likely to occur when conditions havenbeen bad but are rapidly improving.nWhen conditions are bad and stay bad,npeople take their misfortune for granted,nbut when conditions suddenly improvenpeople develop higher aspirationsnand become easily frustrated.nComment: This is an excellentnquestion. It is precisely the sort ofnquestion that the test promises but failsnto deliver throughout: it attacks a beliefnthat really is held by the student (thenworse the conditions, the more likely isnrevolution), surprises him with the correctnanswer, and prepares him for anvaluable sociological finding (rising expectationsnoutpace improvement inncondition). It is not coincidental, however,nthat this question and misconceptionnhave to do with a correlationn(improvement and revolution) muchnfurther removed from the student’sndaily observation than are the subjectsnof most of the other questions. Thenmore abstract the correlation in question,nthe more likely that the studentnholds an incorrect beliefnQ. Lower-class youths are morenlikely to commit crimes than middleclassnyouths?nA. Lower-class youths are not morenlikely than middle-class youths to commitncrimes. Middle-class youths are atnleast as likely to engage in delinquentnacts, but they are less likely to benarrested, and therefore do not show upnas frequently in the court statistics.nComment: To most beginning students,nas to the general population,n”crime” means “violent crime” or, atnmost, “violent crime and major whitecollarncrime.” The student is hardlyngoing to be knocked off his chair by then56/CHRONICLESnPLAY BALLnLIBERAL ARTSnfact that for every mugging by a membernof the lower class there is a marijuananjoint smoked by a high-school student.nIn other words, this questionncould achieve its goal of surprising thenstudent (and giving a correct answer)nonly by showing that middle- andnlower-class youths commit equally seriousncrimes at equal rates. And, ofncourse, it cannot show this because it isnnot true.nQ. The best way to get an accuratenassessment of public opinion is to pollnas many people as possible?nA. The number of people involvednin a public opinion poll is largelynirrelevant. What matters is that thensample should be fully representativenof the population whose opinion isnwanted. A property chosen sample ofntwo or three thousand Americans canngive a highly accurate test of nationalnopinion; a poorly chosen sample ofnthree million, or even 30 million, couldnbe hopelessly off target.nComment: This answer is true onlynin the sense that it would be correct tonsay that a football player’s size is unimportantnto his ability because large, butnuncoordinated, people are less likely tonmake the National Football Leaguenthan are superb athletes of average size.nCleariy, when we say that size is importantnto football ability we mean that,nother things being equal, size is important.nLikewise, when we say that a largensample is better than a small sample,nwe mean “when members of the samplesnare equally representative.” And,nin saying this, we are correct; in thisnsense — the only sense in which thenclaim makes any sense — the larger thenAccording to a new “ex-gay” effort called Evergreen, sports isnthe way out of homosexuality. As reported in the summernnewsletter of Evangelicals Concerned. Evergreen recentlynsponsored a two-day conference in Salt Lake City entitledn”You Don’t Have to be Gay.”, Evergreen director AlannSeegmiller said that in addition to teaching basketball andnSoftball to those who want to become “ex-gay,” some havenfound it helpful to seek training in automobile mechanics.”nnnsample, the better.nQ. People who are regular Christiannchurchgoers are less likely to be prejudicednagainst other races than peoplenwho do not attend church?nA. Regular churchgoers are generallynnot less prejudiced than nonchurchgoers;nin fact, they tend to be morenprejudiced.nComment: This question and answernseeM to address the question ofnthe effect of religion (or at least ofnchurchgoing) on prejudice, a questionnof monumental scientific, moral, andnpolitical importance.nThe question and answer actuallyngiven simply reflect the fact that Protestantnchurchgoers tend (statisticallynspeaking, as always) to come fromngroups tending, for reasons havingnnothing to do with Christianity ornchurchgoing, to be more prejudicedn(rural Southern Fundamentalists, asnopposed to urban nonbelievers, fornexample). This no more demonstratesnthat churchgoing increases prejudicenthan the fact that Japanese churchgoersnare shorter than Canadian nonchurchgoersndemonstrates that going tonchurch makes you short.nThe important question that thenanswer seems to address but does not isnthe effect on prejudice of churchgoing ^inwhen all other variables are heldnsteady. In other words, one does notnwant to compare churchgoers fromnBirmingham with nonchurchgoersnfrom New York, but churchgoers fromnBirmingham with nonchurchgoersn(matched for race, income, class, residentialnarea, and the like) from Birminghamn(or churchgoers from NewnYork with nonchurchgoers from NewnYork). It is the answer to this questionnthat is likely to demonstrate whethernchurchgoing has the effect of increasingnprejudice, decreasing it, or leavingnit unaffected.nQ. The number of federal governmentnemployees has grown sharplynover the past two decades?nA. The number of federal civiliannofficials has remained almost constantnfor twenty years, although the numbernof state and local-government officialsnhas risen significantiy.nComment: The intuition of thenstudent (and the rest of us) is thatnbureaucracy has burgeoned. That thisnintuition derives from our experiencenwith state and local bureaucracies, andn