hypotheses.
B)dependent variables must often be lagged for proper interpretation.
C)confounding factors present alternative explanations to original hypotheses.
D)adding new variables can change the results regarding the null hypothesis.

Question 21

In her study of married people,Aiko finds an association between more frequent religious service participation and staying married longer.She also explores whether stronger community ties explain at least part of this association.For this example,which of the following scenarios is MOST illustrative of using a control variable in regression analysis?

A)testing whether the association between religious participation and marital stability is evident when looking separately at higher and lower religious participation groups
B)testing whether community ties predicts religious participation for both married and unmarried people
C)testing whether controlling for community ties reduces the association between religious participation and marital stability
D)testing whether religious participation predicts community ties for both married and unmarried people

Question 22

A preliminary study shows an association between conservative attitudes and a higher likelihood of having a degenerative condition,such as heart disease or high blood pressure.The researcher decides to perform a simple elaboration of the association by running cross-tabulations of the attitudes-degenerative conditions link separately for those who are under the age of 50 and those who are 50 years of age or older.The reason the researcher does this is that

A)degenerative conditions may actually cause people to be more conservative.
B)older people may be more likely to be conservative and to have degenerative conditions.
C)age might be a mediating variable between attitudes and degenerative conditions.
D)it can help to get rid of any bias introduced if the sample is not probabilistic.

Question 23

In sociological research,suppression is

A)more common than confounding.
B)less common than confounding.
C)the equivalent of confounding.
D)dependent on the presence of confounding.

Question 24

Frederick is studying employees in a large firm to assess whether length of employment with that firm differs between professional employees and service employees.Further,he tests the possibility that occupation type might affect employment duration indirectly through job commitment.He uses an existing survey that asks the respondents their job title,their length of time with the firm,and one item that asks,"Overall,how committed are you to this job?" The problem of incorrectly measured differences would occur if

A)job commitment varies between occupational groups depending on the dimension.
B)all employees have a low likelihood of answering that they have low job commitment.
C)job commitment actually predicts whether someone is in a professional or service job.
D)professional and service employees have equal job commitment on average.

Question 25

Consider the results of a study that suggest more frequent marijuana use is associated with a higher number of hours of TV viewing.If the study tests whether both the frequency of marijuana use and the number of hours of TV viewing are predicted by whether respondents are employed,the study has tested whether employment

A)is actually the dependent variable.
B)has a practical level of significance.
C)is a confounding variable.
D)is a suppressor variable.

Question 26

Sarah is developing her study of the effects of neighborhood poverty and household poverty,which takes into account how children's psychological strain might explain how poverty predicts more risk behaviors,like smoking and drinking.In this example,the control variable is

A)neighborhood poverty.
B)household poverty.
C)psychological strain.
D)risk behaviors.

Question 27

One challenge in sociology is that the experiences being studied are often strongly related to individuals' characteristics.This causes difficulty because

A)sociology studies many factors that cannot be directly measured.
B)sociology is not able to test for reverse causality.
C)sociological research can rarely use experimental designs.
D)social experiences tend to be extremely random.

Question 28

Once a statistically significant correlation has been confirmed between an independent and dependent variable,________ are used to examine why the correlation occurs,and ________ are used to examine under what conditions the correlation occurs.

A)mediators;moderators
B)mediators;confounding
C)moderators;confounding
D)moderators;moderators

Question 29

Bianca is reviewing the research literature on marriage and health and learns that marriage is predictive of better health outcomes on a wide range of health measures.The possibility that healthier people are more likely to get married and stay married would illustrate the important potential for ________ in this area of study.

A)an inverse association between the two variables
B)reverse causality between the two variables
C)lagged independent variables over time
D)exogenous variation in the dependent variable

Question 30

Using an equation to express the relationship between the dependent variable and many independent variables simultaneously is a powerful feature of

A)regression analysis.
B)a lagged dependent variable.
C)normal distribution.
D)significance testing.

Question 31

Zeynep hypothesizes that sleep quality predicts weight gain and wants to be as sure as possible that any association she observes is consistent with her hypothesis.To do this,Zeynep should

A)use careful elaboration techniques to rule out any alternative explanations.
B)pay close attention to the practical significance her sample allows her to have.
C)draw her sample specifically from people who have reported sleep problems.
D)simultaneously model several control variables in a regression analysis.

Question 32

A research team is testing whether mothers' mental health is predicted by their own ratings of their parenting quality,and finds that mothers who have a more positive perception of their parenting report fewer depressive symptoms.However,they also find that employed mothers have lower parenting ratings and lower reported depressive symptoms.In this example,mothers' employment is a possible ________ variable.

A)suppressor
B)confounding
C)mediating
D)control

Question 33

There is a wealth of research on the U.S.population showing that people with higher incomes have longer life expectancies.However,if we include gender in the model,the story can become more complicated because men have higher average incomes than women do,but have shorter average lives than women do.Here,we can see that gender could be a suppressor variable of the relationship between income and life expectancy because

A)we are not sure if health affects income or if income affects health.
B)once gender is controlled,life expectancy becomes a lagged dependent variable.
C)being female is still a disadvantage because it is associated with lower average income.
D)being male is positively associated with income,but negatively related with life expectancy.

Question 34

Kai's data suggests that U.S.babies who were breastfed longer have higher standardized test scores and higher grades across childhood and adolescence.However,some of this association was explained by the higher average household income among mothers who breastfed longer and among children who had higher academic achievement on average.Yet again,some of the association between breastfeeding duration and academic achievement remained until immune system measures were included.Here,longer breastfeeding was associated with stronger average immune systems,which were,in turn,related to higher academic achievement (through fewer childhood illnesses that can affect grades and test scores).This hypothetical study demonstrates that

A)moderating variables can sometimes also be confounding variables.
B)confounds explain all of the association between the independent and dependent variables.
C)once a control variable is shown to be a confound,all other added variables will be confounds.
D)some variables in one model can be confounding while others can be mediators.

Question 35

A study examines the hypothesis that women perform more unpaid household labor than men on average because women also have lower average income than men.However,the researchers are encouraged to consider the possibility of reverse causality,such that the higher average time women spend in unpaid labor predicts women's lower average income.In order to respond to this suggestion,the researchers in this study will most likely need to

A)employ careful elaboration.
B)be sure their data has statistical power.
C)select a probability sample.
D)use longitudinal data.

Question 36

Consider a study on how sibling conflict might predict poorer academic performance.If this study is trying to avoid the problem of unmeasured differences,a good strategy would be to

A)study the question using an experimental design.
B)select the sample at random to avoid selection bias.
C)get all of the information from the siblings' parents.
D)gather data on many dimensions of a large number of variables.

Question 37

Faye's dissertation research will focus on how the hiring of new employees may influence the productivity of those already employed by the firm or company.If she chooses to test for the importance of job satisfaction after the hiring cycle,she is hypothesizing that

A)job satisfaction is a mechanism between hiring and productivity.
B)job satisfaction is a confound that predicts hiring and productivity.
C)productivity is actually a predictor of both job satisfaction and hiring.
D)job satisfaction will increase after new employees are hired.

Question 38

A researcher observes a link between his hypothesized independent variable and dependent variable.He then adds a new variable to his model that occurs after the independent variable.He is modeling this added variable as a(n)

A)lagged dependent variable between Time 1 and Time 2.
B)mechanism between the independent variable and the dependent variable.
C)test of the null hypothesis that the independent and dependent variables are related by chance.
D)statistical significance test of the association between the independent and dependent variables.

Question 39

One major reason that sociological research must consider the possible impact of confounding factors is that

A)survey questionnaires can rarely tap a wealth of in-depth information.
B)many sociological questions cannot be tested with experiments.
C)many sociological variables are related in a reciprocal way.
D)the sample type can often influence what observations are obtained.

Question 40

Reverse causality is an issue that comes up when

A)the independent variable occurs at the same time as the dependent variable.
B)we have our independent and dependent variables switched around.
C)the dependent variable is inversely related to the independent variable.
D)the dependent variable is actually measured at multiple time points.

Question 41

A researcher observes an association between hours spent at work and the risk of divorce.She controls for personal distress,thinking distress might explain why work hours would cause divorce.To assess this,the researcher will look for a change in the relationship between work hours and divorce once distress is added to the regression model.This is different from testing for a confounding variable because a test for confounding variables is conducted to determine if the association between the independent and dependent variables

A)is due to the common influence of the confounding variable.
B)is contingent upon the value of the confounding variable.
C)actually occurs in the reverse direction than hypothesized.
D)is actually an indirect pathway through the confounding variable.

Question 42

Consider the relationship between education and income,such that more education is associated with higher income,on average.Identify 2-3 control variables that should be included in the study,making sure to explain why these are potentially important controls.Explain how this would specifically operate with the use of regression analysis,including the regression analysis equation as it corresponds to all the variables in your model.Finally,explain what you would look for regarding the potential of control variables on your results.

Question 43

The reason a moderating effect is also called an interaction effect is because

A)the influence of the independent variable operates indirectly through the moderator.
B)the influence of the independent variable differs based on the value of the moderator.
C)the moderator actually influences the independent variable at certain values.
D)at some values of the moderator the dependent variable influences the independent variable.

Question 44

Consider the past research showing that if a child's parents divorce,the child is at higher risk of his own divorce when he reaches adulthood.One factor that might be relevant here is the adult offspring's relationship skills (think in terms of being able to communicate with an intimate partner in healthy ways,having positive relationship behaviors,etc. ).Write 4-6 sentences in which you describe two distinct studies: the first,in which adult offspring relationship skills are a mediator of the relationship between parents' divorce and offspring's divorce risk;and the second,in which relationship skills are a moderator of the relationship between parents' divorce and offspring's divorce risk.In your answer,be clear what research question is being asked in each study and how you would know that mediation had occurred in the first study and that moderation had occurred in the second study.

Question 45

Research shows that parents' occupational prestige is a strong predictor of their adult offspring's occupational prestige.In 3-4 sentences,consider the potential importance of neighborhood-level occupational prestige (in terms of what are the typical levels of occupational prestige across the neighborhood),and offer two contrasting scenarios: one in which neighborhood occupational prestige could be tested as a confounding variable and one in which it could be tested as a mediator.In your answer,be sure you clearly identify the key difference between a confounding and a mediating variable and demonstrate how your examples illustrate that difference.

Question 46

If a study finds evidence that a variable mediates the association between an independent and dependent variable,it means that once the mediating variable is added to the statistical model,the results would show a(n)

A)significant relationship between the independent variable and the mediating variable.
B)reduction or elimination of the association between the independent and dependent variables.
C)significant relationship between the mediating variable and the dependent variable.
D)increase or reversal in the association between the independent and dependent variables.

Question 47

If a researcher tests for whether chronic stress is a mediator of the association between socioeconomic resources and health,she is hypothesizing that

A)resources only affect health at certain levels of stress.
B)stress causes both resources and health.
C)resources predict health indirectly through stress.
D)resources interact with stress in their influence on health.

Question 48

Sean is studying a group of female high schoolers to test the relationship between changing schools during high school and adolescents' grades.Once he confirms that changing schools is associated with lower grades on average,he decides to include a measure of the resources in both the old and new schools to see if it interacts with the effect of moving schools on grades.Sean has decided to analyze the data in this way because he wants to test whether

A)being female is a mediator.
B)being female is a moderator.
C)school resources are a mediator.
D)school resources are a moderator.

Question 49

Write 4-5 sentences in which you describe a research project that relies on a nationally representative sample to study the relationship between gender and time spent on housework.Focus primarily on the null hypothesis that this study would test and the importance of the p-value for testing the null.Be sure to identify the independent and the dependent variable (your only two variables for this example are gender and time spent on housework),and the specific null hypothesis for this example.Be sure to describe a scenario about the study's observed relationship between gender and housework that clearly illustrates the null hypothesis for this example,and identify what p-value would be required for rejecting the null.

Question 50

Write 3-4 sentences explaining the difference between statistical and practical significance,including the description of a scenario (either hypothetical or one you learned about in this course)in which you could have statistical but not practical significance.In your answer,make it clear how your example demonstrates statistical and practical significance as well as the reason to be cautious about findings that are statistically,but not practically,significant.

Question 51

Consider the relationship between impulsivity,social isolation,and mental health,and consider the three following potential ways to study these associations.In one hypothetical study,a researcher considers whether impulsivity predicts poorer mental health,but only at high levels of social isolation.In a second study,a researcher tests whether social isolation explains why impulsivity predicts poorer mental health.In a third study,a researcher tests whether social isolation predicts both impulsivity and poorer mental health.Consider how social isolation is being modeled in each of these studies,and choose the correct description from the options below.The first study models social isolation as a ________;the second study models social isolation as a ________;and the third models social isolation as a ________.

A)moderator;mediator;confounding
B)suppressor;mediator;moderator
C)moderator;confounding;suppressor
D)confounding;mediator;moderator

Question 52

Biyu wants to know whether parental divorce predicts lower social adjustment for middle schoolers.She thinks divorce will influence adjustment indirectly through a decrease in interpersonal skills among children in divorced families.She also hypothesizes that the influence of parental divorce on school adjustment will be stronger for boys than for girls.To test for moderation in her hypotheses,Biyu should test whether

A)interpersonal skills interact with child's gender.
B)gender interacts with social adjustment.
C)interpersonal skills interact with parental divorce.
D)child's gender interacts with parental divorce.

Question 53

Jeanette is studying the association between middle-aged adults' education level and the education level of their parents.She has access to some genetic information about both generations,and decides to control for how shared genetics might explain the relationship between the education levels of middle-aged adults and their parents.In a separate study,Bowden is analyzing whether economic resources predict elementary school students' grades indirectly through the nutritional quality of students' diets.In these two studies,shared genetics

A)and nutrition are both possible confounding variables.
B)and nutrition are both possible mediators.
C)is a possible confounding variable and nutrition is a possible mediator.
D)is a possible mediator and nutrition is a possible confounding variable.

Question 54

A research team wants to study the influence of childhood neighborhood poverty on long-term health,such as life expectancy.As with many sociological questions,there are constraints on how this can be studied.Researchers interested in finding the most developed and reliable answers to such questions should

A)carefully design multiple experiments to establish causality.
B)use multiple strategies to analyze multiple sources of data.
C)target a very large sample for one well-elaborated study.
D)randomly select several neighborhoods,all of which are poor.

Question 55

Past research shows that the positive influence of education on income is smaller for African Americans and Latinos than it is for non-Latino Whites and Asian American-Pacific Islanders.In 4-6 sentences,explain how this finding demonstrates moderation.Identify the independent,dependent,and moderating variables in this example and explain what would have to be observed in the statistical model to conclude that moderation is occurring.Finally,briefly note how this finding might inform policies to equalize educational attainment as a way to decrease income inequality.