17.6 Return to the study fi rst described in Question 16.5 on page 336, where a psychologist tests whether shy college students initiate more eye contacts with strangers because of training sessions in assertive behavior. Use the same data, but now assume that eight subjects, coded as A, B, . . . G, H, are tested repeatedly after zero, one, two, and three training sessions. (Inci-dentally, since the psychologist is interested in any learning or sequential effect, it would not make sense—indeed, it’s impossible, given the sequen-tial nature of the independent variable—to counterbalance the four ses-sions.) The results are expressed as the observed number of eye contacts:
| WORKSHOP SESSION
SUBJECT ZERO ONE TWO THREE SUBJECT A 1 2 4 7 14 B 0 1 2 6 9 C 0 2 3 6 11 D 2 4 6 7 19 E 3 4 7 9 23 F 4 6 8 10 28 G 2 3 5 8 18 H 1 3 5 7 16 G5 138
|
- Summarize the results with an ANOVA table. Short-circuit computa-tional work by using the results in Question 16.5 for the SS terms, that is, SS between 5 154.12, SS within 5 132.75, and SS total 5 286.87.
- Whenever appropriate, estimate effect sizes with □ 2p and with d , and conduct Tukey s HSD test. (c) Compare these results with repeated measures with those in Question 16.5 for independent samples.
- Compare these results with repeated measures with those in Question 16.5 for independent samples.
17.7 Recall the experiment described in Review Question 16.11 on page 314, where errors on a driving simulator were obtained for subjects whose orange juice had been laced with controlled amounts of vodka. Now assume that repeated measures are taken across all fi ve conditions for each of fi vesubjects. (Assume that no lingering effects occur because sufficient time elapses between successive tests, and no order bias appears because the orders of the fi ve conditions are equalized across the fi ve subjects.)
| DRIVING ERRORS AS A FUNCTION OF ALCOHOL CONSUMPTION
(OUNCES) SUBJECT ZERO ONE TWO FOUR SIX T A 1 4 6 15 20 46 B 1 3 1 6 25 36 C 3 1 2 9 10 25 D 6 7 10 17 10 5 E 4 5 7 9 9 34 T 15 20 26 56 74 SX 5 G 5 191 SX2 5 2371 |
- Summarize the results in an ANOVA table. If you did Review Ques-tion 16.11 and saved your results, you can use the known values for SS between , SS within , and SS total to short-circuit computations. (b) If appropriate, estimate the effect sizes and use Tukey’s HSD test.
17.8 While analyzing data, an investigator treats each score as if it were contrib-uted by a different subject even though, in fact, scores were repeated mea-sures. What effect, if any, would this mistake probably have on the F test if the null hypothesis were (a) true? (b) false?
*18.8 For the two-factor experiment described in the previous question, assume that, as shown, mean bar press rates of either 4 or 8 are identified with three of the four cells in the 2 3 2 table of outcomes.
| FOOD
DEPRIVATION (Hrs.) 0 24 Reward 1 Amount 2 (Pellets) |
| 8
4 8 |
Furthermore, just for the sake of this question, ignore sampling variability and assume that effects occur whenever any numerical differences corre-spond to either food deprivation, reward amount, or the interaction. Indi-cate whether or not effects occur for each of these three components if the empty cell in the 2 3 2 table is occupied by a mean of (
- a) 12
(b) 8
(c) 4
18.11 In what sense does a two-factor ANOVA use observations more efficiently than a one-factor ANOVA does?
18.12 A psychologist employs a two-factor experiment to study the combined effect of sleep deprivation and alcohol consumption on the performance of automobile drivers. Before the driving test, the subjects go without sleep for various time periods and then drink a glass of orange juice laced with controlled amounts of vodka. Their performance is measured by the num-ber of errors made on a driving simulator. Two subjects are randomly assigned to each cell, that is, each possible combination of sleep depriva-tion (either 0, 24, 48, or 72 hours) and alcohol consumption (either 0, 1, 2, or 3 ounces), yielding the following results:
(a) Using the .05 level of significance, test the null hypothesis that survival rates are independent of the passengers’ accommodations (cabin or steerage).
(b) Assuming a signifi cant c2 , estimate the strength of the relationship.
(c) To more fully appreciate the importance of this relationship, calculate an odds ratio to determine how much more likely a cabin passenger is to have survived than a steerage passenger.
19.14 In a classic study, Milgram et al . “lost” stamped envelopes with fi ctitious addresses (Medical Research Association, Personal Address, Friends of Communist Party, and Friends of Nazi Party).* One hundred letters with each address were distributed among four locations (shops, cars, streets, and phone booths) in New Haven Connecticut, with the following results:
| NUMBER OF DRIVING ERRORS
SLEEP DEPRIVATION ALCOHOL HOURS CONSUMPTION (OUNCES)0 24 48 72 Trow T²row
0 0 2 5 5 29 841
1 1 3 6 5 36 1296 3 3 7 8 2 3 2 8 7 53 2809 5 5 11 12
3 4 4 10 9 68 4624 6 7 13 15
SX² 5 1466 T column 25 30 64 67 G5 186 T2column 625 900 4096 4489 G25 3496 |
(a) Summarize the results with an ANOVA table.
(b) If appropriate, conduct additional F tests, estimate effect sizes, and use Tukey’s HSD test.
21.4 DISTINGUISHING BETWEEN THE TWO TYPES OF DATA The distinction between quantitative and qualitative observations is crucial, and it is usually fairly easy. First, a l ways make the distinction between qua n titative and qualitative data . In those cases where you feel uncomfortable about making this distinction, consider the following guidelines.
Focusing on a Single Observation When you have access to the original observations, focus on any single observation. If an observation represents an amount or a count, expressed numerically, it is quantitative; if it represents a class or category, described by a word or a code, it is qualitative.
Focusing on Numerical Summaries When you do not have access to the original observations, focus on any numeri-cal summaries specified for the data. If means and standard deviations are specified, the data are quantitative; if only frequencies are specified, the data are qualitative.
Focusing on Key Words When you have neither access to the original observations nor numerical sum-maries of data, as in the case of the questions at the end of this chapter, read the description of the study very carefully, attending to key words, such as scores or means, which, if present, typify quantitative data, or, if absent, typify qualitative data. If All Else Fails If all else fails, try visualizing the value of a single observation, whether a meaningful number (quantitative) or a word or numerical code (qualitative), based on any information given. A careful evaluation, combined with an occa-sional speculation, usually reveals whether data are quantitative or qualitative.
21.7 An investigator wishes to test whether creative artists are equally likely to be born under each of the 12 astrological signs.
21.8 To determine whether there is a relationship between the sexual codes of primitive tribes and their behavior toward neighboring tribes, an anthro-pologist consults available records, classifying each tribe on the basis of its sexual codes (permissive or repressive) and its behavior toward neighboring tribes (friendly or hostile).
21.12 Over a century ago, the British surgeon Joseph Lister investigated the relationship between the operating room environment (presence or absence of disinfectant) and the fate of about 100 emergency amputees (survived or failed to survive
21.13 A comparative psychologist suspects that chemicals in the urine of male rats trigger an increase in the activity of other rats. To check this hunch, she randomly assigns rats, in equal numbers, to either a sterile cage, a cage sprayed with a trace of the chemicals, or a cage sprayed thor-oughly with the chemicals. Furthermore, to check out the possibility that reactions might be sex-linked, equal numbers of female and male rats are assigned to the three cage conditions. An activity score is recorded for each rat during a 5-minute observation period in the specified cage.
21.14 A psychologist wishes to evaluate the effectiveness of relaxation training on the subsequent performance of college students in a public speaking class. After being matched on the basis of the quality of their initial speeches, students are randomly assigned either to receive relaxation training or to serve in a control group. Evaluation is based on scores awarded to students for their speeches at the end of the class.
21.15 An investigator wishes to determine whether, for a random sample of drug addicts, the mean score on the depression scale of a personality test differs from that which, according to the test documentation, represents the mean score for the general population.
21.16 Another investigator wishes to determine whether, for a random sample of drug addicts, the mean score on the depression scale of a personality test differs from the corresponding mean score for a random sample of nonaddicted people.
21.17 To determine whether cramming can increase GRE scores, a researcher randomly assigns college students to either a specialized GRE test-taking workshop, a general test-taking workshop, or a control (non-test-taking) workshop. Furthermore, to check the effect of scheduling, students are randomly assigned, in equal numbers, to attend their workshop either during a marathon weekend or during a series of weekly sessions.
21.18 A criminologist suspects that there is a relationship between the degree of structure provided for paroled ex-convicts (a supervised or unsuper-vised “rehab” house) and whether or not there is a violation of parole during the fi rst 6 months of freedom.
21.19 A psychologist uses chimpanzees to test the notion that more crowded living conditions cause aggressive behavior. The same chimps live in a succession of cages containing one, several, or many other chimps. After several days in each cage, chimps are assigned scores on the basis of their aggressive behavior toward a chimplike stuffed doll in an observation cage.
21.20 In an extrasensory perception experiment involving a deck of special playing cards, each of 30 subjects attempts to predict the one correct pattern (on each playing card) from among fi ve possible patterns during each of 100 trials. The mean number of correct predictions for all 30 subjects is compared with 20, the number of correct predictions per 100 trials on the assumption that subjects lack extrasensory perception.
21.21 A social scientist wishes to determine whether there is a relationship between the attractiveness scores (on a 100-point scale) assigned to college students by a panel of peers and their scores on a paper-and-pencil test of anxiety.