A research hypothesis is a prediction of the outcoe of a study! The prediction ay
"e "ased on an educated #uess or a fora$ theory! E%ap$e & is a hypothesis for a none%perienta$ study! E%ap$e & It is hypothesi'ed that first #rade #ir$s (i$$ sho( "etter readin# coprehension than first #rade "oys! In E%ap$e &) the author is predictin# that he or she (i$$ find hi#her coprehension aon# #ir$s than "oys! To test it) a none%perienta$ study (ou$d "e appropriate "ecause nothin# in the hypothesis su##ests that treatents (i$$ "e #i*en! A sip$e research hypothesis predicts a re$ationship "et(een t(o *aria"$es! +ro your study of *aria"$es) it shou$d "e c$ear that the t(o *aria"$es in E%ap$e & are ,&- #ender and ,.- readin# coprehension! The hypothesis states that readin# coprehension is re/ $ated to #ender! E%ap$e . is a hypothesis for an e%perienta$ study! E%ap$e . It is hypothesi'ed that chi$dren (ho arc sho(n a *ideo (ith i$d *io$ence (i$$ "e ore a##ressi*e on the p$ay#round than those (ho are sho(n a sii$ar *ideo (ithout the *io$ence! In E%ap$e .) the independent variable is *io$ence ,i$d *s! none-) and the dependent variable is a##ressi*eness on the p$ay#round! The hypotheses in E%ap$es & and . are e%ap$es of directiona$ hypotheses! In a direc/ tiona$ hypothesis) (e predict (hich #roup (i$$ "e hi#her or ha*e ore of soethin#! Soeties (e ha*e a nondirectiona$ hypothesis! Consider E%ap$e 0! E%ap$e 0 It is hypothesi'ed that the chi$d/rearin# practices of Tri"e A are different fro those of Tri"e 1! The author of E%ap$e 0 is sayin# that there (i$$ "e a difference "ut does not predict the direction of the difference! This is perfect$y accepta"$e (hen there is no "asis for a2in# an educated #uess! Instead of a nondirectiona$ hypothesis) (e i#ht state a research purpose! E%ap$e 3 sho(s a research purpose that corresponds to the nondirectiona$ hypothesis in E%ap$e 0! E%ap$e 4 The purpose is to e%p$ore the differences in chi$drearin# practices "et(een Tri"e A and Tri"e 1! A research 4uestion ay a$so "e su"stituted for a nondirectiona$ hypothesis! E%ap$e 5 sho(s a research 4uestion that corresponds to the nondirectiona$ hypothesis in E%ap$e 0! E%ap$e 5 The research 4uestion is 6Ho( do the chi$drearin# practices in Tri"e A and Tri"e 1 differ78 9hen usin# a research 4uestion as the "asis for research) (e usua$$y shou$d "e carefu$ not to state it as a 4uestion that can "e ans(ered (ith a sip$e 6yes8 or 6no)8 as is done in E%ap$e :! E%ap$e : The 4uestion is) 6Do the chi$d/rearin# practices in Tri"e A and Tri"e 1 differ78 E%ap$e : ere$y as2s 6do they differ78 This is not a *ery interestin# research 4uestion! E%ap$e 5 is superior "ecause it as2s how do they differ78 The choice "et(een a nondirectiona$ hypothesis) a research purpose) and a research 4ues/ tion) is pure$y a atter of persona$ taste; a$$ are accepta"$e in the scientific counity! Of course) (hen (e are (i$$in# to predict the outcoe of a study) (e shou$d state a directiona$ hypothesis! If you<*e read research reports) you ay ha*e encountered references to another type of hypothesis;the nu$$ hypothesis! This is a statistical hypothesis, (hich (i$$ "e e%p$ored ne%t! E=ERCISE &! 9hich type of stateent ,hypothesis) purpose) or 4uestion- predicts the outcoe of a study7 .! 6It is hypothesi'ed that co$$e#e students (ho ha*e fir career #oa$s achie*e hi#her >PAs than those (ho do not ha*e fir career #oa$s!8 Is this a directiona$ or nondirectiona$ hypothesis7 0! 9ou$d an e%perienta$ or none%perienta$ study "e "etter for testin# the hypothesis in 4uestion .7 3! 6It is hypothesi'ed that chi$dren of ii#rants and chi$dren of nati*e "orn citi'ens (i$$ differ in their attitudes to(ard schoo$!8 Is this a directiona$ or nondirectiona$ hypothesis7 5! 6The #oa$ of this study is to e%aine co$$e#e students< attitudes to(ard re$i#ion!8 Is this stateent a hypothesis or purpose7 :! 6Are chi$dren of a$coho$ics different in their socia$ ad?ustent than chi$dren of non/ a$coho$ics78 Is this research 4uestion stated appropriate$y7 9hy7 9hy not7 @! 9hen (e are (i$$in# to predict the outcoe of a study) shou$d (e state a directiona$ or nondirectiona$ hypothesis7 Questions for Discussion A! Restate this hypothesis as a research purposeB 6It is hypothesi'ed that there is a difference in ?o" satisfaction "et(een those (ho recei*e re#u$ar feed"ac2 on their ?o" perforance and those (ho recei*e irre#u$ar feed"ac2!8 C! Is the hypothesis in 4uestion C directiona$ or nondirectiona$7 E%p$ain! &D! Cou$d an e%perient "e conducted to test the hypothesis in 4uestion A7 E%p$ain! &&! Restate this hypothesis as a research 4uestionB 6It is hypothesi'ed that those (ho e%ercise re#u$ar$y and those (ho do not e%ercise re#u$ar$y (i$$ differ in other "eha*iors that affect hea$th!8 ANS9ERS &! hypothesis .! directiona$ 0! none%perienta$ 3! nondirectiona$ 5! purpose :! no) "ecause it is stated in a (ay that can "e ans(ered yes or no @! directiona$ A! Sap$e ans(erB The purpose is to e%p$ore differences in ?o" satisfaction "et(een those (ho recei*e re#u$ar feed"ac2 on their ?o" perforance and those (ho recei*e irre#u$ar feed"ac2! C! nondirectiona$ "ecause it does not predict (hich #roup (i$$ ha*e #reater ?o" satisfaction &D! Sap$e ans(erB yes) "ecause (e cou$d physica$$y anipu$ate the re#u$arity of feed"ac2 &&! Sap$e ans(erB Ho( do those (ho e%ercise re#u$ar$y and those (ho do not differ in other "eha*iors that affect hea$th7 INTRODUCTION TO THE NUEE HYPOTHESIS Suppose (e dre( rando sap$es of en#ineers and psycho$o#ists) adinistered a se$f/ report easure of socia"i$ity) and coputed the ean ,the ost coon$y used a*era#e- for each #roup! +urtherore) suppose the ean for en#ineers is :5!DD and the ean for psycho$o#ists is @D!DD! 9here did the fi*e point difference coe fro7 There are three possi"$e e%p$anationsB &! Perhaps the popu$ation of psycho$o#ists is tru$y ore socia"$e than the popu$ation of en#ineers) and our sap$es correct$y identified the difference! ,In fact) our research hypothesis ay ha*e "een that psycho$o#ists are ore socia"$e than en#ineers ; (hich no( appears to "e supported "y the daTA- .! Perhaps there (as a "ias in procedures! 1y usin# rando sap$in#) (e ha*e ru$ed out sap$in# "ias) "ut other procedures such as easureent ay "e "iased! +or e%ap$e) ay"e the psycho$o#ists (ere contacted durin# Dece"er) (hen any socia$ e*ents ta2e p$ace and the en#ineers (ere contacted durin# a #$ooy +e"ruary! The on$y (ay to ru$e out "ias as an e%p$anation is to ta2e physical steps to pre*ent it! In this case) (e (ou$d (ant to a2e sure that the socia"i$ity of "oth #roups (as easured in the sae (ay at the sae tie! 0! Perhaps the popu$ations of psycho$o#ists and en#ineers are the sae "ut the sap$es are unrepresentati*e of their popu$ations "ecause of rando sap$in# errors! +or in/ stance) the rando dra( ay ha*e #i*en us a sap$e of psycho$o#ists (ho are ore socia"$e) on the a*era#e) than their popu$ation! The third e%p$anation has a nae ; it is the nu$$ hypothesis! The #enera$ for in (hich it is stated *aries fro researcher to researcher! Here are three *ersions) a$$ of (hich are consistent (ith each otherB Fersion A of the nu$$ hypothesisB The observed difference was created by sampling error. ,Note that the ter sampling error refers on$y to random errorsnot errors created "y a "ias!- Fersion 1 of the nu$$ hypothesisB There is no true difference between the two groups. ,The ter true difference refers to the difference (e (ou$d find in a census of the popu$ations) that is) the difference (e (ou$d find if there (ere no sap$in# errors!- Fersion C of the nu$$ hypothesisB The true difference between the two groups is zero. Significance tests deterine the pro"a"i$ity that the nu$$ hypothesis is true! ,9e (i$$ "e considerin# the use of specific si#nificance tests in future (ee2s!- Suppose for our e%a/ p$e) (e use a si#nificance test and find that the pro"a"i$ity that the nu$$ hypothesis is true is $ess than 5 in &DDG this (ou$d "e stated as p H!D5) (here p o"*ious$y stands for probability. Of course) if the chances that soethin# is true are $ess than 5 in &DD) It<s a #ood "et that it<s not true! If it<s pro"a"$y not true) (e reect the null hypothesis, $ea*in# us (ith on$y the first t(o e%p$anations that (e started (ith as *ia"$e e%p$anations for the difference! There is no ru$e of nature that dictates at (hat pro"a"i$ity $e*e$ the nu$$ hypothesis shou$d "e re?ected! Ho(e*er) con*entiona$ (isdo su##ests that !D5 or $ess ,such as !D& or !DD&- is reasona"$e! Of course) researchers shou$d state in their reports the pro"a"i$ity $e*e$ they used to deterine (hether to re?ect the nu$$ hypothesis! Note that (hen (e fai$ to re?ect the nu$$ hypothesis "ecause the pro"a"i$ity is #reater than .!5, (e do ?ust thatB (e 6fai$ to re?ect8 the nu$$ hypothesis and it stays on our $ist of possi"$e e%p$anationsG (e never 6accept8 the nu$$ hypothesis as the on$y e%p$anation ; ree"er) there are three possi"$e e%p$anations and fai$in# to re?ect one of the does not ean that you are acceptin# it as the on$y e%p$anation! An a$ternati*e (ay to say that (e ha*e re?ected the nu$$ hypothesis is to state that the dif/ ference is statistically significant. Thus) if (e state that a difference is statistica$$y si#nificant at the !D5 $e*e$ ,eanin# !D5 or $ess-) it is e4ui*a$ent to statin# that the nu$$ hypothesis has "een re?ected at that $e*e$! 9hen you read research reported in acadeic ?ourna$s) you (i$$ find that the nu$$ hy/ pothesis is se$do stated "y researchers) (ho assue that you 2no( that the so$e purpose of a si#nificance test is to test a nu$$ hypothesis! Instead) researchers te$$ you (hich differences (ere tested for si#nificance) (hich si#nificance test they used) and (hich differences (ere found to "e statistica$$y si#nificant! It is ore coon to find nu$$ hypotheses stated in theses and dissertations since coittee e"ers ay (ish to a2e sure that the students they are super*isin# understand the reason they ha*e conducted a si#nificance test! As (e consider specific si#nificance tests in other chapters of the te%t) (e<$$ e%aine the nu$$ hypothesis in ore detai$! E=ERCISE &! Ho( any e%p$anations (ere there for the difference in socia"i$ity "et(een psycho$o#ists and en#ineers in the e%ap$e in this topic7 .! 9hat does the nu$$ hypothesis say a"out sap$in# error7 0! Does the ter sampling error refer to random errors or to bias" 3! The nu$$ hypothesis says that the true difference e4ua$s (hat *a$ue7 5! 9hat is used to deterine the pro"a"i$ities that nu$$ hypotheses are true7 :! +or (hat does p H!D5 stand7 @! Do (e re?ect the nu$$ hypothesis (hen the pro"a"i$ity of its truth is hi#h or (hen it is $o(7 A! 9hat do (e do if the pro"a"i$ity is #reater than !D57 C! 9hat is an a$ternati*e (ay of sayin# that (e ha*e re?ected the nu$$ hypothesis7 &D! Are you ore $i2e$y to find a nu$$ hypothesis stated in a ?ourna$ artic$e or in a thesis7 ANS9ERS &! three .! the o"ser*ed difference (as created "y sap$in# error 0! rando errors 3! 'ero 5! si#nificance tests :! pro"a"i$ity that the nu$$ hypothesis is true $ess than 5 in &DD @! (hen it is $o( A! fai$ to re?ect the nu$$ hypothesis C! a difference is statistica$$y si#nificant &D! thesis