moo m > Answer Answer 1 moogo> PLANNING A STUDY Practice Questions Questions on the Vocabulary of Data Collection Which ofthe following i a characteristic of a census? W's based on anecdotal evidence Iv's generally more accurate than a sample Iruses secondary data 10s always part ofan experiment. It gathers data from every member of a population, E. it gathers data from every member of a population. A census contacts every population member for information, although this is often very difficult and usually not achieved. ‘There's a tendency to undercount individuals who are difficult to locate. Sampling is often a more accurate technique because of the limitations of gathering data from an entire population True or False: Ifa researcher uses a sample to make inferences about a population, the most {important consideration is thatthe sample be large enough to make generalizations from it False. Although lange samples are preferred to small samples, in order to maximize sampling variability, the most important consideration is that the sample be representative oF the population. Which ofthe following are true’? ‘There's always a treatment in an experiment, An observational study is one type of experiment Sample surveys are experiments only only Mand II only Land II only All of these are true reel 273 LearningCHAPTER 6 Answer pow > m Answer Answer ara sold Learning A. Tonly. An experiment requires that some treatment is applied to a group to see if there's, ‘predictable response. An observational study doesn’t have a treatment. Sample surveys fare one type of observation study. A teacher wants to compare the mean GPA of statistics students to mean GPA of all students ther high school. She looks up the GPAs of the statistics students and finds the mean. She then compares this mean to the mean GPA of all students. Which of the following are true? ‘This teacher will be relying on available evidence, or secondary data, ‘This is an observational study, This is an experiment, since there is a comparison of a sample to a population. Lonly Tonly Monly and Hl only Mand II only D. Land Hl only. The teacher isn’t producing data; she's relying on data that already exist. No treatment is being applied, so this is an observational study. ‘Your good friend Bill tells you that he has proof that not wearing your seatbelt is safer than ‘wearing one. His friend, Darlene, was in a bad accident. She wasn’t wearing her seat belt and was thrown clear ofthe car. The car immediately burst into flames. Darlene would have been killed if she hadn't been thrown clear. Which ofthe following statements is true? Since Bill is a trusted friend, this is good evidence that seat belts aren’t as safe as the authorities would have you believe. ‘This is a good example of using available evidence to help make a decision, This is an example of anecdotal evidence and shouldn’t be used to make a decision about ‘wearing seat belts. This is an example of an experiment that proves that it may be safer not to wear a seat bet. This should be considered good evidence for not wearing a seat belt, but you should do a study to be more confident in your decision before reaching a conclusion, C. This isan example of anecdotal evidence and shouldn’t be used to make a decision about \wearing seat belts. This is simply one case and provides no evidence in favor of wearing, ‘or not wearing, a seat belt. Note that you can’t infer the opposite, either, that wearing seat belts is safer. The anecdote provides no good evidence one way or the other.Answer PLANNING A STUDY You want to demonstrate that people are influenced by a cola’s product name over product Quality. You set up two tables on opposite sides of the quad of your school, so people can laste @ name brand and a store brand cola and tell you which they prefer. At one table the ‘asters drink the colas in unmarked cups, so that they don't know which cup contains which cola. At the other table, the subjects drink each cola from its original can. You analyze the results and report your findings in the school newspaper. Which of the following is true: This is an experiment, since you imposed a treatment on some subjects and left others untreated, This is an observational study, since you merely collected data and made @ comparison based on your data. ‘This isan experiment for which you used available data: students who were willing to taste the two colas. ‘There was no experimental design, since you didn’t decide on treatments or on how groups ‘would be assigned to treatments ‘This was not an experiment because your groups weren't selected randomly. A. this is an experiment, since you imposed a treatment on some subjects and left others Untreated. ‘The treatment allowed the subjects 10 know which brand of cola they were drinking, Then you could measure these subjects’ responses against the untreated group who didn’t know which cola they were drinking. The groups weren’t selected randomly but that doesn’t mean it wasn't an experiment-—although it may mean it wasn’t avery good Questions on Data Collection Which ofthe following isan example of using anecdotal rather than available datato support Using the results of a nationwide survey to conclude that an individual's income is related to her occupational status ‘Concluding that @ particular brand of fish food leads to faster fish growth, based on your friend’s observations Using the results of an experiment in different local gardens to conclude that a new plant femilzer leads to sweeter-asting tomatoes Concluding that the fertility rate in third-world counties is related to the labor force pattic- ipation rate for women, based on official census statistics None of the above LearningCHAPTER 6 Answer moos Answer moog? Answer ae oll Tearing B. concluding that particular brand of fish food leads of faster fish growth based, on the ‘observations of your friend. Conclusions based on anecdotal data, which are the results of a single or small number of cases that aren't part ofa deliberate experiment, are usually not reliable. In this case, your conclusion about the effects ofthe fish food is based solely on ‘one friend's observation. ‘You're interested in drawing some conclusions about the political attitudes of college under- ‘graduates. Your objective is to administer a survey to a representative sample of students, ‘Which of the following is true? It's necessary to set up an experiment with control and treatment conditions to properly collect representative data I's likely that using this approach will only identify a sample of males. ‘Your approach will produce anecdotal rather than available data ‘There’s no way to ensure that your sample will be representative of all college seniors. By choosing only seniors, your sample won't be a representative sample of students, E, by choosing only seniors, your sample won't be a representative sample of students Since your intent is to examine the attitudes of undergraduates, you need to select a sample of students from all levels (freshmen, sophomores, uniors, and seniors) to maintain a proper representation of the undergraduate population. A plant researcher identifies a sample of plots of land to examine the effect of a new fertilizer fn the speed of growth of a particular crop. She treats half the plots of crops with the new fertilizer (NewFert), and the other half with the more traditional brand (FertiWhammo). She observes that crop growth in both sets of plots is almost identical. The data collection strategy she has used isa (an): experiment. ‘anecdotal study. ‘observational study. survey analysis, {quasi-study. ‘A. experiment, Since the researcher exerts some control over some of the variables (admin- isters a different treatment to different sets of plots), this is an experiment. ‘You're interested in whether tutoring is more effective when the student volunteers to be tutored, You compare the test scores of two groups of students: those who chose to go to after-school tutoring sessions and those who are required to go to the sessions by the school administration. This is an example of a (an:moop> Answer Answer PLANNING A STUDY experiment. anecdotal study ‘observational study survey analysis. fieldwork study CC. observational study. In this case, you're simply measuring the performance of two different groups. You have no influence on the independent or dependent variable. If this ‘were an experiment, you'd have assigned the students to the two different groups. Which of the following could not be an example of secondary data? Information on attitudes about abortion among senior citizens collected via a short survey by the University of Michigan Data on length of gestation among female monkeys from a sample of monkeys at 200s in the United States Data on individuals’ weight change among those given a placebo pill in a nationwide experiment Information about the effects of heavy metal music on violent behaviors, based on the stories of three juvenile offenders in New York Ci Data on the behavioral patterns of hummingbirds in a large part of Yellowstone National Park, based on observations from five stations across 30 days. 1D. information about the effects of heavy metal music on violent behaviors, based on the stories of three juvenile offenders in New York City. Secondary data are easly available and typically collected in a systematic way from a large sample of individuals, cases, and 0 on, In this situation, you's mnecdotal data from a small number of cases to decide whether there’s a link between heavy metal music and youth violence. A researcher is conducting an experiment on the possible effects of a new pill on weight Toss among obese men and women, He has given the new pill (0a sample of women and an ‘ld pill (0 the men, The researcher is comparing the effects of the old vs. new pills to see which group demonstrates the greatest percentage decrease in weight, You believe that his design is awed, The law in his design i: ‘There's no response variable in the experiment. The experimenter did not give both pills to those in the treatment condition (the females should get both the old and new: pill) You can’t measure the effect of the new pill on the men, since none of them received this treatment, LearningCHAPTER 6 Answer moo w > Answer ‘Learning ‘There are too many treatment conditions in the experiment, None of the above C. you can’t measure the effect of the new pill on the men, since none of them received this treatment. If you want to compare the effects of the old vs. new pills on both men and women, each pill must be administered to different groups of men and women. The experiment right now has two groups: men who get the old pill and women who get the new pill. The experiment should have four groups: women/old pill, women/new pill, men/old pill, and men/new pill ‘An observational study based on survey data concluded that individuals who took more vitamin C were able to recover from the flu faster. You want to replicate this study using an experimental approach. The treatment in this experiment might be: how long it takes t0 recover from the flu, in days ‘whether an individual took vitamin C in a pill form or Tiquid form, the amount of vitamin C taken per day: 0:mg, 1,000 mg, 2,000 mg, or 3,000 mg, the change in body temperature over the period of the experiment, All are acceptable treatments C. the amount of vitamin C taken per day: 0 mg, 1,000 mg, 2,000 mg. or 3,000 mg, Since you're interested in whether the dosage in vitamin C has an impact on recovery time, the treatment is based on the dosage level. Consider the following four studies: A survey of newspaper editors examines their political views about foreign policy issues. ‘The analysis and conclusion involves a summary of the survey results, ‘Third graders are randomly sampled and assigned to an intensive language course, Their performance on a year-end language testis compared to students who did not take the course. A study compares the change in home values between 1990-1999 in ten different neighbor- hhoods in Seattle. The results show that there's a higher percentage change over time in less affluent neighborhoods, A study compares the durability of machine parts, ‘The comparison looks at machines that Were purchased with new metal alloy parts and at those purchased with old metal alloy parts. ‘The results show that durability increases with the use of the new alloy. ‘Which examples are experiments? and I Wand IV Only 1Answer Answer mpo p> Answer PLANNING A STUDY Only Mand 1v only Il Inthis study, the researcher exerts control over the variables by randomly assigning students to different treatment conditions, The researcher then measures whether there are different levels of response associated with each treatment condition, Many statisticians say that the U.S. Census, which attempts to count every population ‘member directly, is significantly less accurate than a count estimated by random sampling. Why mighta countestimated from random samples be more accurate than acensus? (Choose the best answer) Random samples are scientific whereas censuses are not A true census takes so long that by the time all population members are counted, the population has changed ‘A census often can’t find every population member, so some groups (such as the homeless) are often under-represented, A census is a haphazard sample, and census takers may be bribed by households, A census is an old method written into law before anyone knew anything about statist C-acensus often can’t find every population member, o some groups such as the homeless) ‘re under-represented. It's nearly impossible to actualy find and count every member of the USS. population, so the count is likely to have many omissions. A properly drawn random ‘sample takes this into account. In many cases, inferential statistics can use a good random sample to estimate the population better than an actual count. Which of the following isthe best representative sample of the adult population in the United States? ‘Simple random sample of 10,000 adults from different city phone books ‘Simple random sample of 10,000 voters from across the country Sample of 50,000 individuals atthe Super Bow! (which draws from all over the country) ‘Simple random sample of 1,000 adults from across the country Sample of 50,000 members of AARP (American Association of Retired Persons) D. sample of 1,000 adults randomly chosen from across the country. This isa large sample in which all adults have a more-or-less equal chance of being selected. The larger samples Of voters and names in the phone book may be approximately representative (and may work in many cases), but by selecting only from voters and people in the phone book, you lunder-represent some groups of people. roel 20 LearningCHAPTER 6 Answer Answer Questions on the Vocabulary of Experiments and Surveys A block is best described as: the use of chance to divide experimental units into groups, ‘a design in which neither the experimenter nor the subject knows who is in the treatment ‘group and who is in the control group. 1 group of subjects that are similarin some way known to affect the response to the treatment, the policy of repeating an experiment on different subjects to reduce chance variation and to determine the generalizability of the findings. the tendency of subjects to respond favorably to any treatment. C. a group of subjects that are similar in some way known to affect the response to the tweatment. For example, if we know in advance that age has an effect on the response to an ‘experiment, we might divide the sample into age blocks and run the experiment separately ‘on each age block. Replication is best described as: the use of chance to divide experimental units into groups. ‘a design in which neither the experimenter nor the subject knows who is in the treatment ‘group and who is in the control group. ‘group of subjects that are similar in some Way known to affect the response to the treatment the policy of repeating the experiment on different subjects to reduce chance variation and to determine the generalizability of the findings the tendency of subjects to respond favorably to any treatment, D. the policy of repeating the experiment on different subjects to reduce chance variation and to determine the generalizability ofthe findings. Replication is repetition of the experiment, ‘one of the principles of experimental design. Double-blind is best described as: the use of chance fo divide experimental units into groups, ‘a design in which neither the experimenter nor the subject knows who is in the treatment {group and who is inthe control group. ‘a group of subjects that are similar in some way known to affect the response to the treatment,Answer E. Answer moog > Answer 6 Pi NNING A STUDY the policy of repeating the experiment on different subjects to reduce chance variation and to determine the generalizability of the findings, the tendency of subjects to respond favorably to any treatment. B. adesign in which neither the experimenter nor the subject knows who isin the treatment Toup and who is inthe control group. Neither the experimenter nor the subjects know who is receiving the treatment, so neither can be influenced by a placebo effect. The placebo effect is best described as: the use of chance to divide experimental units into groups 4 design in which neither the experimenter nor the subject knows who is in the treatment group and who isin the control group. ‘ group of subjects thet are similar in some way known to affect the response tothe treatment, the policy of repeating the experiment on different subjects to reduce chance vatiation and to determine the generalizability of the findings, the tendency of subjects to respond favorably to any treatment. E, the tendency of subjects to respond favorably to any treatment. The placebo effect is psychological. I's the expectation the subject receiving the treatment might have about the treatments effec, rather than a real effect ofthe treatment Which of the following are considered to be important principles of experimental design? Control Randomization Replication Tonly only Tand Il only I and I only 111, and I E, I, Il, and 1M. All three are important principles of experimental design. Control helps reduce the effects of confounding variables. Randomization and control help to insure that any differences between groups are due to the treatment, and replication allows us to generalize to a population. A primary purpose of randomizati sed 251 TearingCHAPTER 6 moo @ > Answer moo p> Answer moom > Answer ‘Learning to minimize the placebo effect to avoid having to use complicated double-blind procedure. to eliminate bias between treatment groups to reduce variability within treatment groups to assist in developing 1 matched-pairs procedure . to eliminate bias between treatment groups. Bias is introduced when there is some systematic difference between the treatment and control groups. Random assignment of subjects to treatment groups avoids bias. A primary purpose of blocking is: to minimize the placebo effect. to avoid having to use a complicated double-blind procedure to eliminate bias between treatment groups to isolate the separate effects ofthe treatment and another important variable to assist in developing a matched: pairs procedure. D. to isolate the separate effects of the treatment and another important variable. If we block. or divide the sample based on some important variable we know to have an effect on the response (such as gender), we can separate the effects of the treatment from the effect of the blocking variable. ‘An experiment is conducted in which a series of tess are performed on pairs of identical twins who were raised separately. A comparison of scores on each pair of twins is used for analysis. This is best described as: ‘a matched-pairs procedure ‘a double-blind procedure. 4 randomized treatment group procedure ‘procedure with two explanatory variables. ‘an example of the placebo effect. ‘A. a matched-paits procedure. The twins make a natural matched pair for some kinds of ‘experiments in which one might want to match subjects as closely as possible on biological characteristics.9. Answer a) b) °) PLANNING A STUDY True of False: A doctor gives «pill to patient, and the patient gets better. This must be an example of the placebo effect. False. The question doesn’t give any information about the design ofthe study, so we don’t know whether or not this observation is a true effect of a treatment or an example of the placebo effect. Free-Response Questions on Aspects of Experiments ‘There are three important principles of experimental design: control, randomization, and replication. ‘Sometimes bias can affect the results ofthe experiment. Bias may come from the behaviors of the experimenters; other times bias may result from the placebo effect, the Hawthorne effect, or from a flawed experimental design, An experimenter identifies a random sample of individuals for an experiment to test the effects of a new memory pill. He explains the purpose of the experiment to the subjects, telling them that they will take a set of pills to improve their ability to retain knowledge. He randomly assigns half the participants to receive a daily dose of the memory pill. He gives the other group a daily dose of a sugar pill (a pill with no known effect on memory improvement). He administers a memory test at the end of the experiment and finds that ‘memory skills improved at about the same level for each group. What might explain the results ofthis study? How might one improve the study to reduce the likelihood of any bias? Professor Syssaphys conducts an experiment on the possible effects of room temperature ‘on the speed at which a machine produces rubber snakes, He identifies 100 machines and sets up the following design: 50 machines are placed in a “warm” room (85 degrees) 30 of the machines in the “warm” room are phiced near the windows 20 of the machines in the “warm” room are placed away from the windows 50 machines are placed in a “cold” room (55 degrees) 15 of the machines in the “cold” room are placed near the windows 35 of the machines in the “cold” room are placed away from the windows ‘What are the factors in this experiment? What are the levels of these factors? How many treatment conditions are there in this experiment? What are the treatment conditions? How many machines are in each treatment condition? What is the response variable? raph 2 nes earningCHAPTER 6 a) b) °) ‘You're interested in whether white Americans’ attitudes about prejudice towards African “Americans change after they lear more about the history of African Americans. “You identify a diverse sample of whites that contains men and women of varying ages. incomes, religions, and levels of education. You decide to use a documentary as way of teaching {Individuals more about the history of the African American experience. Recall that there fare two different ways of conducting a matched pairs experiment. Describe in detail how You'd use each type of matched pairs experiment fo examine your hypothesis on attitudes about prejudice AA botanist is interested in the effects of a new plant nutrient on the brightness of the colors of a particular type of tulip. She believes thatthe brightness ofthe colors will improve if she uses the new nutrient (X4TS) rather than the old nutrient (X3Y8). She also wants to ‘make sure that the effect of the nutrient is present after controling for two possible lurking ‘variables—location of garden (indoor greenhouse or outdoor farm) and acidity of the soit (high, moderate, low), She's identified 60 indoor and 60 outdoor locations to conduet an experiment on the effects of these factors on the response variable. ‘What are the factors? What are the levels of these factors? What is the response variable? Create an experimental design that measures the effect of the plant nutrient on color bright- ness, controlling for garden location, and soil acidity. Identify the treatment conditions, and if possible, assign the same number of units into each condition. You're a scientist, You've conducted a double-blind replication experiment examining the effects of factor A on response variable B, blocking for gender. Explain what you've done, in simpler terms, to someone who knows nothing about experimental design. ‘Answers to Free-Response Questions on Aspects of Experiments ‘The researcher is most likely observing some type of Hawthome effect. Ifthe participants know they're being observed, even those who receive the placebo might make an extra effort to remember things because they know thatthe researcher is hoping to see better memory. ‘A way to compensate for this problem is to design the study so participants don’t know ‘what effect is being measured. (Though in some cases this isn't possible for ethical reasons, People often must be informed about why they're being studied.) NOTE: Don't confuse the Hawthorne effect with the placebo effect. The placebo effect re- sults from subjects" expectations about what the treatment will do for them. The Hawthorne effect, on the other hand, results from the subjects knowing that they're being observed. Placebo effect is a more specific term relating to subjects’ expectations about a treatment; Hawthorne effect isa more general term relating to any change in behavior that results from subjects knowing they're being observed. These two effects are similar in some ways, but technically they aren't the same thing,a) ») ° PLANNING A STUDY Factor Room Temperature Proximity to Window vs. “Cold” room s. “Away from” windows ‘There are four treatment conditions, each combination of the levels of the two factors—room temperature and proximity to window: ‘Treatment Condition # of Machines, ‘Warm" Room! 30 [Near the Windows “Warm” Room! 20 ‘Away From Windows ‘old” Room 1s Near the Windows “Cold” Room! 35 ‘Away From Windows ‘The response variable in this experiment is speed at which a machine produces a rubber snake, One way of measuring this is to count the number of snakes produced by the machine in a specified period (# of snakes per hour) Matched Pairs ‘An example of a matched pairs design would involve matching individuals based on a set of characteristics suchas gender, age, income, level of education, and religion, and placing them into treatment and control groups. For instance, you could collect information on gender, ‘age, income, level of education, and religion from a sample of $00 white Americans. Then ‘you could identify two individuals with a similar set of characteristies; for example, two men around 30 years old with medium incomes, college educations, and who are Catholic. ‘Once you've matched these individuals, one is randomly assigned to the treatment condition (a group that watches the documentary), and the other is assigned to the control condition (a group that receives no treatment). It's important that the assignment be random. This matching procedure continues until all individuals are assigned to either the treatment or control group. The assumption is that the two groups should be similar in terms of the factors on which they were matched, periment #1 At this point, those in the treatment group watch the movie, while those in the control ‘group don’t view the documentary. Ata period after the administration of the treatment (for ‘example, wo weeks later) survey is used to measure individuals” attitudes on prejudice. If the treatment has an-effect, you should expect to find that atitudes differ for those who saw the documentary compared to those who didn't see it Matched Pairs Experiment #2 Another way of using a matched-pairs design involves measuring the same individuals at two different times, In this instance, we could identify a sample of 100 individuals and have them complete a survey on attitudes about prejudice before any treatment is administered, Apex fad LearningCHAPTER 6 a) b) 285 Ay 2 e ‘Then we'd show all 100 individuals the documentary on the African American experience. [At some point after the treatment (for example, two weeks later), we'd have the same 100 individuals complete the same survey. If the treatment has an effect, we should expect to see that atitudes in this group will have changed from the baseline period (before they saw the movie) tothe follow-up period (after they saw the movie). The matching in this instance involves collecting information from the same individuals at two times, The control is their attitudes before the documentary, nd the response variable is the change in attitudes demonstrated after the documentary. Factor Levels Plant Nutrient New (X4TS) vs. Old (X3Y8) Garden Location Indoor vs. Outdoor Soil Acidity High vs. Moderate vs. Low ‘The response variable isthe brightness of the color of the tulip in that location. In this case you've got three different factors—plant nutrient, garden location, and soil ‘acidity, To properly control for these possible lurking factors, you should create treatment conditions for each of the combinations of the levels of the factors, thus giving you a 2 by 2 by 3 design, or 12 possible conditions, Since there are 120 locations (60 outdoor and 60 indoor locations), you might be able to randomly assign 10 locations to each of the 12 treatment conditions. Then you could randomly assign bulbs to each location. The tweatment conditions are: New Nutrient / Indoor Location / High Acidity New Nutrient / Indoor Location / Moderate Acidity New Nutrient /Indoor Location / Low Acidity New Nutrient / Outdoor Location / High Acidity New Nutrient / Outdoor Location / Moderate Acidity New Nutrient / Outdoor Location / Low Acidity (Old Nutrient / Indoor Location / High Acidity (Old Nutrient / Indoor Location / Moderate Acidity (Old Nutrient / Indoor Location / Low Acidity (Old Nutrient / Outdoor Location / High Acidity ‘Old Nutrient / Outdoor Location / Moderate Acidity Od Nutrient / Outdoor Location / Low Acidity NOTE: It’s important that the assignments ofthe locations and the bulbs be random (though the random assignments ofthe locations can happen within the constraints of the conditions seb.moope Answer PLANNING A STUDY First, the fact that you're conducting a replication implies that you're in some way repeating of redoing an experiment that was done in the past. The presumption is that someone in the past did an experiment examining the effect of A on B, and perhaps your replication involves seeing whether you'd get similar results on a different population or under slightly different conditions. ‘The statement “effects of A on B, blocking for gender” suggests that you're interested in whether those who receive treatment A (or different levels of A) will show different levels of response B. For example, let's say that A is medicine dosage and B is severity of flu symptoms. ‘The question is whether those who receive different dosage levels (for example, 200 mg vs. 400 mg) demonstrate more or less severe symptoms. The assumption is that some group of individuals will receive one dosage level while the second group will receive the other dosage level “Blocking for gender” suggests that you believe the effect of A on B may be much different for males and females. Thus, there is some concern that the effect of the explanatory variable (medicine dosage) may be confounded by the effect of gender. The process of blocking indicates that you first split the sample into males and females and them randomize individuals within each group into different treatment conditions. “The idea of a “double-blind” procedure tells us that neither the experimenter nor the subject has any information about the level of the treatment the subject receives. Such an approach is often used if there's a concern about the experimenter having an unwanted effect on the response variable Questions on Designs for Experiments ‘You're going to test two new varieties of fish food vs. s commonly used fish food. You set ‘up an experiment as follows: 60 fish are randomly assigned to each of three different tanks, (One tank is randomly selected to receive one of the new foods, another tank to receive the ‘other new food, and the third tank to receive the common food. Fish growth is measured lover time, This is an example of: randomized block desig. 4 double-blind matched pairs design. 4 completely randomized design with no contrl group. 4 comparative block design. ‘completely randomized design with a control group. E. a completely randomized design with a control group. This is the simplest type of experiment. Units (the fish) are completely randomly assigned to treatment conditions: regular food (the control) or one ofthe types of new food (the two treatments). The responses (growth) of both treatment groups can be compared to each other and to the control (the roup that receives the common food). If this were a blocked design, the subjects would divided according to particular characteristics, such as male or female, and then subjects, from each division (each block) would be randomly assigned to a treatment,CHAPTER 6 Answer po > moo Answer 288 geld Learning Say you're designing an experiment that looks at the effect of different types of newly ‘developed fish food on the growth rate ofthe fish. The treatments are new food A and new food B; the control is the old food. You believe that the effects ofthe different types of food ‘on fish growth will vary for goldfish vs. tiger fish vs. guppies. How could you design an experiment so that you're blocking on the species of fish? Eliminate one of the treatment conditions. ‘Match three different fish (one goldfish, one guppy, and of one tiger fish) on a set of charactersties such as weight and length and then randomly assign them to different tanks. Only measure fish growth in the two treatment conditions, First separate the fish into the three species. Then within each species, randomly assign each fish to a treatment All of the above. DD. first separate the fish into the three species and then randomly assign them to tanks that receive the treatments. A block design first distinguishes the suibjects based on some characteristic that might confounding variable (inthis case species of fish). After separating the fish by species, units within each block are randomly assigned to a treatment or control ‘group. The design resembles doing three separate experiments: one for just guppies, one for just goldfish, and one for just tiger fish, ‘You're designing an experiment that looks at the effect of different types of newly developed fish food on the growth rate of three species of fish: tiger fish, guppies, and goldfish. The treatments are new food #1, new food #2, and old food. For a randomized block design, blocked on species of fish, how many treatment groups would you have? (Think of food” and one of the treatments.) 3 s 6 9 2 1-9. You've first separated the sample of fish into three species: guppies, tiger fish, and oldfish. The fish in each of these species are now randomly assigned within their block to tone of the three treatments: new food #1, new food #2, and old food. This would give you athree by three design or nine possible conditions: guppy/new food |, guppy/new food 2, guppy/old food, golufish/new food 1, goldfish/new food 2, goldfish/old food, tiger fish/new food 1, tiger fish/new food 2, and tiger fishfold food.moom> Answer pom > E Answer PLANNING A STUDY '50 boy's and 50 girls with ADD (Attention Deticit Disorder) were selected for an experiment to test a new drug for the treatment of ADD. Half of the boys and half of the girls were selected at random to receive the new drug, and the other half of each group received @ placebo. A reduction in symptoms of ADD was measured for each subject. The basic design of this experiment is: completely randomized completely randomized with one factor, gender randomized block, blocked by drug and gender randomized block, blocked by drug randomized block, blocked by gender. E. randomized block, blocked by gender. Because the design makes sure that exactly half the boys and half the girls are assigned to a treatment or control group, this experiment isn't a completely randomized design. The subjects are blocked, or divided, by gender, and then ‘each block is randomized into treatment and control groups. Let's say you're interested in the effects on boys of different dosage levels of a new drug for the treatment of Auention Deficit Disorder (ADD), You set up an experiment to consider the factor of dosage with two levels (300 mg vs, $00 mg). What would be the different treatment groups of the experiment within each block? ‘Three groups: no drug/300 mg of new drug/500 mg of new drug Three groups: placebo drug/300 mg of new drug/500 mg of new drug ‘Two groups: 300 mg of new drug/500 mg of new drug, Four groups: 300 mg of new drug/300 mg of placebo drug/ 500 mg of new drug/S00 mg of placebo drug ‘Two groups: placebo drug/either 300 or 500 mg of new drug. B. three groups: placebo drug/300 mg of new drug/500 mg of new drug. If you're interested in whether the drug itself has an effect and whether the level of dosage also has an effect, you need three different treatment groups. Remember, one of these groups will stil receive the placebo drug, though you're not interested in varying the dosage of the placebo drug. ‘The two “new drug” groups simply vary on the dosage level You've read a story in The New York Times claiming that individuals who engage in aerobic exercise for atleast an hour a day demonstrate fewer symptoms of depression. You read that ‘an experiment was conducted in which a researcher first administered a survey on depression ‘and self-esteem (0 100 individuals and then taught them some proper techniques of aerobic exercise. The 100 individuals were then sent off to exercise at least one hour a day. After two months, the depression and self-esteem survey was administered again and showed that depression symptoms declined and self-esteem increased. The experimental design used here isa reall 280 LearningCHAPTER 6 ‘A. matched pairs design in which subjects are matched on a set of characterstics before random assignment B. matched pairs before and after design. ©. completely randomized design D, randomized block design, blocked on form of exercise and gender. E, quasi design. Answer B, matched pairs before and after design, This is a matched pairs before and after design Information on depression and self-esteem is collected from the same individuals before and after the treatment (exercise), and the change in depression and self-esteem is measured in this group. In this design, each subject acts as its own control. While this isn’t an ideal design (it’s often better to have a separate control group to compare with the treatment group) itis an acceptable way to conduct a study if separate control group isn’t possible, Anexperiment was conducted in which a researcher first administered a survey on depression nd self-esteem to 100 individuals and then taught them some proper techniques of aerobic exercise, The 100 individuals were then sent off to exercise at least one hour a day. After ‘ovo months, the depression and self-esteem survey was administered again and showed that depression symptoms dectined and self-esteem increased. You're a bit skeptical of the results. You believe there are two problems. One, you really know the effects of the treatment without a control group, and two, you suspect there's interviewer bias. ‘Which type of experimental design might best reduce these problems? ‘A. Matched pairs design, in which subjects are matched on gender and age before starting the exercise routine B. Randomized block design, in which the interviewers or researchers are randomiy assigned {o teach different groups of subjects ise” for one hour ©. Completely randomized design with two treatment conditions: “new exe a day vs, “no exercise" for 1/2 hour a day D. Completely randomized design comparing those who exercise vs. those who don'texercise, and a blind procedure : E. Completely randomized design comparing those who use the “new exercise” vs. those who don’t use the “new exercise” Answer D. completely randomized design comparing those who exercise vs. those who don’t ex- ercise, and a blind procedure. You could randomly assign the subjects to one of two cconditions—exercise one hour a day vs, no exercise (the control group). And with a blind ‘procedure, the interviewer doesn’t know which treatment condition the subjects are in. This is useful for reducing interviewer bias. 290 rook earning eerrrrrrr———irrerrrrrr—hLEA B. c. D. E Answer mooa> Answer PLANNING A STUDY [A researcher identifies a sample of teenagers and chooses two males, age 16, from lower- income families. She randomly selects one of the males to be in an intensive tutoring course: she places the second male in a standard classroom, She continues doing this with pairs of ‘males and females. This isan expample of: blocking based on gender. ‘matching pairs on responses before and after the tutoring course ‘matching pairs for random assignment on gender, age, and income status completely randomizing teenagers to treatment and contol conditions. blocking teens, xs based on gender, age, and income status. CC, matching pairs for random assignment on gender, age, and income status. When you choose pairs based on some common characteristics and then randomly assign one member (of the pair to one conditon and the other member toa second conditon, you're matching. In this case, the experimenter is matching on gender, age, and income status, and the goal is to find pairs of teenagers that are most similar on these characteristics before assigning them to different conditions ‘The following diagram describes an experiment in which subjects have been divided into three groups. Each of the three groups is randomly assigned to one of two treatments, Group 1 European Ps compare ‘Americans > Group 2 (no tering) Group 1 Avican Seompare Americans > croup 2 (no tering) as ‘~scomeare + eroup 2 i toring) Asin The design of this experiment is ‘matched paits before and after design completely randomized design with two factors, race and tutoring, randomized block design, blocked on tutoring completely randomized design with one factor, grades. randomized block design, blocked on race. E. randomized block design, blocked on race. In this case you can see that the sample is first distinguished by race, then those inthe different blocks are randomly assigned to either the tutoring or no tutoring conditions. Thus, we are blocking based on race onl 291 eSmingCHAPTER 6 moop> Answer Answer earning Questions on Types of Samples for Surveys A company wants to obtain information on students aged 17 t0 22. This company randomly selects 2,000 students from the mailing ists of four universities. (Every student attending tone of these universities is on the respective mailing lists.) Then the company sends the questionnaires to the sample of 2,000. What type of sample is this? A representative sample of students aged 17 t0 22 at these four universities, A stratified random sample of students aged 17 to 22 at these four universities, A simple random sample of all students at these four universities. A stratified random sample of all students at these four universities. A multi-stage random sample of all students at these four universities. a simple random sample of all students at these four universities. The researchers would hhave to select the 17 to 22 year olds from the total sample after they got the questionnaires back, since there is no way to know the ages of the students from the mailing lists alone FFor a semester project, a student needs to select a random sample of 10 students from his senior class of 250. He carefully numbers the class list from 000 to 249 and then uses a random number generator to obtain 3-digit random numbers, The 10 unique numbers are his sample. He notices that they all belong to the same honors AP Calculus class. Another student claims that this could not be @ random sample. Which of the following is true? ‘The sample drawn is so unlikely that it could not be considered a random sample. Since the selected students are not representative the entire senior class, this is mot a random sample. ‘Whether a sample is a random sample or not is determined by the sampling method, not the results. The method used here is OK. ‘A sample size of 10 is too small to be a random sample of 250. ‘The class should have been numbered from 001 to 250 rather than from 000 to 249 to produce a better random sample C. whether a sample is a random sample or not is determined by the sampling method, not the results. The student does have a random sample. All members of the group had an equal chance of being selected, and each member is independent. Sometimes random samples aren't representative samples, AAs primary research for one of her books, Shere Hite distributed 100,000 questionnaries to women’s groups; 4,500 women responded. Hite found that 96% of the women felt they ‘give more emotional support to than they get from their husbands or boyfriends. Which of the following best describes her sample?9 Oo p> Answer moo p> Answer pom > Answer PLANNING A STUDY Representative sample Setfselected sample Satfed random sample Systematic sample ‘Cluster sample B. selfselected sample. These questionnaires were distributed in much the same way ‘magazine questionnaries get distributed, to a group of people who are easy to reach. The respondents chose whether or not {0 respond, which makes this a type of convenience sample, An important reason a market researcher collects data using a stratified random sample rather than @ simple random sample is to-collect data at a lower cost to eliminate, ora least reduce, bias tomake arepresentative sample more likely than one produced by simple random sampling convenient data collection to have a systematic way of obtaining the data C. to make a representative sample more likely than one produced by simple random sam- pling, Stratified random samples ensure that we get the correct proportions of groups that have certain characteristies we want to measure. Sex. age, gender, and race are characte istics often stratified when sampling, Which of the following would generate an SRS of 50 integers from 5 to 25 on the TI-83? randNorm(5.25,50) randln(50. ) { rand(5,25.50) 50 randint(5,25) randin(5,25,50) E. randint(5.25.50), The randInt function produces a random number between the first and second numbers you specify. The third number is the number of times you want a random i number generated. eel 20 LearningCHAPTER 6 moo p> Answer Answer A 8. . ‘A major reason to use a systematic random sample rather than a simple random sample would be: It’s often easier to do than other types of random sampling, ‘The sample is mote likely to be random. Ithas a good chance of being representative of the population Tonly Henly Ul only 1 and I only Mand II only . 1 and Il only. A systematic random sample usually takes less time to generate than a simple random sample, especially if a randomly ordered list is already available. If done correctly a systematic random sample has a good chance of drawing a representative sample of the population, ‘True or False: Blocking in an experiment and stratifying in a survey accomplish the same thing. They control the amount of variation in key characteristics within the total sample, such as race or gender, that are likely to be important to the outcome of interes. True. The main difference between them is that no treatments are applied to samples that are stratified, whereas experimental treatments are applied within blocks, Consider the following four rows from a random number table: 5558899404 70708-41098 46563 5693448304 51719 1297513258 1308845144 72321 81940 00360 02428 Use the random number table above to draw a sample size of 10. Twenty percent of the population has a trait you wish to find. A success is defined as a 0 or 1. Draw eight samples of size 10 by reading each row from left to right. What are the greatest and fewest number of suecesses you might expect in a sample of size 10? al 3.0 4.0 3.1E. Answer Answer ANNING A STUDY 42 C.4,0, The winning blocks of ten are in bold [55588 99404] [70708 41098] (4 instances) [46563 56934] [48394 51719] (0 instances) (12975 13258] [13048 45144] (72321 81940] {00360 02428} (4 instances) Questions on Bias in Surveys Which of the following are true statements?” Under-coverage bias can be overcome by greatly increasing the sample size. Well-designed probability sample always eliminate bias. Response bias results only from the wording of survey questions, only only U1 only Wand It None of these E, None of these. There may be some truth in each statement, but none of them are always true, ‘True or False: The bigger the sample, the smaller the bias False, The bigger the sample, the less the sample variability, or error due to chance. Sample size has nothing to do with bias, which is error induced by some systematic process A random sample of 15,000 people is selected from local telephone books; 3,000 return the ‘questionnaire. This survey suffers from: Sampling bias Response bias Question-wording bias Tonly only rood baad TearingCHAPTER 6 Answer moogm > Answer u uL 296 rool Learning Monly and Il only Hand II only D. and Honly. There's certainly under-coverage, which isa sampling bias, and there’s non- response bias, which is one kind of response bias. There's no indication that the wording is ‘A surveyor for a vegetarian magazine asks the following question as part of a survey’ “The U.S. Government subsidizes the cost of food and water for cattle. Given that most agricultural land is for feeding and grazing cattle, while people starve to death, do you favor taking away goverment subsidies, so the consumer pays the full price for beef?” The survey appeared in @ national magazine and readers were requested to return their response ‘oa provided address. This survey will most likely suffer from: Sampling bias Response bias Question-wording bias only only Ul only and only 1, Mand I E.1, IT and IIL Since this was in a vegetarian magazine, meat eaters were under-represented inthe sample, which creates sampling bias. Readers who feel most strongly about the issue ‘will more likely to return the survey, resulting in response bias. The question is designed to elicit agreement with the statement, so there’s question-wording bias, too. A poll found that 81% of U.S. parents say they have spoken with their wenagers about the {dangers of drinking and driving. Only 64% of teens ofthe same families say they remember such a discussion, What type of bias is most likely at work here? Sampling bias Response bias Question-wording bias only Tonly MI onlyD. E. Answer Answer I. i 99 p> Answer Answer PLANNING A STUDY Mand If only 1, Wand I B. only. Parents are probably falling prey to the uncruthful, or socially acceptable, answer. ‘They don't want anyone to think they're bad parents. You're looking at the results of @ sample, which are several measurements all taken under the same conditions. True or False: By looking at the numbers you can tell ifthe sampling or the way the measurements were taken was biased. False, You may be able to get a few clues about potential bias this way, but you really have to examine the entire study design process to understand bias, not just look at the study resus, Virginia is going to use a systematic sample to choose a sample of 1/10 of her sophomore class. She looks in the first ten names and sees the name of her friend Elena there. She figures that Elena would like to be in the survey, So she begins her systematic sample there. ‘Then she selects every tenth name on the alphabetical list after that. Which ofthe following Statements are true? Her By including Elena, she has increased the likelihood of a response bias. sample is not a probability sample. Her sample suffers from under-coverage. Tonly only and Il Tand itl 111, and 11 C. Land I. A systematic sample is a probability sample, but only when the first name is selected randomly. Some response bias has probably been increased by choosing Elena {although the bias may be very small). Because Elena is a friend of Virginia's, she's more likely to answer the survey and to share Virginia's views. “True or False: A finding is statistically significant iF it’s unlikely that it could have occurred by chance. ‘True, When there’s a very low chance that a result could have occurred by chance. then it’s statistically significant. This indicates that something other than chance may be influencing the result. reel = LearningCHAPTER 6 Questions on Factors Causing Bias 1. You're shopping for a car, and you log on to a Web site that gives reviews for automobiles Reviews are written by average people who bought the cars and want to share their experi- ences with others. You look up the reviews for a Toyhushusta Bearcat sport truck, and find that 79% of the reviewers checked the “I would not buy this truck again” box. Based on that and nothing else, you decide not to buy the truck. Why was your decision too hasty? The poll question is leading and poorly worded, ‘The sample size is too small to make any statistical conclusions. {I's reasonable to assume that a large percentage of the respondents are lying, Undercoverage and voluntary response bias. moo > ‘There is no control condition in this study. Answer D. undercoverage and voluntary response bias. This kind of survey may be fine for giving you ideas to think about, but it’s not reliable enough to base a decision solely on it. Visitors tothe Web site don’t represent the entire population, Also, people who feel most strongly about the issue are more likely to respond. In some cases people who volunteer to respond are the ones who feel negatively about the issue, and that might explain why so many ofthe respondents wouldn’t buy the truck again. 2. Alll of the following are examples of a sample that suffers from an undercoverage bias except ‘A. _anestimate of an election outcome based on a sample of voters’ names taken from automobile registrations, 8. _apoll on Missouri dentist's attitudes, based on questionnaires sent toa registry of attendees atthe langest dental convention in the state C. the mean income of U.S. citizens, based on Census data, D. arandom sample of 5,000 U.S. adults with survey data collected via e-mail, E. _asimple random sample of licensed drivers, aken from driver-licensing records, Answer _E. a simple random sample of licensed drivers, taken from driver-licensing records. In this ‘example, the sampling frame covers the entire population: licensed drivers, 3. Consider a survey in which 79% of the respondents said yes to the following question Agree or disagree: Since our economy can’t sustain itself without a healthy environment, it’s important that Congress pass laws to protect the environment. What's a reasonable reaction to the survey results? 298 seed temoog > Answer moo > Answer moog > UDY ‘They're not valid, due to undereoverage ‘They're not valid, due to voluntary response bias They're ot valid, due to a leading question. Tally agree, We need laws that protect the environment. 1 disagree, We don't need laws that protect the environment C..they'e not valid, due to leading question. This question makes the agree position seem | more reasonable. A telephone research agency has been contracted to collect survey information on buying habits of teenagers, aged 16 and 17. The agency conducts phone calls between 3 PM. and 4:30 PM, each weekday. ‘The survey finds that the majority or teenagers aged 16 and 17 have access to a credit card. Which of the following is the most likely problem with this analysis? This is voluntary response sample, and only affluent youths will answer the survey “The respondents were not randomly selected The survey questions ate leading and influence the way afluent youths respond “The sample suffers from undercoverage due tothe hours when polling took place ‘The respondents were being untruthful D. the sample sufers from undercoverage. Many teenagers aren't at home during those hours, and many of those who are at home are likely to share traits that could bias the survey results A university involved in conducting social science experiments and surveys places an ad q vertisement in the classifieds to hire a racially and ethnically diverse set of individuals to administer surveys. The surveys will assess opinions on welfare policy and discrimination, ‘The university plans to use a random sample of these new employees to help conduct an j ‘upcoming experiment with a random sample of individuals in a major city. By hiring this di- verse group of employees, the university is hoping the reduce the problems of. in the experiment. undereoverage interviewer-induced bias a Hawthome effect reverse blocking | Voluntary response biasCHAPTER 6 Answer moop > Answer A B «. D. E Answer B. interviewer-induced bias. The concer in such an experiments that the interviewers’ race Will affect the respondents’ answers, Since responses to the topic of discrimination could potentially be affected by interviewer-induced bias, a way to reduce some of the problems. is to use a diverse mix of interviewers and to examine whether the race of the interviewer is adversely affecting the response variable, ‘You're a wildlife biologist who wants to measure the top speed of a particular species of dragonfly. You set up « measurement apparatus in a portable wind tunnel, where the insect is suspended in place while air blows over it. The wind tunnel blows air at an increasing speed, and you track the point at which the beating ofthe insect’s wings can't keep pace with the rate of air flowing over them. You plan to travel all over South America, setting up your portable wind tunnel in exotic locations, capturing insects, and measuring their maximum, speeds. ‘What is potential source of bias inthis study? Voluntary response bias A biased sample Experimenterinduced bias Undercoverage Response bits BB. a biased sample. The insects you capture are likely to be easier to catch (and possibly slower) than the population as a whole, ‘A firm administers a survey in the state of New York. The survey is mailed out toarandom set (9 5,000 households throughout the state, and responses are received from 200 households. ‘The two key questions on the survey are: 1) Whom do you plan to vote for in the upcoming Senate election, the Democratic or Republican candidate?” 2) Do you agree with the idea of imposing limits on the amount of time a Senator can spend in office? ‘What isa possible problem with this survey? Voluntary response bias Undercoverage bias Leading questions Stratified original sample Allof the above A. voluntary response bias, This is a type of response bias in which the people who choose 10 respond may hold opinions that are different from the majority of the population,moop> Answer moop> Answer 10. PLANNIN A STUDY ‘A 9th grade class is chosen for an experiment. A random half of the students receive specialized one-on-one tutoring in the classroom twice a week. The other half of the students work in study groups in the same classroom as the one-on-one students. The ‘experiment shows that test scores increased at the same rate for both the experimental and ‘control conditions. This is surprising, given that almost every other study of this specialized training showed dramatic improvements in the treatment group, What might be going on in this experiment? The blocking design was faved, The double blind procedure masked the differences Both groups experienced a placebo effect. ‘There's no real treatment ‘The treatment group was not separate enough from the control group. E. the treatment group was not separate enough from the control group. It seems the main problem in this experiment is that both groups are experiencing their “weatment” inthe same setting (the classroom), The control group (the study group students) likely see that the other students are receiving this spectal treatment and perhaps work harder to compensate for the fact that they're not receiving the specialized attention A survey is conducted ata local mall, An interviewer stops every 20th person that walks by add asks them to answer some questions about attitudes on transportation use. OF the 200, individuals who complete the survey, 80 say they prefer the bus as a transportation option. From this analysis, you have reason to conclude that: [About 40% of the sample responded to a representative survey [About 40% of the local residents prefer the bus a a transportation option [About 80% ofthe mall gers prefer the bus as transportation option About 40% ofthe mall goers prefer the bus asa transportation option About 40% of those who ride the bus shop atthe local mal D. about 40% of the mall goers prefer the bus as u transportation option. ‘The sampling method is a systematic sample, which ean represent the population reasonably well if it's «done correetly. The population for this sample is people who shop at the mall. There's no presumption that mall goers are a representative sample of the entie local population. [An experiment is conducted examining the effects of a new type of toothpaste on reducing cavities. Individuals in the treatment condition are given the new toothpaste; individuals in the control condition are given the standard toothpaste, but are told that they're receiving the new toothpaste, At the end of the experiment, the control group participants demonstrate a slightly higher decline in cavities, This is an example of a(n): seed 30 TearingCHAPTER 6 moogm > A B. c D. E. Answer oversampling effect Hawthome effect placebo effect voluntary response bias ‘wording bias C. placeboeffect. A placeboeffect occurs when one group believes that they’realso receiving the treatment in the experiment. In this instance the control group participants think they're receiving the new toothpaste, which in turn might influence their response—in this case it ‘may influence their brushing habits. Questions on Aspects of Studies ‘The idea that subjects in an experiment change their performance on a task simply because they're being observed is potential bias known as the: placebo effect. lurking variable effect. cause effect Hawthorne effect. replication effect. DD, Hawthorne effect. The Hawthorne effect describes « situation in which the response of subjects change because the subjects realize theyre being observed, A. way to alleviate this problem isto use a blind design, in which the subjets don't know whether or why they're being observed. In many cases it's impossible to do this, so the Hawthome effect i controlled fr (but not eliminated) by using contol groups or other means. The Hawthorne effect is similar tothe placebo effect, bu they aren't the same, The Hawthorne effect isa result of knowing you're being observed the placebo effects a result of knowing you're being treated. If subjects in an experiment are separated into groups of whites and non-whites prior tothe random assignment stage, we should think of race asa (an) bias, blocking variable, matched pair. explanatory variableAnswer ore 9 Answer moogmp Answer PLANNING A STUDY response variable, BB. blocking variable. When you separate subjects into groups based on some characteristic, you're creating blocks in the experiments. This approach is common when the researcher ‘thinks that the effects of the treatment are different for subjects with diferent characteristics (for example, whites vs. non-whites) Ina cluster sample: a subset of naturally occurring groups are randomly sampled, an alphabetic list of individuals is used for random selection into the sample. individuals are distinguished based on some characteristic such as gender before they're randomly selected from the population ‘oly convenient members of the population are chosen, None of the above A, a subset of naturally occurring groups are randomly sampled. A cluster is @ naturally ‘occurring group of subjects. In a clustering approach, first a random sample of the group- ings is chosen (for example, a random sample of schools), and then a random sample of individuals within each of the groups is selected. If we collect experimental data on alcohol consumption from the same 100 individuals before and then after a course on the dangers of overdrinking, we are using a: randomized block design post-test only design matched pairs design, randomized comparative design control-only design, C. matched pairs design. One version of a matched pairs design involves collecting data from the Same individuals before and after a treatment, This is known as a before and after design ‘The process of replication in an experiment often involves: ‘matching pairs on characteristics such as height and randomly assigning each pair member to experiment | and experiment 2. ‘ensuring that all laboratory animals can reproduce, so that the same experiment can be done ‘on the progenyCHAPTER 6 E. Answer pop > Answer moo > Answer climinating any interviewer bias present in the experiment. accounting for the effects of lurking variables by using @ matched pairs before and after design, repeating an experiment on different units or under different conditions. E, repeating an experiment on different units or under different conditions. Replication is ‘an important way of reducing the error due to chance in experimental studies. I's also an important way (0 see if the experiment’s results can be generalized to the population oF to other populations. ‘An experiment examines the effect of a new pill on reducing blood pressure. Half the subjects are given the new pill, the other half are given a sugar pill and told they're receiving ‘a new medicine for reducing blood pressure. The new pill is considered a (an), and the sugar pill is called a (an) control, treatment treatment, bias treatment, factor treatment, placebo placebo, control . treatment, placebo. In this experiment, the treatment is the new drug. The “ake” treatment (the sugar pill) is called a placebo. An important advantage of using «randomized block design in an experiment is it controls forthe effects of factors that may confound your results eliminating of all possible lurking variables dhat may confound the effect ofthe treatment. reducing bias associated with using multiple interviewers. collecting information from the subjects before and ate the administration of treatment All of the above A. it controls for the effects of factors that may confound your results. A block design is 1 good way to isolate the effects of the variable you want to study. For example, if you're studying the effects of a new drug for treating heart disease and you block your subjects by age and amount of exercise, you won’t have to worry about variations in age and exercise confounding the effects of the new drug.moog > Answer E. Answer 10. PLANNING A STUDY A police department commissioned a door-to-door survey of citizens on whether they com: mit minor law infractions, such as speeding or ilegal parking. The interviewers told the interviewees that they would not be investigated or ticketed for any minor infractions they may have committed, An analysis suggested that there was bias in the responses. This \would likely be an example of a (an) placebo effect. imterviewer-induced bias under-coverage bias. ‘voluntary response bias. double-blind effect. B. interviewer-induced bias, Interviewer-induced bias is likely here because many of the respondents may not want to tel a stranger (who's standing on their doorstep) about all the legal infractions they've committed. Many respondents will want to give answers that make them appear more law-abiding in the eyes of the interviewer. Which of the following is not an example of a leading question’ Do you agree with the NRA that the integrity of the U.S. constitution is beneficial to U.S. citizens and therefore that 2nd Amendment rights should be upheld? Since most homeless individuals are not intelligent, should we eliminate welfare benefits until they obtain more educational skills? How strongly do you agree that drugs are a problem in our public schools? ‘The World Trash Organization found that 80% of the trash in landiills could be recycled. Given this fact, should we fine companies that dump recyclable trash in local landfills? All are leading questions. how strongly do you agree that drugs are a problem in our public schools? This question js the most neutral of the bunch because it doesn’t include statements that make one response seem more reasonable than the others, You construct a sample in which whites, blacks, and Latinos are randomly selected from the U.S. adult population. ‘The composition of your final sample is 25% white, 30% black, and 30% Latino, You've sampled from all over the country, but the actual composition of the population has a much higher proportion of whites. This is likely an example of a: ‘multi-stage cluster sample. convenience sample, eel ed LearningCHAPTER 6 C. simple random sample . voluntary response sample, E. stratified random sample. Answer _E. stratified random sample. We'd expect that the overall population from which the sample is chosen would be comprised mostly of whites (much more than 25% of adult population nationwide). Since the sample has disproportionately high percentages of blacks and Lati ‘os, it's likely thatthe population was frst stratified based on race before random selection ‘This approach is often used when we believe that a simple random sample will draw too few individuals of a certain characteristic (such as blacks of Latinos). 306 Learning