Developmental Psychology Copyright 1984 by the1984, Vol 20, No 4. 653-666 American Psychological Association. Inc

Pace and Continuity of Television Programs: Effects onChildren's Attention and Comprehension

John C. Wright, Aletha C. Huston, Rhonda P. Ross, Sandra L. Calvert,David Rolandelli, Lee Ann Weeks, Pouran Raeissi, and Richard Potts

Center for Research on the Influences of Television on Children (CRrrc),Department of Human Development, University of Kansas

Children's tv programs can be distinguished by the processing demands of theirformat and the pace with which changes occur. High-continuity programs arestories requiring temporal integration of successive scenes for full comprehension.Low-continuity programs are in "magazine" format: Successive bits are independentof one another, and temporal integration across bits is not required for compre-hension. Pace is denned as rate of scene and character change in stones and rateof bit change in magazine shows. Sixteen children's tv programs varying in continuity(high vs. low), pace (high vs low), and animation (cartoon vs. live production)were made from broadcast material. Each lasted 15 minutes. Children (N = 160),half in grades K-l and half in grades 3-4, viewed two of the programs and werethen tested for recall. The recall task required sequential senation of still photostaken from the program. Older children attended longer and reconstructed sequencesbetter than younger children. High-continuity (story) programs led to greater at-tention and better recall than low-continuity (magazine) programs. Low-pacedshows were recalled better than high-paced shows. Older children recalled bestwhen shown either low pace or story format or both. Young children showedadditive increments in recall due to low pace and high continuity: Either alonewas better than none, and both was better than either alone. Younger childrenattended somewhat more to high- than to low-paced shows if they were in magazineformat, a weak effect of perceptual salience. Older children attended to high-con-tinuity programs in synchrony with their pace: longer looks to low-paced andshorter looks to high-paced programs. This effect was attributed to schematic andstrategic processing by older children when the format justified it. Regression analysesindicated higher correlations between attention and recall for animated stories thanfor other types of programs, an effect attributed to their relatively high stereotypyin the medium. Age, continuity, and pace effects on recall were not fully accountedfor by their effects on attention. Results were interpreted as indicating evidencefor development of active, schematic processing of television by children and forstrategic attending by older children, based on perceived processing demands.

Rapid pace is often used in both commercial characterized by very rapid rates of changeand educational television to attract and hold (Huston et al., 1981). Many educational pro-children's attention. Cartoons and other com- grams, such as Sesame Street and The Electricmercial children's programs are short and are Company, have also adopted rapidly paced

formats comprised of short "bits" on the as-— — sumption that young children have brief at-

This research was made possible by a grant from the tention spans and need frequent change toSpencer Foundation, which is gratefully acknowledged, maintain their interest. Indeed many of theThe authors would also like to express their gratitude to c r i t i c i s m s o f fo^ programs are based on thethe teachers, administrators, students, and parents of the , , Z ~~ _Uoii~,,Baldwin Elementary School, Baldwin, Kansas. We also Premise that such formats encourage shallowthank Mabel L. Rice, Alice Leary Reitz, Valeria Lovelace, or passive processing, reduced mental effort,and Ellen Wartella for their assistance short attention span, and a taste for percep-

Requests for reprints should be sent to John C. Wnght, t u a j gimmickry, all of which may be inimicalCenter for Research o ^ e Influence of Television on Chil- t h o u g h t f u l processing and constructivedren, Department of Human Development, University of . ^Kansas, Lawrence, Kansas 66045. learning.

653

654 WRIGHT ET AL.

The effects of pace on cognitive processingare of theoretical as well as applied interest.We have proposed earlier (Wright & Huston,1981) that very young children's attention totelevision is governed more by the perceptualsalience of the audiovisual events than by theviewer's agenda for information search andmeaningful interpretation of messages. Per-ceptual salience refers to such properties asBerlyne's (1960) "collative properties"—in-tensity, contrast, change, movement, novelty,or incongruity—which are exemplified inchildren's television by such features as rapidpace (rate of scene and character change), highrate of action (physical movement of charactersthrough space), frequent use of special effects,auditory and visual embellishment, and thelike. The perceptual salience model suggeststhat for young children at least, high pace,with its short bits and frequent changes, di-rectly elicits and maintains children's attentionmore effectively than low pace. The saliencehypothesis is meant to describe initial percep-tually based determinants of attention, and isexpected to account for less attentional vari-ance as the child's cognitive level and viewingexperience come to play more important rolesin sustaining attention and interest (Huston& Wright, 1983; Wright & Huston, 1983).

Although there is empirical support for theeffects of many salient television features onattention, existing data do not provide strongsupport for the effects of rapid pace, per se,on attention. Anderson and Levin (1976) ini-tially found that children paid more visualattention to short bits than to long bits whilewatching Sesame Street, a magazine formatprogram consisting mostly of unconnectedbits. However Lorch, Anderson, and Levin(1979) found no such differences when theycompared specially constructed Sesame Streetprograms composed of long versus short bits(equivalent to low and high pace, respectively).

In a pilot study (Wright, Calvert, Huston-Stein, & Watkins, 1980), children attended tohigh-paced cartoons more than to low-pacedcartoons, but when the content was changedto a magazine format, segments taken fromThe Electric Company, pace had no effect onvisual attention. Moreover, children in middleelementary school were more attentive to thehigh-paced cartoons than were preschoolers,an age trend opposite to that predicted.

One reason for the inconsistent effects of

pace on attention is suggested by more mi-croscopic analysis. Pace is actually composedof changes to a new scene (not previouslyshown in the program), changes to a familiarscene, and changes in the cast of characterspresent. These changes are often marked byvisual features, such as fades, dissolves, zooms,or pans, or by auditory features, such aschanges in music, sound effects, or speech.Thus changes of scene sometimes elicit atten-tion from a child who is not looking, but fora viewer who is attending, scene and characterchanges can sometimes prompt a temporaryinterruption of looking (Alwitt, Anderson,Lorch, & Levin, 1980; Wright et al., 1980).Older children, presumably better able to parsethe flow of a program, used scene-changepoints as the occasion for both starting andstopping a look more regularly than did youn-ger children in one of these studies, suggestingthat they were more aware of the formal fea-ture cues marking a breaking point in the pro-gram. It appears that at least for school-agechildren, scene and character changes are at-tentional decision points. Therefore, for them,rapid pace may lead to frequent shifts in at-tention, rather than to longer durations oflooking (Huston & Wright, 1983).

One formulation (Huston-Stein & Wright,1979), proposed a developmental shift fromperceptual salience as a basis for attentionaldecisions toward search for information basedon processing the content of the program. Thatis, the cognitive-processing requirements of theprogram should become increasingly impor-tant for attention as children become older,more cognitively sophisticated, and more ex-perienced with television.

Two other theoretical models concerning therelation of cognitive processing to attentionhave been proposed. Krull and Husson (1979)suggested that high pace and unpredictabilityof content would increase the processing de-mands of the program, resulting in cognitivearousal, a prediction they confirmed with re-spect to autonomic activity while viewing byadolescents. They predicted and found greaterattention to high- than to low-paced programs,an effect they attributed to the higher levelsof arousal induced by the former.

Anderson and his colleagues (1981) haveproposed an alternative model in which at-tention is determined by the perceived com-prehensibility of the content being presented.

CHILDREN'S TV: PACE AND CONTINUITY 655

Both the content itself and certain accompa-nying features can serve to signal whether aprogram is intended for children and thuswhether a child is likely to be able to under-stand it. If the scenes and characters in a pro-gram change rapidly, a child may not be ableto make the connections and inferences nec-essary for temporal integration and logicalcomprehension of the narrative. If it is all hap-pening too fast there may be a correspondingreduction in comprehensibility, with a con-sequent loss of attention relative to a lowerpaced program of otherwise equal difficulty.The comprehensibility hypothesis appears topredict exactly the opposite of the salience andarousal hypotheses regarding the effects of paceon attention: that low-paced programs, beingeasier in general to understand, should receivemore attention than high-paced programs, asthe latter prove to be less comprehensible,especially for younger children. It is likely thatsalience and cognitive-processing requirementsboth play a role in determining attention, foras Salomon (1981) has shown, these same at-tributes in concert with the child's expectationscan determine the level of mental effort in-vested in processing television.

In the present study, we investigate the hy-pothesis that certain aspects of a program de-termine the kind and depth of processing thata child invests in understanding it. When thecontent is not cumulative, that is, when littlelogico-temporal integration of content mes-sages appears to be required, then children'sattention and recall should be more influencedby superficial perceptual attributes of the pro-gram. But when more cumulative processingis called for, signaled by cues that events seenand heard are interconnected and require log-ico-temporal integration, then attention andsubsequent processing should be governedmore by the informational needs of the childas he or she makes an effort to understand(cognitive arousal), and more by the successshe or he experiences in so doing (compre-hensibility), but less by the superficial percep-tual attributes. High pace should enhance at-tention in the former case (low continuity ofcontent), but whether high or low pace wouldlead to greater attention in the latter case (highcontinuity) is not clear, as reasonable predic-tions can be made in both directions.

The two kinds of processing just describedcorrespond to Mandler's (1979) distinction

between schematic and categorical organiza-tion. A magazine format program, that is, onecontaining separate bits that are essentially self-contained and could logically be presented inany order, does not lend itself to organizationby a schema based on temporal or logical con-nections among its parts. It probably elicitsdiscrete or noncumulative processing of bits,each enjoyed for its self-contained attributes,but not integrated into a larger meaningfulmessage. To be sure, each bit may require itsown internal integrative processing, but witheach bit change there is a fresh start.

However, much of television consists of high-continuity stories that do lend themselves toschematic processing (Collins, 1982). Under-standing a television story is inherently a mat-ter of processing sequenced, connected, time-locked event structures and narratives (Wright,Watkins, & Huston, 1978). The connectednarrative or story format both demands effortfor children to achieve temporal integrationof its plot and offers them an opportunity toexercise existing schemes and scripts drawnfrom their real-world and television knowledgebases to help them prestructure, anticipate,rehearse, and organize the content into a co-herent tale.

Collins and his associates have delineatedmany of the cognitive processes children useto understand narrative, portrayed on televi-sion (Collins, 1982; Collins, Sobol, & Westby,1981). They have suggested that children underage 8 or 9 do not readily achieve logico-tem-poral integration of televised stories. Youngchildren do not make inferences that serve toconnect explicitly presented bits of informa-tion with content that is implied but notshown. Nor do they make appropriate infer-ences about characters' feelings and motivesas explanations for their behavior. AlthoughCollins and his colleagues have identified manyof the cognitive processes involved in under-standing televised stories, it seems likely thatthey have underestimated young children'sabilities by conducting their research withadult programs. Not only are the story scriptscomplex (e.g., they have extensive subplots),but they also require real-world knowledge thatyoung children do not have (e.g., understand-ing the interpersonal implications of testifyingagainst someone in court).

In the present study, we investigated thepossibility that the effects of pace on attention

656 WRIGHT ET AL.

are different in stories than in magazine formatprograms. That is, rapid versus slow paceshould have different effects when children areengaging in processing that is noncumulative.Within each type of program, pace could in-fluence attention and processing on the basisof its perceptual salience or on the basis ofthe processing demands it makes on the viewer.But it is likely that salience should have stron-ger effects for magazines (low continuity) thanfor stories (high continuity). In addition, wehypothesized that older children (9 and 10)are likely to be less responsive to salience thanyounger children (5 to 7). Older children, ofcourse, may also be better able to discriminatethe two formats and their differential pro-cessing demands. Thus, the formats may havemore distinctive effects on their attention thanon that of younger children.

Children's attention to programs varying inpace (rate of scene and character change) andcontinuity (story vs. magazine format) wasmeasured. In addition, recall of the temporalsequence of program events was measured asa test of the amount of schematic processingthe child had done. If children organized storymaterial schematically as they viewed, thenrecall of the temporal sequence should be bet-ter for stories than for magazine formats, asthe latter cannot as readily be organized intotemporal schemes. If children did not useschematic processing, then recall for the twotypes of programs should not differ. Two agegroups (kindergarten and first grade vs. thirdand fourth grades) were compared to assessdevelopmental differences.

Table 1Rates Per Minute of Scene and CharacterChanges for Programs Varying in Pace,Continuity, and Animation

Method

Subjects

Participants were 160 children who attended a publicschool in a small midwestern city They were equally dividedby age (kindergarten and first grade vs. third and fourthgrades) and by sex.

Television Programs

Sixteen programs were selected from a library of ap-proximately 150 broadcast children's television programsthat had been scored for formal features Eight high-con-tinuity programs were stories designed for children. Eachhad a relatively well-developed plot, no programs consistingof simple chase plots or disconnected sight gags were in-cluded. All stories were originally made for commercial

Continuity

LowAnimatedLive

HighAnimatedLive

LowAnimatedLive

HighAnimatedLive

Bifchange

.34

.45

1.251 30

Scenechange

Low pace

2.27.65

1.25.95

High pace

3.293.23

5.165.01

Characterchange

3.121.37

3.231.31

5.153.56

5 455.36

Total

5.392.02

4.482.26

8.446.79

10.6110.37

• only applies to magazine (low-continuity) format

television and lasted 20 to 25 mm in their broadcast version(one lasted 50 minutes). Each was edited so that it lastedapproximately 15 minutes, and so that each story plotwas meaningfully developed and resolved within the 15-mm edited version, including titles and credits.

Of the eight high-continuity programs, four were highpaced and four were low paced Pace was denned as therate of scene and character changes. Half the programsat each pace level were animated and half were live. Becauseanimated programs have a higher average level of pacethan live programs, the absolute levels of pace for theanimated and live programs differed. Nevertheless therewere marked differences between the high- and low-pacedprograms. Moreover, editing was used to further increaseor reduce pace for the high- and low-paced programs,respectively. The mean rates of scene and character changefor each level are shown in Table 1.

The low-continuity programs were magazine formatshows in which the individual "bits" had no causal re-lationship to one another and no logical sequence. Fourof the original programs from which they were taken wereoriginally made for commercial broadcast and four forpublic television. High and low pace were denned by thelength of bits. For the high-paced shows a series of shortbits was edited together, for the low-paced shows a seriesof long bits was assembled. As much as possible, selectionswere made that contained relatively few or no scene changeswithin bits. Again, half the shows within each level of pacewere animated and half were live. Each show lasted ap-proximately 15 min in its edited version, including be-ginning and ending titles and credits. The rates of sceneand character change and the rates of bit change are alsoshown in Table 1.

CHILDREN'S TV: PACE AND CONTINUITY 657

Experimental Design

The 16 programs formed four cells of a design thatcrossed high-low continuity (story vs. magazine) with high-low pace. Within each cell two programs were animated(Programs A and B) and two were live (Programs C andD). Each child was assigned to see two programs (oneanimated and one live) from one of the four cells of thepace by continuity design. Within each continuity by pacecell, half the subjects saw Programs A and C; the otherhalf saw Programs B and D. Thus the complete experi-mental design included Age (2) X Sex (2) X Continuity(2) X Pace (2) X Pairs (5) as between-subjects variables,and animation (2) as a withm-subjects variable. Programset (A-C vs B-D) was nested within continuity and pace.

ProcedureChildren were brought to a two-room mobile laboratory,

parked on school grounds, in same-sex pairs. When theyentered the viewing room, they were told that they wouldsee some television. They were seated facing each otherat a small table, arranged so that each child's chair wasat a 90° angle to the color tv monitor (though the chairscould easily be turned by the children, and often were)Paper, crayons, magic markers, and a variety of small toyswere on the table. The participants were told that theycould play with any of the materials, talk, or watch TV,but they were asked to remain seated at the table Thepurpose of the toys and of viewing in pairs was to provideenough alternative activities so that attention would varyand to make the situation similar in some respects tonatural home viewing. Children were told that they wouldbe asked some questions about the programs later. Theadult experimenter announced that he or she would be inthe next room if needed and went into the control room.

Shortly after the adult left the room, the first programbegan. When the first program ended, the adult enteredthe viewing room very briefly to ask if they were com-fortable or needed anything. She or he then explained thatanother program would start soon and returned to thecontrol room. The second program was then shown Withineach cell of the design the order in which the two programswere shown was counterbalanced across pairs of subjects

Children's visual orientation to the monitor was scoredcontinuously as off or on by two observers (one assignedto each child) on a Datamyte recorder. The one-way mirrorthrough which the observer watched was located directlybehind, and just above the tv monitor The time of eachonset and offset of looking at the screen was recordedautomatically. Videotapes of all sessions were made In afew cases, attention was rescored from the tapes becauselive scoring had been disrupted or because data were lostfrom the Datamyte. Observer agreement was assessed byhaving two observers score videotaped records indepen-dently. Agreement occurred when both observers codedan onset or an offset within 4.8 s of one another Agreementwas 96%, using the following formula, number of agree-ments X 2, divided by the total number of scores by bothobservers. Subsequent reliability checks on live scoringhave yielded 97% agreement.

The following three attention scores were calculated foreach child: (a) duration—the total looking time dividedby the length of the program. The score can be thoughtof as percentage of time spent looking, (b) frequency—the total frequency of looking (number of looks) per minute,

(c) duration per look—the average length of looks, dennedas the total looking time divided by total frequency

Comprehension of Temporal Sequences

At the end of the second program, tests of comprehensionwere administered to each child individually. The seriationtask was to arrange photos in the order in which theyappeared in the program. Seriation measures of compre-hension were selected because the main theoretical focusof the study was the child's recall of logical and temporalsequences within programs. Seriation measures also havethe advantage of not requiring verbal description by theexperimenter or by the child. Two levels of seriation weretested" whole-program seriation and segment seriation.

Whole-program seriation For each of the 16 programs,five still photographs were made representing scenes sam-pled from throughout the program. The scenes were chosento be distinctive and to represent central (as contrastedwith incidental) program events For the low-continuityprograms, each photograph was taken from a differentbit. For the high-conunuity programs, each depicted adifferent scene in the story.

Segment seriation One segment, 1 to 3 min long, wasselected from each of the 16 programs for a mini-seriationitem. Five photographs were taken from within this seg-ment For low-continuity programs, the segment-seriationitems were taken from a single bit.' For the high-continuityprograms, they were taken from one portion of the story,comprising one or two contiguous scenes.

Administration All seriation sets were administeredafter children had viewed both programs, because testingafter the first program might have influenced attention tothe second program One of the children was invited intothe control room for an interview with one adult, whilea second adult interviewed the other child in the viewingroom

Six seriation sets were administered to each child. Thefirst two were the whole program and segment-seriationsets, respectively, from the last program the child had seenduring the viewing session Hence, they represented fairlyimmediate recall Next the whole-program and segment-senation sets from ihejirst program shown in the viewingsession were administered Thus, they measured delayedrecall, with possible interference from the intervening tele-vision viewing and testing For each set the child was shownthe pictures in scrambled order and asked to arrange themin the same order in which they occurred within the tele-vision programs. Children were encouraged to verbalizethe basis for their picture-arrangement decisions.

Finally, each child responded to the whole-program andsegment-seriation sets for one additional program that heor she had not seen The purpose of this procedure wasto determine the effects of cues in the pictures, previousfamiliarity with the program, or other factors extraneousto the experimental viewing experience These scores wereused to control for possible differences in difficulty amongthe seriation sets for different programs The "unseen"program was always selected from the same pace and con-tinuity cell of the design, as were the programs that the

' Except for one animated and one live program in thelow-continuity, high-paced cell, where they were taken fromadjacent bits.

658 WRIGHT ET AL.

I Duration of attention

PACE

CONTINUITY

AGE

High Low High Low

Magazine 1 Story

Youngir Subjects

High

1 Magazir

Low High Lowi

w | Story

Okler Subject!

Figure I Attention to programs as a function of age, pace, and continuity. (Difference between meansmarked "a" is of borderline significance, p < .10 Difference between means marked " b " is significant,p < .05.)

child had seen When children were given these final "un-seen" seriation items, the adult explained that they camefrom a program the child had not seen but that the childshould try to arrange them as she or he thought they wouldoccur in a television program.

Scoring The senation score for each set was calculatedby comparing the temporal sequence of the pictures withthe correct sequence on two criteria: how close each picturewas to its correct absolute position and how many pictureswere sequenced correctly, regardless of absolute position.For this purpose, the pictures were correctly numberedfrom 1 to 5. Then the seriauon score was calculated asfollows, (a) For each picture in turn, one point was givenfor every picture with a lower number that was placed tothe left of it The maximum number of possible pointswas 10. This part of the score correlates very highly (.98)across all possible sequences, with scores derived by cal-culating the rank-order correlation (rho) between eachorder and the correct one. Such an index, however, basedsolely on the absolute position of each picture, fails to givesufficient credit for correct partial sequences that are mis-placed For example, the response sequence 23451 receivesa rho of 0, whereas the sequence 13254 receives a rho of.80. Therefore a second component of the seriation scorewas devised, (b) The second part of the score consisted ofone point for each pair that was placed in correct adjacentorder, regardless of overall position. The maximum numberof points on this part was 4. The correlation of this partwith rho across all possible sequences is only moder-ate— 45.

The total senation scores were calculated by adding thetwo parts. Although the maximum number of possiblepoints is 14, all 14s were reduced to 12, because obtainedscores of 12 and 13 are numerically impossible. The finaltotal scores were normally distributed. The total score forthe sample sequence 23451 is 9, and for the sample se-quence 13254 it is 8. The correlation of total seriationscore with rho across all possible sequences is .89.

Results

Attention to Programs

Initial analyses indicated that frequency andduration of looking were normally distributed,but duration per look was positively skewed.Accordingly, duration per look scores weresubmitted to a log transformation, which nor-malized their distribution. Initial analyses alsoindicated no important effects of the order inwhich programs were seen, and so orders wereignored in all subsequent analyses.

Each measure of attention was submittedto an analysis of variance (ANOVA) in whichbetween-subjects effects were Age (2) X Sex(2) X Pace (2) X Continuity (2) X Pairs (5),and the within-subjects effect was animation(2). Pairs entered as a random variable; there-fore error terms for all other effects were basedon pair scores. Within each level of the Pace XContinuity X Animation classification, the twoprograms (A vs. B, or C vs. D) were not dis-tinguished. Hence, any differences attributableto the content of particular programs withincells contributed to error variance.

Duration of looking. The proportion oftime spent looking at the monitor is shown inFigure 1. Children spent more time lookingat high-continuity programs than at low-con-tinuity programs, F(l, 64) = 6.24, p < .05.They looked 52% of the time during the high-

CHILDREN'S TV: PACE AND CONTINUITY 659

continuity programs and 44% of the time dur-ing the low-continuity programs. There wereno significant main effects or interactions as-sociated with program pace, but a plannedcontrast between attention to high- and low-paced magazine programs for younger childrenapproached significance (p < .10). For them,high-paced magazine shows got more attentionthan low-paced magazine shows. Older chil-dren spent more time looking than youngerchildren, F\\, 64) = 9.86, p < .01. The meanproportions of time spent looking were 53%and 42%, respectively. There were no signifi-cant interactions of age with program char-acteristics.

Boys spent more time looking than girls,F{\, 64) = 5.43, p < .05. A borderline inter-action of Sex X Animation, F{1, 64) = 3.69,p < .06, signified that the sex difference oc-curred primarily on the animated programs.For the animated programs boys attended 56%of the time, whereas girls looked 43% of thetime. The corresponding figures for live pro-grams were 47% and 45%, respectively.

Duration per look. The two age groups dif-fered in their mean log duration per look dur-ing high- and low-paced programs, as indicatedby the significant interaction of Age X Pace,i=Xl, 64) = 4.42, p < .05. Older children hadlonger looks during low-paced programs (.081)than during high-paced programs (.066).Younger children showed the reverse pattern:Their looks were shorter during low-pacedprograms (.051) than during high-paced pro-grams (.068). Subgroup comparisons indicatedthat older children looked significantly longerper look at low-paced than at high-paced storyprograms (p < .05). There were no other sig-nificant main effects or interactions for du-ration per look. The means appear in Fig-ure 1.

Frequency of looking. Looking frequencywas submitted to an analysis of covariance(ANCOVA) in which the covariate was durationof looking. In previous studies, including thepilot studies for this one, frequency was neg-atively correlated with duration, perhaps be-cause children who spend most of their timelooking do not look toward and away fromthe screen as often as those who spend lesstotal time oriented to the screen. In any case,looking frequency is of interest only as an indexof the pattern of looking. It, of course, dupli-

3 !S | 350

PACE High Low Ugh Low

CONTINUITV Magazne I Story

AGE Younger Sublets

High Low High Low

Magazine | Story

Older Subteeu

Figure 2 Adjusted mean frequency of looking (per minute),with duration of looking covaried, as a function of age,pace, and continuity

cates in reverse some of the variance containedin the duration-per-look index.

With duration controlled, children shiftedgaze more often during high-continuity thanduring low-continuity programs, F(l, 63) =11.46, p < .01. Older children shifted gazemore often than younger children, F\\, 63) =8.69, p < .01. There was a significant inter-action of pace by age, F{\, 63) = 5.43, p <.05, which replicated in reverse the patternfound for duration per look. During low-pacedprograms younger and older children shiftedgaze about equally often. However, youngerchildren shifted less often during high-pacedthan during low-paced programs, whereasolder children shifted more often during high-paced than during low-paced programs. Theadjusted means are shown in Figure 2.

Similarity of children viewing together.Children had patterns of attention that weremore similar to those of their partner in thesame pair than to those of other children. Fortotal duration of looking and duration perlook, there was a significant effect of pairs inthe ANCOVA, i=\64, 80) = 5.23, p < .001, and

, 80) = 2.90, p < .001, respectively.

Comprehension of Temporal Sequences

Whole-program seriation. The total pro-gram-seriation scores were submitted to AN-COVAS in which age (2), sex (2), pace (2), con-

660 WRIGHT ET AL.

s

i

10

I Whole Program Sedation

1 Segment S«f lotion

sS

PACE

CONTINUITY

AGE

High Low High Low

Magazine I Story

Younger Subjects

High Low High Low

Magazine | Story

Older Subjects

Figure 3 Comprehension of programs indexed by senation scores as a function of age, pace, and continuity.

tinuity (2), pairs (5), and animation (2) wereindependent variables. The covariate for eachtelevision program was the mean "unseen"sedation score obtained by the control groupof children who had not seen that program.2

The seriation performance data are shown inFigure 3.

Children performed whole-program seria-tion better on the low-paced than on the high-paced programs, F{1, 62) = 27.70, p < .001,and they performed better on the high-con-tinuity than on the low-continuity programs,F\l, 62) = 36.18, p < .001. Older childrenperformed better than younger ones, F(l,62) = 44.20, p < .001. Of borderline signifi-cance was the interaction of pace by continuityby age, F(l, 62) = 3.76, p < .06.

Although both age groups performed betteron high- than low-continuity shows and onlow- than high-paced programs, the pattern ofintermediate scores was somewhat different.For older children, either of the "favorable"attributes—high continuity or low pace—ledto performance that was almost as good aswhen both favorable attributes were combined.For younger children, by contrast, there wasan additive effect: Low-paced stories were se-quenced best; high-paced stories or low-pacedmagazine programs were intermediate; andhigh-paced magazines were sequenced very

poorly. There were no other significant effectson whole-program seriation scores.

Segment seriation scores. Parallel analyseswere performed on the segment seriationscores shown in Figure 3. Again children per-formed better on low-paced than on high-paced programs, 7=1(1, 62) = 18.39, p < .001;and they performed better on high-continuitythan on low-continuity programs, F\ 1, 62) =30.08, p < .001. There was a significant in-teraction of pace by continuity, F(\, 62) =47.51, p < .001, and a borderline three-wayinteraction of pace by continuity by age F{ 1,62) = 3.85, p < .06. The pattern of findingsis similar to that for whole-program seriation.

Unlike the whole-program scores, however,there were differences in seriation of segmentsassociated with animation. The means areshown in Table 2. There was a significant maineffect of animation, F\l, 62) = 14.14, p <.001, and significant interactions of Anima-

2 It is obviously impossible to obtain such a controlscore from a child who has seen the program. Accordingly,the mean for other children who had not seen the programwas considered to be the most appropriate estimate of thescore that might be obtained by use of the cues in the stillpictures and general world and media knowledge. Therewere no effects of age, sex, pace, or continuity on the"unseen" scores.

CHILDREN'S TV: PACE AND CONTINUITY 661

Table 2Adjusted Segment Senation Scores for ProgramsVarying in Pace, Continuity, and Animation

Pace

LowHigh

Continuity

Animated

Low

8.958.18

High

8.058.89

Low

9.816 06

Live

High

117810 09

tion X Pace, F(l, 62) = 24.60, p < .001, andAnimation X Continuity, F(l, 62) = 26.41, p< .001. Children performed better on live thananimated programs for all combinations ofpace and continuity, except those with highpace and low continuity. For those rapidlypaced magazine shows, children were able torecall the serial order within a short segmentbetter for animated than for live shows. Be-cause there were only two television programsin each cell, this result may reflect idiosyncraticproperties of particular programs.

Order effects. An ANOVA of age (2) X order(2) X pace (2) X continuity (2) X pairs (5) Xanimation (2) was performed for each type ofseriation score. Order was coded 1 when theanimated program was shown first and 2 whenthe live program was shown first. In the anal-ysis, therefore, the interaction of Animation XOrder reflects the real effects of order of pre-sentation for particular programs. No covari-ates were entered because the comparison oftwo orders of administration always involvedcomparing scores for the same programs.

There were no effects of order for whole-program seriation, but there were some sig-nificant effects for segment seriation scores.The relevant means appear in Table 3. Therewas a significant interaction of Continuity XOrder X Animation, F\ 1, 64) = 5.63, p < .05,and a five-way interaction of Age X Pace XContinuity X Order X Animation, F( 1, 64) =4.33, p < .05. For magazine programs only,segment seriation scores were better for theprogram seen last and tested immediately afterviewing than for the program seen first andtested after a delay. This recency effect formagazine programs only was confined to cellswhere performance was intermediate. That is,it appeared for younger children in low-pacedprograms and for older children in high-paced

programs. It did not appear either for youngerchildren in high-paced programs, where scoreswere consistently low, nor for older childrenin low-paced programs, where scores wereconsistently high.

For high-continuity programs (stories), or-der of presentation and testing had no signif-icant effects on either whole-program or seg-ment seriation performance.

Similarity of children viewing together. Al-though comprehension was tested individuallyfor each child in a separate room from thepair-mate, pairs of children viewing togethercould not be assumed to have independentscores. If, for example, attention during view-ing influenced comprehension, then the mu-tual influence of two children on each other'sattention during viewing could lead to theirhaving similar patterns of comprehension. Thetwo children in a pair were also free to discussand rehearse program content during viewing.In the ANOVAS of whole-program and segmentseriation scores, the main effect of pairs wassignificant for whole-program seriation, F(64,80) = 1.84, p < .01, but not for segment se-riation scores. Even for whole programs, how-ever, pairs accounted for a relatively smallproportion of variance.

Relation Between Attentionand Comprehension

The relation between individual patterns ofattention and comprehension scores were ex-amined in a series of multiple regression anal-yses formatted in two different ways: predicting

Table 3Mean Segment Seriation Scores by Orderof Testing/Viewing

Pace

LowYoungerOlder

HighYoungerOlder

All subjects

Order of viewing

Low continuity

First

6.9510.55

6.557 35

7.85

Second

8 7510 65

6.909.25

8.89

High continuity

First

9.359.90

8.7510.70

9.68

Second

8.9010.35

8.4510.25

9.49

662 WRIGHT ET AL.

Table 4Multiple Regression of Whole-Program Senation Scores

Live programs Animated programs

Predictor

1. Duration2. Frequency3. Age group4. Continuity5. Pace

Overall FOverall multiple R2

Beta

.10

.21

.29

.28-.37

Increment in R2

.06

.08

.07

.08

.13(5, 74) 10.87**

.42

F to enter

4.60*7.75**6.42*8 36**

17.35**

Beta

.38

.26

.28

.22-.17

Increment in R2

.18

.14

.07

.04

.03(5, 74) 12.53**

.46

F to enter

17.49**16.16**7.83*5.55*3.88

* />< .05 * * / > < . 0 1 .

comprehension from attention and predictingattention from comprehension, in each caseusing age and treatment as additional predic-tors. Because there was some relationship be-tween the attention and whole-program com-prehension scores of pairs of children whoviewed together, pair means were used as en-tries in the multiple regressions. Analyses wereperformed separately for animated and liveprograms because each child saw one programof each type.

Prediction of comprehension from attention.In the prediction of comprehension scores, theattention measures, duration and frequencyof looking, were forced to enter first, followedby age, continuity, and pace. This procedurewas designed to determine the extent to whichthe effects of age and treatments might be me-diated by attention.

The regressions for predicting whole-pro-gram sedation in live and animated programsare shown in Table 4. Both attention durationand attention frequency predicted significantproportions of the variance in both types ofprograms. Even with attention statisticallycontrolled, older children performed betterthan younger ones; senation scores were higherfor high-continuity than for low-continuityprograms; and, for live programs, children re-called low-pace better than high-pace pro-grams. That is, the effects of age and treatmentson comprehension were not entirely accountedfor by their effects on attention.

The major difference between the regres-sions for animated and live programs was thatattention accounted for twice as much vari-ation in comprehension of animated, as com-pared to live programs (32% vs. 14%). For

animated programs, duration of attention wasa somewhat stronger predictor than frequency,whereas the reverse was true for live programs.In live programs, low pace was associated withmuch better seriation than high pace, evenwith all other variables controlled. In animatedprograms, pace did not account for compre-hension after the effects of other variables wereremoved.

Subanalyses were carried out for high- andlow-continuity programs separately. The re-sults appear in Table 5. The effects of attentionon comprehension were in general stronger forstory than for magazine formats. By contrast,the independent effects of pace on compre-hension occurred primarily in live magazineformats.

Subanalyses were also performed for olderand younger children separately. They appearin Table 5. Both attention and treatments ac-counted for more variance in older than inyounger children's seriation scores.

In summary, these analyses show that at-tention, age, continuity, and pace accountedfor just under half the variance in the whole-program seriation scores.

When segment seriation was the dependentvariable, fewer effects reached significance thanwhen whole-program seriation was analyzed.Attention was not a significant predictor ofcomprehension even when entered first, butthe patterns of relationship were similar. Ac-cordingly, they are not reported here.

Prediction of Attention From ComprehensionThe second set of multiple regressions were

designed to predict attention as a dependentvariable. On the grounds that actual compre-

CHILDREN'S TV: PACE AND CONTINUITY 663

Table 5Multiple Regression of Whole-Program Sedation Scores for High- and Low-Continuity Programsand for Older and Younger Children

Live programs Animated programs

High continuity Low continuity High continuity Low continuity

Predictor

Incre- Incre-ment F to ment Flo

Beta in R2 enter Beta in R2 enter

Incre- Incre-ment Flo ment Flo

Beta in R2 enter Beta in R2 enter

1. Duration2. Frequency3. Age group4. PaceOverall FOverall multiple R2

1. Duration2. Frequency3. Continuity4 PaceOverall FOverall multiple R2

.20

.24

.42- .12

(4,

.07

.13

.14

.0135).35

3.075.54»7.90"0.80

4.81*

Younger children

.23

.01

.22- .40

(4,

.03

.04

.04

.1335).24

1.251.641.645.69*

2.72*

.04

.26

.20-.63

(4,

.00

.07

.05

.3735)

.49

0.012.712 17

25.498.46**

Older children

.01

.40

.40-.56

(4,

.02

.08

.19

.2735)

.56

0.803.399 4 9 "

21.38"11.10"

.57

.39

.21-.04

(4,

.30

.19

.04

.0135)

.54

16.28"14.14"3.140.13

10 10"

Younger children

.26

.3026

-.12(4,

.11

.13

.06

.0135)

.31

461*6.42*3.030.68

3 9 7 "

.17

.08

.44-.26

(4,

.07

.02

.17

.0735)

.33

2.760.898.57"3.34

4.29"

Older children

.57

.34

.23-.26

(4,

.19

.14

.07

.0435)

.44

8.74"7.65"2.984.17*

6.97"

*P<-05. - p<m.

hension scores may serve as one index of com-prehensibility, and that such differential com-prehensibility might be perceived early in aprogram and thus have an opportunity to in-fluence attention to the remainder of the pro-gram, the comprehension scores were enteredas predictors of attention, along with age, con-tinuity, and pace, in two different forced orders:comprehension first and comprehension last.None of the analyses significantly predictedfrequency of looking. Accordingly, only the

analyses of duration of attention are reported.The results appear in Table 6.

For live programs, even when comprehen-sion was forced to enter first, only whole-pro-gram seriation entered, and it accounted foronly a 6% increment in R2. Continuity wasthe only other significant predictor. For ani-mated programs, comprehension (whole-pro-gram seriation) was the only significant pre-dictor. The total amount of variance predictedwas considerably smaller when attention was

Table 6Multiple Regression of Attention Duration Scores

Predictor Beta

.05

.20

.17

.24

.23

Live programs

Incrementi n * 2

.06

.04

.02

.07

.04(5,74) 4.34"

.23

Ftoenter

4.60*3 471.886.56*3.60

Beta

.42

.06

.04-.04

.07

Animated programs

IncrementinR2

.18

.01

.00

.00

.01(5, 74) 3.64"

.20

Floenter

17.49"0.670.140 070.47

1. Whole-program seriation2 Segment seriation3 Age group4. Continuity5. Pace

Overall FOverall multiple R2

* P < .05. " p < .01.

664 WRIGHT ET AL.

the dependent variable than when compre-hension was the dependent variable.

DiscussionWe began with the general notion that at-

tention to, and comprehension of, televisionprograms was a complex resultant of programfeatures such as pace and continuity in inter-action with the active processing skills andmotivation to employ them brought to thesituation by the child. Some features, like pace,create both perceptual salience and processingdifficulty. Program formats, like magazine vs.story and animated vs. live production, candemand and support different kinds anddepths of processing of content. That antici-pated complexity is supported by the results,for few effects took the same form across alltypes of programs and both ages of viewers.

Regardless of animation, pace, and conti-nuity, older children visually fixated the screenduring greater portions of the programs andachieved higher seriation scores than youngerchildren. The differences in comprehensionwould have been remarkable only if they hadfailed to occur.

Contrary to assumptions often made byboth producers and their critics, high-pacedprograms did not receive consistently greaterattention than low-paced programs. A nearlysignificant effect in that direction occurredonly once: for younger children when they werewatching magazine format shows. By contrast,the effects of pace on comprehension wereclear: High-paced shows were more difficultto integrate and sequence over time than werelow-paced shows.

Pace did significantly affect the patterningof attention for older children, primarily whenthey viewed high-continuity programs. In thatone case, where processing may have beenmost intensive and integrative, length of looksappeared to depend on length of scenes or bitsas indexed (negatively) by pace. This findingalso supports the hypothesis that older childrenare more sensitive to the formal feature cuesfor transitions in content than are youngerchildren.

Stories received more overall attention thandid magazine programs. Both seriation mea-sures of comprehension also favored storyprograms over magazine shows for both age

groups, suggesting that younger as well as olderchildren benefited from the schematic orga-nization of the stories. The view that the storieswere processed schematically is further sup-ported by the finding that there were no re-cency effects for story recall. For magazineshows, there were recency effects on segmentseriation. Close attention to stories was morelikely to improve serial recall; presumably theschematic organization provided by the storycould be used by an attentive child. In mag-azine shows, attention did not produce largeimprovements in seriation; instead, the overallnumber of segments (per unit time) indexedby pace seemed to govern the ease of recall.

These findings suggest that both age groupswere able to use schematic processing and toachieve temporal integration of the content ofstones. The major difference between agegroups was the extent to which "supportiveattributes" (i.e. high continuity and low pace)were important. For the older children, eitherattribute was sufficient to enable them toachieve high levels of recall. For the youngerchildren, the contributions of story format andlow pace to comprehension were additive: Ei-ther one was better than neither, and togetherthey were even more helpful.

The findings provide some support for thehypothesis that ongoing comprehension ofcontent increases attention (Anderson, Lorch,Field, & Sanders, 1981). Comprehensionscores predicted attention. They accounted forthe effects of age on attention, and, for ani-mated programs, they accounted for the maineffects of continuity. As there were no maineffects of pace on attention, there was littlevariance associated with pace to account for.In live programs, however, the effects of con-tinuity on attention remained after compre-hension was removed. Children's attraction tolive stories was apparently based on more thanthe comprehensibility of the content.

Animated programs in America are a dif-ferent genre than live programs. A much nar-rower and more stereotyped set of expectationsis aroused by knowing that a program is an-imated than by knowing it is live. The rangeof pace variation we were able to find in an-imated programs was smaller, and the overallpace was higher than in live programs. Thesedifferences are part of the reality of broadcast

CHILDREN'S TV- PACE AND CONTINUITY 665

television. Although animation had only oneweak effect on attention (boys preferred ani-mated to live programs in overall attention;girls did not) and had no effects on compre-hension, it nevertheless seemed to amplify therelationship between attention and compre-hension. It is as if animated stories are morepredictable and familiar than either animatedmagazine bits or live programs, especially tothe older children, who presumably have moreexperience with the medium and its conven-tions. They are thus able to make more nearlydichotomous decisions about the appeal of aprogram early in the show. If they like thegenre, they attend more closely and recall cor-rectly; if not they do neither. Hence, in thosecells and for those children, the correlationsbetween attention and comprehension are ex-ceptionally high.

In summary, with increasing age, youngtelevision viewers came to attend more con-sistently and strategically, and they compre-hended more effectively what they viewed, atleast as measured by recall of temporal se-quences. Pace of events in a program, the in-verse of average scene or bit length, determineddifficulty for comprehension and controlledthe tempo of looking by older children forstory programs but tended to enhance atten-tion only for younger children—and then onlyto the less schematically organized magazineprograms. Continuity of format played a con-sistent role in both attention and comprehen-sion. Stories elicited more attention and bettercomprehension than magazine shows. Finally,animation was a cue to program stereotypy:Attention and comprehension covaried withone another more in animated than in liveshows. This effect was especially pronouncedwhen animated stories were viewed by olderchildren, the case where the most stereotypedprogramming was viewed by the children withthe best knowledge of those stereotypes.

The implications of this study for improvedprogramming of television for children areclear. High pace does not often enhance at-tention and may interfere with comprehension.Content presented in story format is in generalmore effectively communicated than contentpresented in a magazine format. Live formatselicit less all-or-none processing than animatedformats.

The results contain significant implicationsfor models of basic cognitive development aswell. Not only is television viewing a contextin which active cognitive processing takes placeand can be profitably studied, but it is a contextin which children use a variety of subtle cuesto determine the level of processing that isnecessary and sufficient for comprehension.Neither comprehensibility nor salience alonedetermine the nature and extent of processing.Rather, it appears that they are determinedinstead by a combination of schematic knowl-edge (e.g., animated story scripts) and strategicdecision making at key points in response tothe level of processing demanded or supportedby the program's format and structure. Formand format of a familiar medium play rolesin cognitive processing that are in part inde-pendent of content. A more precise delineationof what children know who know how to pro-cess their favorite medium awaits further in-vestigation.

References

Alwitt, L F , Anderson. D R., Lorch, E. P., & Levin,S R (1980) Preschool children's visual attention toattributes of television Human Communication Re-search. 7. 52-67.

Anderson. D R ,& Levin. S R (1976) Young children'sattention to "Sesame Street" Child Development. 47,806-811

Anderson, D R., Lorch E P, Field, D E., & Sanders, J.(1981) The effects of TV program comprehensibilityon preschool children's visual attention to television.Child Development. 52. 151-157

Berlyne, D E. (1960). Conflict, arousal, and curiosityNew York McGraw-Hill

Collins, W A (1982) Cognitive processing in televisionviewing. In D Pearl, E Bouthilet, & J. Lazar (Eds),Television and behavior Ten years of scientific progressand implications for the 80s (Vol 2). (National Instituteof Mental Health, Vol 2; pp. 9-23) Washington, DC:U S. Government Printing Office

Collins. W A , Sobol. B L., & Westby, S. (1981). Effectsof adult commentary on children's comprehension andinferences about a televised asggressive portrayal. ChildDevelopment. 52. 158-163

Huston, A. C, & Wnght, J C (1983) Children's processingof television The informative functions of formal fea-tures. In J Bryant & D. R Anderson (Eds.), Children'sunderstanding of television Research on attention andcomprehension (pp. 35-68). New York- Academic Press

Huston, A C , Wright, J C , Wartella, E., Rice, M. L.,Watkins, B A., Campbell, T, & Potts, R (1981). Com-municating more than content Formal features of chil-dren's television programs. Journal of Communication,J/(3), 32-48.

666 WRIGHT ET AL.

Huston-Stein, A. C , & Wright, J C. (1979). Children andtelevision: Effects of the medium, its content, and itsform Journal of Research and Development in Edu-cation. 13, 20-31.

Krull, R., & Husson, W (1979) Children's attention: Thecase of TV viewing. In E Wartella (Ed ), Children com-municating Media and development of thought, speech,understanding (Sage Annual Reviews of CommunicationResearch, Vol. 7, pp 83-114). Beverly Hills Sage

Lorch, E P., Anderson, D R., & Levin, S R. (1979). Therelationship of visual attention to children's compre-hension of television. Child Development, 50, 722-727.

Mandler, J M (1979). Categorical and schematic orga-nization in memory. In C. R Puff (Ed ), Memory or-ganization and structure (pp. 259-299) New York. Ac-ademic Press

Salomon, G. (1981). Introducing AIME: The assessmentof children's mental involvement with television. In HGardner & H. Kelly (Eds ), Viewing children throughtelevision (pp 89-102; A quarterly sourcebook in theseries. New Directions in Child Development) San Fran-cisco: Jossey-Bass

Wright, J. C , Calvert, S. L., Huston-Stein, A., & Watkins,B. A (1980, May). Children's selective attention to tele-vision forms Effects of salient and informative productionfeatures as functions of age and viewing experience Paperpresented at the annual meeting of the InternationalCommunication Association, Acapulco, Mexico.

Wright, J. C , & Huston, A. C. (1981). The forms of tele-vision. Nature and development of television literacy inchildren In H. Gardner & H. Kelly (Eds), Viewingchildren through television (pp. 73-88; A quarterlysourcebook in the series, New Directions in Child De-velopment) San Francisco' Jossey-Bass.

Wright, J C , & Huston, A. C. (1983) A matter of form:Potentials of television for young viewers AmericanPsychologist. 38. 835-843.

Wright, J C , Watkins, B A., & Huston, A. C. (1978,August) Active vs passive television viewing- A modelof the development of TV information processing in chil-dren Paper presented at the annual meeting of theAmerican Psychological Association, Toronto

Received January 5, 1983 •