Homework 1

1. A psychologist wants to know if adults with normal vision can be fooled by a certain optical illusion. She recruits 50 students from her PSY 120 class and finds that 42 of them are fooled by the illusion. The population in this study is all adults with normal vision. 2. When we take a census, we attempt to collect data from every individual in the population. 3. A 2008 national sample survey interviewed 1007 people age 18 and older nationwide by telephone. One question asked was whether on the whole, would you say you are satisfied or dissatisfied with the way the nation is being governed? The population for this sample survey appears to be all adult residents of the United States. 4. You want to know the opinions of American school teachers about establishing a national test for high school graduation. You obtain a list of the members of the National Education Association (the largest teachers' union) and mail a questionnaire to 2500 teachers chosen at random from this list. In all, 1347 teachers return the questionnaire. In this situation, the population is all American school teachers. 5. See Above. The sample is the 1347 teachers who mail back the questionnaire. 6. In a (n) _____________, the environments of the subjects are controlled or manipulated by the researcher. experiment 7. The General Social Survey (GSS), conducted by the National Opinion Research Center at the University of Chicago, is a major source of data on social attitudes in the United States. Once each year, 1500 adults are interviewed in their homes all across the country. The subjects are asked their opinions about sex and marriage, attitudes toward women, welfare, foreign policy, and many other issues. The population for the GSS is all adult residents of the United States. 8. See Above. The questions on the GSS are considered variables. 9. A Gallup Poll asked 1002 randomly chosen adults to rate the honesty and ethical standards of 21 occupations. Nurses were at the top, with 82% of the poll respondents rating them very high or high. Telemarketers were at the bottom, with just 7% very high or high ratings. The population for this poll is all adults. 10. A sportswriter wants to know how strongly Columbus, Ohio, residents support building a new stadium for the local minor league baseball team, the Columbus Clippers. She stands outside the stadium before a game and interviews the first 25 people who enter the stadium. The population for this survey is all residents of Columbus, Ohio. 11. See Above. The sample for the survey is the 25 people who gave the sportswriter their opinion. 12. Can pleasant aromas help people work more efficiently? Researchers did this study to find out. Twenty-two students worked a paper-and-pencil maze six times. On three tries, they wore a mask with floral scents. On the other three tries, they wore a mask with no scent. The six tries were done in random order and each used a different maze. The researchers found that the subjects took less time to complete the maze when wearing the scented mask. This study is a (n) experiment. 13. See Above. The population for this study is people. 14. See #11. The response variable for this study is time to complete maze. 15. The administration of Virginia Commonwealth University has been asked to extend the Drop/Add Period to two weeks (instead of the current one week period). VCU will do so if it

16. 17. 18.

19. 20.

is convinced that more than half of the student body is in favor of this change. Of the 1500 students surveyed, 803 are in favor of extending the Drop/Add Period as proposed. The study is a (n) sample survey. See Above. The population for this study is all VCU students. The response variable for this study is opinion toward proposed change. The New York Giants' 2117 victory over the New England Patriots in Super Bowl XLVI, which was watched by approximately 111.3 million viewers, became the most-watched broadcast in television history up to that point. But how many of those viewers were actually more interested in watching the commercials than the game? In a simple random sample of 800 Super Bowl XLVI viewers, 380 said that they were more interested in watching the broadcast for the commercials than in watching because of the actual game. This study is a (n) sample survey. See Above. The population for this study is 111.3 million viewers The response variable for this study is reason for watching broadcast.

Homework 2
1. The student newspaper runs a weekly question that readers can answer online or by campus mail. One question was Do you think the college is doing enough to provide student parking? Of the 82 people who responded, 79% said No. When we say that the newspaper poll is biased, we mean that repeated polls would give results that are very different from each other. 2. In a table of random digits, any pair of entries is equally likely to be any of the onehundred possible pairs 00, 01, 02, ..., 99. 3. Every conceivable group of people of the required size has the same chance of being the selected sample when we use a simple random sample. 4. A table of random numbers is used to select 30 students from a statistics class to rate a statistics video. These 30 students are a simple random sample of the class. 5. A table of random numbers is used to select 30 students from a statistics class to rate a statistics video. The ratings that these students give are used to estimate the ratings that would be given if the entire class were asked to rate the video. This type of sample is a sample that avoids bias. 6. For a sample to be a simple random sample of size n, every collection of n individuals must have the same chance to be the sample actually chosen. 7. When Ann Landers asked her readers to tell her if your sex life has gone downhill after marriage, more than 100,000 people responded. This is an example of a voluntary response sample 8. You want to take an SRS of 50 of the 816 students who live in a college dormitory. You label the students 001 to 816 in alphabetical order. In the table of random digits you read the entries 96746 12149 37823 71868 18442 35119 62103 39244 The first three students in your sample have labels 461, 214, 718. 9. See Above. Another valid choice of labels for the 816 students is 001 to 816 in order of the student ID numbers and 000 to 815 in alphabetical order. 10. Which of these statements about the table of random digits is true? None of these is true

11. Which of the following is not true of a simple random sample of size 1000 chosen from a population of size 4 million? Every individual of the population has chance 1 in 1000 of being included in the sample. 12. A table of random numbers is used to select 25 students from a large class to rate a CD by country singer Taylor Swift. The ratings that these students give are used to estimate the ratings that would be given if the entire class were asked to rate the CD. This type of sample is a simple random sample 13. Bias in a sampling method is any systematic error that tends to occur in the same

direction every time you use this sampling method

14. The mayor's office wants to know how people in the city feel about the condition of the city's roads. They place an announcement in the newspaper asking residents to email their opinions to the mayor's office. Five hundred people send emails, and about 84% of the responses indicate displeasure with the condition of the city's roads. This is a voluntary response sample. It will almost certainly overestimate the level of displeasure among the city's residents. 15. A professor surveys all of her 63 students to inquire about their understanding of a recently presented topic. These 63 students the population. 16. Suppose you want to take a simple random sample of size 6 from the 20 participants in your Zumba exercise class. You label the students 01 to 20 in alphabetical order by last name. In the table of random digits, you read the entries 45149 32992 75730 66280 03819 56202 02938 70915 The six participants in your selected sample have labels 14, 06, 03, 02, 09, 15. 17. See Above. Another valid choice of labels for the 20 participants is Answers A, B, and C are correct. 36 to 55 in alphabetical order by last name or 01 to 20 in reverse alphabetical order by last name or five labels for each person (0004, 0509, 1014, , 9599) assigned in alphabetical order by last name 18. A local talk radio station conducts a poll to determine if its listeners favor or oppose the President's proposed tax plan. To express their opinions, listeners are asked to call or textmessage the radio station. The poll results in 71.82% of the responders opposing the proposed tax plan. In this situation, 71.82% is likely to overestimate the true percentage since typically

only people with strong (and usually negative) opinions will respond.
19. This type of sampling uses the idea of drawing names out of a hat to produce a sample of individuals. Simple random 20. For a class project, you want to survey students at your school for their opinions about the importance of studying. You go to the campus library and survey two hundred students as they are leaving the library. What type of sample was used in this situation? This is a convenience sample.

Homework 3
1. The student newspaper runs a weekly question that readers can answer online or by campus mail. One question was Do you think the college is doing enough to provide student parking? Of the 82 people who responded, 79% said No. The number 79% is a statistic. 2. A Gallup Poll asked, What is your favorite Summer Olympics event? Of the people asked, 27% said swimming. Here is what Gallup says about the accuracy of this poll: The results are based on telephone interviews with a sample of 1007 national adults, aged 18 years and older, conducted July 2527, 2008. For results based on the total sample of national adults, one can say with 95% confidence that the maximum error attributable to sampling and other random effects is plus or minus 3 percentage points. In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls. In Gallup's statement, 95% confidence means (assuming there is no bias) that if Gallup repeated this poll many times, 95% of the results would be within plus or minus 3 percentage points of the truth about the population. 3. See Above. Suppose 500 people in the sample of 1007 adults were women. Gallup asked these 500 women, What is your favorite Summer Olympic event? Thirty-five percent said swimming was their favorite. Gallup gave a margin of error for this result. Applying the quick method, we find that the margin of error for 95% confidence changes when the sample size drops from 1007 to 500 from: 3.2% to 4.5%. 4. Do doctors in managed care plans give less charity care? Researchers chose 60 communities at random, then chose doctors at random in each community. In all, they interviewed 10,881 doctors. Overall, 77.3% of the doctors said they had given some care free or at reduced rates because of the patient's financial need in the month before the interview. Doctors who received at least 85% of their practice income from managed care plans were significantly less likely than other doctors to provide charity care. Some doctors who did not give any charity care may say that they did. If so, the study suffers from bias: The sample result will systematically overestimate the true percent of doctors who give charity care. 5. For a simple random sample of size 10,881, the margin of error for 95% confidence is about 0.96%.

6. A Census Bureau report on the income of Americans says that with 90% confidence the median income of all California households in 2007 was $67,484, with a margin of error of $375. This means that the Census Bureau got the result $67,484 $375 using a
method that will cover the true median income 90% of the time when used repeatedly 7. Increasing the size of an SRS has these beneficial effects: The margin of error is smaller than it is for smaller SRSs.

8. To reduce the variability of estimates from a simple random sample, you should use a larger sample. 9. Sale of eggs that are contaminated with salmonella can cause food poisoning among consumers. A large egg producer takes an SRS of 200 eggs from all the eggs shipped in one day. The laboratory reports that 9 of these eggs had salmonella contamination. Unknown to the producer, 0.1% (one-tenth of one percent) of all eggs shipped had salmonella. In this situation, 0.1% is a parameter and 9 is a statistic. 10. When constructing a 95% confidence interval, the margin of error means that, if there is no bias, about 95% of all samples taken will come at least this close to the truth. 11. If a Gallup Poll surveys a national sample of 3000 people rather than 1500 people, the margin of error of the sample result would be less than it would be for 1500 because the sample is now larger. 12. A Gallup Poll recently showed that 69% of Americans believe there is a hell. (The question was, Please tell whether hell is something you believe in, something you're not sure about, or something you don't believe in.) The poll contacted 1003 adults by telephone. The margin of sampling error was plus or minus 3 percentage points. Assume a 95% confidence level and no bias. From the information given above, you can make which of the following statements? (Note that 69% 3% is 66% to 72 %.) We are 95% confident (but not certain) that between 66% and 72% of all American adults believe there is a hell. 13. The Gallup organization conducts a poll to see how much support there is for a constitutional amendment guaranteeing the right to abortion. They announce that 23% of Americans are in favor of such an amendment with margin of error of 3.4% for 95% confidence. If the results of the above poll were described using a 90% confidence statement, what would happen to the margin of error? It would be less than 3.4%. 14. __________ in a sampling method means that the sample results will systematically misrepresent the population in the same way when we take repeated samples. Bias 15. Which of the following is correct? Statistics describe sample characteristics 16. A recent survey of 35,101 randomly selected U.S. adults studied the religious affiliation

of Americans. The survey found that nationwide 23.9% said that they were Catholic. But among Latinos, 55% were Catholic. The sample included 4212 Latinos. The margin of error in a 95% confidence interval for the proportion of all adults who are Catholic is larger for Latinos than for the nation.
17. A recent survey of 35,101 randomly selected U.S. adults studied the religious affiliation of Americans. Of those questioned in the poll, 78.4% said that they were Christians. The number 78.4% is a statistic. 18. Does coaching raise SAT scores? Because many students score higher on a second try even without coaching, a study looked at a simple random sample of 500 students who had taken coaching courses between their two attempts at the SAT. The study looked at the proportion of students who saw an increase in scores (out of the total possible score of 2400). We are 95% confident that between 65.5% and 74.5% of students who are coached raise

their SAT scores. If we want to be 99% confident, the range of points would be wider because higher confidence requires a larger margin of error. 19. For a simple random sample of size 4761, the margin of error for 95% confidence is about 1.45%. 20. A local planning commission is interested in finding out what proportion of its city's residents are opposed to constructing a new baseball stadium in the downtown area. A random sample of 1870 residents is obtained, and 41.2% of them are opposed to the stadium. In this sample, the margin of error is 2.3%.

Homework 4
1. Scotland is considering independence from England. An opinion poll showed that 51% of Scots favor independence. Another poll taken at the same time showed that only 34% favored being separate from England. The primary reason these results differ by so much is that the wording of questions has a big effect on poll results. 2. We divide the class into two groups: first-year students and others. We then take random samples from each group. This is an example of stratified random sampling 3. You want to know the opinions of American school teachers about establishing a national test for high school graduation. You obtain a list of the members of the National Education Association (the largest teachers' union) and mail a questionnaire to 2500 teachers chosen at random from this list. In all, 1347 teachers return the questionnaire. The sampling frame is all members of the National Education Association. 4. Which of the following is a source of nonsampling error in a sample survey? None of the above 5. Some common sources of nonsampling error in samples of human populations are some

subjects lie; some subjects can't be contacted

6. Increasing the size of a probability sample decreases the variability of a sample statistic. 7. To obtain student opinions on a particular issue, students were identified by major, and five students were randomly selected from within each major to be interviewed. This is an example of stratified sampling. 8. An overnight opinion poll calls randomly selected telephone numbers. This polling method misses all people without a phone. Which of the following types of errors is present in this situation? Sampling 9. Professor Lilli Gans wishes to study the relationship between a person's zodiac sign and his or her political opinions. She obtains the birthdays of all 816 students enrolled in her Astrology 101 course. Then she uses a random digits table to choose 10 students from each of the twelve zodiac signs. For example, students born from March 21 until April 19 have the Aries zodiac sign, and Professor Gans chooses 10 Aries students at random. She uses a different part of the table of random digits to choose students having each sign. After obtaining her sample, Professor Gans has the selected students fill out a questionnaire. Professor Gans has obtained a stratified random sample, where the strata are students with different zodiac signs

10. One of the questions in Professor Gans's questionnaire asks whether the student has ever given money to the American Nazi Party. Any students who have done so are likely to lie and say No, rather than admit the truth. This is an example of nonsampling error 11. When we take a census, we collect data from every individual in the population. 12. A sample that is chosen in stages is a sample in which we choose the final sample in several steps, for example, first states, then counties in those states, then households in those counties 13. A poll conducted telephone interviews with 1497 adults. These 1497 people make up the sample 14. A poll questioned 1025 women and 472 men. The design of the sample chose separate samples of men and women and planned to interview more women than men. This is a stratified random sample. 15. Some common sources of nonsampling error in samples of human populations are can't contact some subjects; some subjects refuse to answer 16. One source of error in pre-election polls is that some people in the sample say they will vote, but they end up not voting. This is an example of a nonsampling error. 17. Which of the following is NOT a potential problem with Internet surveys? Multimedia content 18. A probability sample is a sample that is chosen by chance. 19. The announced margin of error for a sample survey covers random sampling error 20. Bias can result from how the sample was chosen or how the survey question was worded

Homework 5
1. Does using a cell phone while driving make an accident more likely? Researchers compared telephone company and police records to find 699 people who had cell phones and were also involved in an auto accident. Using phone billing records, they compared cell phone use in the period of the accident with cell phone use the same period on a previous day. Result: The risk of an accident was four times higher when using a cell phone. observational study, but not a simple random sampling 2. The explanatory variable in this study is whether the subject was using a cell phone 3. A researcher studied whether meeting regularly with a weight-loss counselor helps dieters lose weight. Eighty dieters were available for the study. Half the dieters, assigned at random, received weight-loss counseling. The other half did not. After six months, the weight lost by each dieter was measured. On the average, those who met regularly with a weight-loss counselor lost more weight than those who did not randomized comparative experiment. 4. Does coaching raise SAT scores? Because many students score higher on a second try even without coaching, a study looked at an SRS of 4200 students who took the SAT twice. Of these, 500 had taken coaching courses between their two attempts at the SAT. The study compared the average increase in scores (out of the total possible score of 2400) for students who were coached with the average increase for students who were not coached. Because students' SAT scores vary a lot when they retake the test, the difference between the average increases for coached and uncoached students was not statistically significant. This means that a difference this big could easily occur just by chance even if coaching really has no effect

5. Do doctors in managed care plans give less charity care? Researchers chose 60 communities at random, then chose doctors at random in each community. In all, they interviewed 10,881 doctors. Overall, 77.3% of the doctors said they had given some care free or at reduced rates because of the patient's financial need in the month before the interview. Doctors who received at least 85% of their practice income from managed care plans were significantly less likely than other doctors to provide charity care. This study is a(n) sample survey. 6. Students in a large statistics class were randomly divided into two groups. The first group took the midterm exam in the morning with a symphony by Mozart playing in the background while the second group took the exam in the afternoon with rock music playing. The scores of the two groups on the exam were compared. This study is a(n) randomized comparative experiment 7. In this study, the explanatory variable is the type of music that was playing during the exam 8. An example of a lurking variable that might affect the results of this study is the time of day test was administered 9. A newspaper article was headlined, Marrying Young Can Be Beneficial. The article said that marriage is a good thing when it comes to cutting back on drinking and drugs. A University of Michigan Institute for Social Research study of 33,000 young adults showed that young, unmarried adults usually increased their alcohol, marijuana, and cocaine use when they left home, often to attend college. Those same people, however, decreased their drug and alcohol use when they got engaged or married. Couples who lived together but were not engaged or married showed no such drop in drug use. The strongest reason for having reservations about the claim made in the first sentence of the quote is that it's an observational study 10. The explanatory variable in this study is marital status 11. Does taking large amounts of vitamins protect against cancer? To study this question, researchers enrolled 29,000 Finnish men, all smokers over the age of 50. Half of the men, selected at random, took vitamin supplements, and others took a dummy pill that has no active ingredient. The researchers followed all the men for eight years. At the end of the study, men in the vitamin group were no less likely to have cancer than men in the other group. This study cast doubt on the popular idea that taking lots of vitamins can reduce the risk of cancer. The study design looked like this:

The label for the dummy pill treatment should appear in the outline at the point marked Question B. What is this label? Placebo 12. The response variable should be named in the outline at Question C. The response variable in this study is whether a subject developed cancer 13. A researcher claims that the mean resting pulse rate of all college basketball players in the United States is less than the mean resting pulse rate of all professional basketball players in the United States. The resting pulse rates of a random sample of 115 college basketball players were measured as were the resting pulse rates of a random sample of 80 professional basketball players. The mean resting pulse rates of the two groups were compared. This study is a(n) observational study In this study, the response variable is the resting pulse rate An example of a lurking variable that might affect the results of this study is the age of the players A psychologist recently said that, For relatively mild medical problems, the placebo effect will produce positive results in roughly two-thirds of patients. The placebo effect is the effect of a dummy treatment on a patient An experiment on the effect of vitamin A on cancer uses two randomly chosen groups of 200 men each, one given vitamin A and the other a placebo. An estimate from a similar experiment using two groups of 1000 men each would have less variability A recent issue of the New England Journal of Medicine reports a study of all 122,754 infants born over an 8.5-year period at Parkland Hospital in Dallas, leaving out multiple births and infants with birth defects. The researchers wanted to know if there is a specific birth weight below which infant death and illness increases sharply.

14. 15. 16.

17.

18.

This study is a(n) observational study, but not a probability sample 19. The explanatory variable in the study is birth weight 20. Does taking vitamins prevent colon cancer? A study assigned 864 subjects at random to four groups. One group took beta-carotene, another took vitamins C and E, a third took all three, and the fourth group took only a dummy pill. After four years, there was no difference among the groups in the formation of polyps in the colon that precede cancer. Does the new study (properly designed) give good reason to think that vitamins prevent colon cancer? No, there was no significant difference among the groups

Homework 6
1. A study of a drug to prevent hair loss showed that 86% of the men who took it maintained or increased the amount of hair on their heads. But so did 42% of the men in the same study who took a placebo instead of the drug. This is an example of the placebo effect. 2. To control for the power of suggestion when a subject takes an experimental drug, use a placebo. 3. Students in a large statistics class were randomly divided into two groups. The first group took the midterm exam with a symphony by Mozart playing in the background, while the second group took the exam with rock music playing. The scores of the two groups on the exam were compared.

This experiment was not double-blind because the students knew the type of music that was playing while they were taking the exam 4. It is difficult to establish the causal link between cigarette smoking and lung cancer because All of the above 5. A professor believes that students who smoke cigarettes tend to have lower grades. He collects data from 1326 randomly selected students at his university and discovers that, on an average, students who smoke cigarettes do indeed tend to have lower grade point averages than students who do not smoke. This study was based on a probability sample 6. Does this study give strong evidence that smoking cigarettes causes lower grades? No, because people who make bad decisions may choose not to study well and may choose to smoke cigarettes 7. Which of the following statements about a randomized block design with two treatments is not true? Block A is chosen randomly from among the available subjects 8. To see if eating just before going to bed causes nightmares, volunteers are recruited to spend the night in a sleep laboratory while connected to machines that monitor their sleep patterns. They are randomly assigned to be given a meal before bed or not. Numbers of nightmares are recorded and compared for the two groups. The critics claim the study gives results that do not generalize. 9. In a randomized block design the blocks often represent levels of a lurking variable suspected of being confounded with the explanatory variable.

Homework 7
1. Ethical standards for randomized, controlled clinical trials include not asking subjects to agree to participate without first informing them of the nature of the study and possible risks and benefits 2. Study in which information about individual subjects is never made public is said to be confidential 3. The Dean of Students at Mountain Tech is interested in finding out how prevalent cheating is. She sends a questionnaire to 1000 randomly chosen students and receives 700 replies. Those asked to fill out the questionnaire are told not to put their names on either the questionnaire or on the return envelope. Thus, no one in the Dean's office has any idea who sent in which response. After the replies are processed, statistics describing the entire sample are published, but no one outside the Dean's office is allowed to see individual questionnaires. This survey is both anonymous and confidential 4. Before any study that uses human subjects can be done at a university, the institutional review board must agree that subjects are adequately protected from possible harm. 5. Suppose you want to survey students to find out what proportion have illegally used prescription drugs. In order to maximize the number of truthful responses, your survey should be anonymous. 6. A student asks 27 of her friends to tell her their grade point averages (GPA). This student writes down only the answers and not the name of the friend with whom she is speaking, and she reports on the average GPA of her friends. This survey is considered to be confidential

7. The student-run newspaper asks students to visit a web page and respond to questions regarding a proposed tuition increase. Only responses to the questions are recorded. Summary statistics based on the survey responses are used in an article published the following week, and no one outside of the newspaper has access to the individual responses. The newspaper's survey is considered to be both anonymous and confidential 8.

Notes
1. nonadherer: A subject who participates in an experiment but does not follow the experimental treatment. 2. dropout: A subject who begins an experiment but does not complete it. 3. Sample surveys suffer from nonresponse due to failure to contact some people selected for the sample and the refusal of others to participate. Experiments with human subjects suffer from similar problems 4. explanatory variable: A variable that we think explains or causes changes in the response variable 5. statistically significant: When differences among the effects of the treatments are so large that they would rarely happen just by chance. 6. completely randomized design: Experimental design where all the experimental subjects are allocated at random among all the treatments. 7. random: Phenomenon when individual outcomes are uncertain but there is nevertheless a regular distribution of outcomes in a large number of repetitions. 8. placebo: A dummy treatment with no active ingredients 9. completely randomized design: Experimental design where all the experimental subjects are allocated at random among all the treatments 10. matched pairs: A type of experiment that compares just two treatments and combines matching with randomization. Each subject receives both treatments in a random order, or the subjects are matched in pairs as closely as possible, and one subject in each pair receives each treatment. 11. block: A group of experimental subjects that are known before the experiment to be similar in some way that is expected to affect the response to the treatments. In a block design, the random assignment of subjects to treatments is carried out separately within each block. Blocks are another form of control. 12. variable: Any characteristic of an individual. It can take different values for different individuals. 13. placebo effect: When an individual responds to a dummy treatment, perhaps due to the individuals expectation that the treatment will produce an influential outcome. 14. clinical trial: A medical experiment involving human subjects.

15. doubleblind experiment: An experiment where neither the subjects nor the people who work with them know which treatment each subject is receiving. The double-blind method helps achieve a basic requirement of comparative experiments: equal treatment for all subjects except for the actual treatments the experiment is comparing. The most common weakness in experiments is that we cant generalize the conclusions widely. Some experiments apply unrealistic treatments, some use subjects from some

special group such as college students, and all are performed at some specific place and time. We want to see similar experiments at other places and times confirm important findings. Many experiments use designs that are more complex than the basic completely randomized design, which divides all the subjects among all the treatments in one randomization. Matched pairs designs compare two treatments by giving one to each of a pair of similar subjects or by giving both to the same subject in random order. Block designs form blocks of similar subjects and assign treatments at random separately in each block. The big ideas of randomization, control, and adequate numbers of subjects remain the keys to convincing experiments.
16. randomized comparative experiment: An experiment in which two or more treatments are compared and chance is used to decide which subjects get each treatment. In this type of experiment, enough subjects are used so that the effects of chance are small. 17. bias: Consistent, repeated deviation of the sample statistic from the population parameter in the same direction when we take many samples.

18. nonresponse: The failure to obtain data from an individual selected for a sample. This could be because the subject does not cooperate, or because the subject could not be contacted. 19. Data ethics begin with some principles that go beyond just being honest. Studies with human subjects must be screened in advance by an institutional review board. All subjects must give their informed consent before taking part. All information about individual subjects must be kept confidential.