CATHERINE A. COLE and MICHAEL J. HOUSTON*

Alternative cognitive explanations of memory deficiencies in the elderly suggest

certain conditions in which elderly might be expected to learn less than younger

adults in a consumer information processing context. The authors investigated these

conditions using the levels-of-processing framework for the study of memory. The

results indicate greater difficulties on the part of the elderly in deeply processing

information from print media and in realizing the learning benefits of television.

These findings suggest that marketers face unique problems in reaching this increas-

ingly attractive market segment.

Encoding and Media Effects on ConsumerLearning Deficiencies in the Elderly

Despite the substantial amount of research on con-sumer information processing and the growth of the el-derly segment of the consumer market, knowledge ofelderly consumers' information processing tendencies islimited. In their useful review of information processingby the elderly, Phillips and Stemthal (1977) note the lackof focused research on the elderly in a consumer infor-mation context. They further point out that of the studiespertaining to the elderly very few include other age groupsfor comparison. The situation has changed little in thetime since their review. Additional efforts to understandthe elderly consumer have been primarily in the form ofmore literature reviews (e.g., Meadow, Cosmas, andPlotkin 1981; Ross 1982) or position statements speci-fying areas in which applied research on the elderly wouldbe particularly relevant (e.g., Visvabharathy 1982).

Research in the general area of cognitive psychologyand the specific areas of gerontology and developmentalpsychology shows important differences between agedand younger individuals in their information processingtendencies. Much of this research considers the relativelearning capabilities of the elderly by examining differ-ences between younger and older individuals in theirmemory for previously seen information. The generalfinding suggests memory deficiencies in the elderly incomparison with younger adults. However, the magni-

*Catherine A. Cole is Assistant Professor of Marketing, Universityof Iowa. MichaelJ. Houston isKesearch Professor of Marketing, Uni-versity of Minnesota.

tude of the memory deficiency appears to vary acrosscertain conditions related to the pace at which infor-mation is presented, how it is placed in memory (en-coding), and how it is recovered from memory (re-trieval). Burke and Light (1981) provide an effectivereview of this body of research.

The body of research on memory deficiencies in theelderly provides a useful structure for studying consumerinformation processing in the elderly. It suggests pos-sible conditions leading to changes in the amount of thelearning decrement experienced by elderly consumers.These conditions translate readily into specific condi-tions found in a consumer information processing con-text. The general purpose of our article is to report re-search that identifies conditions leading to larger or smallerdifferences between the elderly and younger consumersin the learning of consumer information. First, we fur-ther develop these conditions by summarizing altemativeexplanations of why elderly individuals have a memorydeficit in comparison with younger individuals. We thendescribe the design and procedures of a laboratory ex-periment to determine certain information processingconditions leading to larger or smaller leaming deficien-cies in elderly consumers. Finally, we discuss the find-ings and their implications

EXPLANATIONS OF MEMORY DEFICIENCIESIN THE ELDERLY

Of the major explanations for memory deficiencies inthe elderly, two center on encoding, the process by whichinformation is placed in memory, and two center on re-

55

Journal of Marketing ResearchVol. XXIV (February 1987), 55-63

56 JOURNAL OF AAARKETING RESEARCH, FEBRUARY 1987

trieval, how previously seen information is recovered fk)mmemory. Collectively, these explanations and the re-search based on them suggest several relevant consumerinformation processing conditions under which to studythe elderly.

Encoding Deficiencies

The explanations with a common emphasis on the en-coding stage are based on the levels-of-processing (LOP)framework for the study of memory (Craik and Tulving1975). The essence of LOP is that information may beencoded at varying levels ranging from shallow to deep.Shallow processing occurs when the structural or sen-sory features of a stimulus are encoded, whereas deepprocessing involves a more elaborative, semantic-ori-ented form of encoding. According to LOP, a deeperlevel of processing at the encoding stage leads to im-proved memory performance in recalling the encoded in-formation.

LOP has had a prominent role in memory research,but is not without criticism The major criticism is thecircularity of LOP: memory improves because of a deeperlevel of processing, but a deeper level of processing isdefined by improved memory performance There is noindependent measure of processing level other than thedependent variable that level is posited to influence. Still,since its formulation, LOP has been the dominant par-adigm used to study encoding explanations of reducedmemory performance in the elderly. Furthermore, be-cause of the variations in sensory-oriented versus se-mantic-oriented content in consumer information stim-uli, LOP has strong intuitive appeal and has been usedin a consumer research context (e.g, Childers and Hous-ton 1984; ZeithamI 1982). Consequently, we used it inour research with an attempt to overcome its major crit-icism.

Production deficiency. This view of poorer memoryperformance by the elderly suggests that they do notspontaneously employ deep processes when encoding in-formation to the same extent as younger adults. The po-sition here is not that the elderly are unable to processdeeply but that, for whatever reason, they simply do not.When they are induced to process at a semantic levelthrough orienting tasks, their memory equates to that ofyounger adults (Craik and Simon 1980). Often impli-cated as a reason for the failure to encode spontaneouslyat a semantic level are the school-like tasks used in mostpsychological experiments on memory. Younger sub-jects are less removed from school settings and more in-clined to adapt to the processing requirements of the sit-uation (Baltes, Reese, and Lipsitt 1980).

Processing deficiency. Eysenck (1974) suggests thatmemory problems in the aged stem from more than justa failure to process at deeper levels. Rather, the elderlysuffer from an inability to process deeply. He bases thisconclusion on a study in which he found that older adultsperformed as well as younger adults in free recall whengiven nonsemantic encoding instructions, but performed

less well under semantic encoding instructions. Thoughthe actual cause of processing deficiencies in the elderlyis beyond the scope of our research, one explanation ofEysenck's findings may be that a slowdown of centralnervous activity contributes to a decline in the rate ofcognitive processing.

Memory Search Deficit

Research has been reported that points to retrieval def-icits rather than encoding deficits as the source of poormemory performance by the aged. Evidence for such aview is provided by White (see Craik 1977). She con-ducted a study similar to that of Eysenck (1974), butwith one important extension. The memory of elderlyand younger adults was compared across semantic andnonsemantic instructions by using both a recall and arecognition test. The recall measure yielded the same re-sults that Eysenck found: under semantic processing el-derly subjects recalled less, but under nonsemantic pro-cessing there were no age differences. However, therecognition measure revealed no age differences underany processing condition, though recognition scores im-proved for both groups under semantic versus nonse-mantic conditions. Craik (1977) interpreted these find-ings as supportive of a retrieval deficit rather than aprocessing deficit. He reasoned that encoding deficien-cies should show up in both recall and recognition mea-sures. This was not the case; only recall indicated poorermemory for the aged. Because recall requires retrievalprocesses but recognition bypasses the retrieval process,the explanation must be that the elderly encode infor-mation equally well but cannot retrieve it as well asyounger adults. One or both of two specific aspects ofretrieval, neither of which would show up in a recog-nition measure, may be the source of retrieval deficitsin the elderly: organization and self-generated retrievalcues.

Organization. Organization of encoded informationoccurs when order and pattem, often through the use ofcategorization, characterize its storage in memory. Or-ganization allows easier recall of the information frommemory, presumably because an organizational schemeprovides easier access to stored information (Lynch andSrull 1982), but it is not important for recognition. Itmay be that the elderly find it more difficult to utilizeorganization strategies spontaneously when processingnew information.

Organization may be implicated as a source of re-trieval deficits in the elderly by the frequent finding thatthe rate of presentation of a stimulus influences age dif-ferences in recall—more rapid pacing of informationpresentation leads to greater recall superiority of the youngover the old. When self-pacing characterizes the rate ofpresentation, age differences in leaming diminish or vanish(Phillips and Stemthal 1977). Self-pacing allows greateruse of organization strategies, whereas increased rates ofextemally paced information presentation limit their use(Burke and Light 1981). If the elderly have greater dif-

CONSUMER LEARNING DEFICIENCIES IN THE ELDERLY 57

ficulties using organization strategies, these effects wouldbe magnified for them.

Self-generated retrieval cues. A second potential soureeof retrieval deficits in the elderly may be in their abilityto self-generate associative cues to aid retrieval. Suchcues enhance recall but do not have a role in recognition(Lynch and Srull 1982). If the elderly are less able thanthe young to self-generate retrieval cues, their perfor-mance on free recall tests will be worse. In aided recalltests with extemally provided retrieval cues (as in paired-associate leaming tasks), the leaming decrement of theelderly should be reduced but not as much as in a rec-ognition test.

RESEARCH PROBLEM

Collectively, the preceding explanations indicate sev-eral conditions under which to examine consumer infor-mation processing differences between elderly and youngeradults. The encoding-based explanations suggest the is-sue of whether aged consumers do or do not have thecapability of extracting semantic content from informa-tion stimuli when encouraged to do so. Investigating thisissue in the common, "everyday" context of consumerinformation helps to avoid the problem of whether theschool-like tasks used in previous research account fora production deficiency

The retrieval-based explanations point to other issuesand conditions relevant to consumer information pro-cessing by the aged. Both retrieval-based explanationssuggest that recognition reduces, if not eliminates, theleaming decrement in the elderly. Because recognitionis most useful for in-store decision making (Bettman1979), consumer behavior by the elderly and marketingefforts directed to them would be aided by increasedmarketing attention to packaging and other in-store dis-plays that are consistent with advertising content. If the

Table 1EMPIRICAL RESULTS PREDICTED BY EACH EXPLANATION

Explanation Major prediction

Encoding deficienciesProduction deficiency

Processing deficiency

Memory search deficitOrganization

Self-generatedretrieval cues

Semantic encoding instructions willremove age differences in learningfor both self- and externally pacedinformation

Age differences in leaming will beobserved in all conditions

There will be no age differences inlearning when recognition is thedependent variable

Age differences in recall will belargest when information isextemally paced

Age differences in leaming will belargest for recall tests, smaller foraided recall tests, absent forrecognition tests.

specific source of the retrieval deficit is a greater diffi-culty on the part of the elderly in organizing information,further implications for advertising media are suggested.Because they allow self-pacing, print media would bemore appropriate for the placement of advertising mes-sages aimed at the elderly. These media would give theelderly a full opportunity to organize the infomiation andenable them to recall it better. Less emphasis on in-storestimuli would be necessary.

We undertook an experiment in which the independentvariables of age, level of processing, and media weremanipulated to correspond to the various conditions.Memory effects were assessed by using free recall, aidedrecall in which an extemal retrieval cue is provided, andrecognition. Table 1 summarizes the predicted results as-sociated with each of the four explanations of age dif-ferences in memory. Such an experiment offers insightsinto the approaches by which marketing managers canreach the elderly consumer and contributes to the generalliterature on leaming by the elderly.

DESIGN AND PROCEDURES

Overview

The effects of level of encoding, pacing, and age onmemory were studied in a full 3 x 2 x 2 factorial de-sign. Level of encoding was manipulated at three levelsby instructing subjects to process information at a sen-sory level, semantic level, or as they normally would.Pacing was manipulated at two levels by exposing sub-jects to material presented either through an extemallypaced medium (television) or self-paced medium (news-paper). Two levels of age were represented by includingseparate groups of young and old subjects. The younggroup included 131 subjects between 18 and 45 years(median = 32). The elderly group included 136 subjectsbetween 60 and 91 years (median = 72). Members ofeach age group were assigned randomly to one of thesix combinations of encoding level and medium and wereexposed to the information contained in the medium.Memory for the information was tested after exposurethrough a series of free recall, aided recall, and recog-nition measures.

Subjects

The 267 subjects who participated in the study wererecruited from various organizations, including churchgroups, drop-in senior centers, hot lunch centers, parentgroups, and housing cooperatives. Subject payment con-sisted of a monetary donation to each organization basedon the number of subjects recruited from it. Data werecollected over a two-month period at meeting sites of theorganizations.

Procedures

The procedures used in the experiment were pretestedto ensure their smooth conduct and to establish the amountof instruction necessary for each age group to understand

58 JOURNAL OF AMRKETING RESEARCH, FEBRUARY 1987

and complete the measuring instruments. Procedures wereadministered to subjects in small groups, typically 7 to12 people. All instructions were given to subjects bothorally on an audiotape player and in written form. De-pending on their media assignment, subjects were toldthat the study was concemed with how people formedimpressions of different TV news shows or newspapersand that they would be evaluating three different shows/papers. Subjects were given a practice series of Likert-type items to familiarize them with the major type ofmeasuring procedure used in the study. Subjects then weregiven instructions on how they should evaluate each show/paper and were exposed to the specific Likert-type itemsthat they would use to provide their evaluations.

Exposure to the respective media followed. Each me-dium was divided into three parts, each part representingone of the three versions to be evaluated. Each part ofthe TV medium was 10 minutes long. Subjects exposedto newspapers spent as much time as they wanted witheach part. After exposure to each part, subjects com-pleted at their own pace a common set of evaluation items.After exposure to the entire media presentation, subjectscompleted a measure of which of the three shows/papersthey preferred most and a thought-listing measure for thetest message. A series of additional measures includingunexpected tests of memory and a self-report measure oflevel of encoding followed. The memory tests included,in order, unaided recall, aided recall, and recognition.Subjects were allowed to complete these measures at theirown pace. Each session ended with a debriefing andquestion-answer period. Most sessions were completedwithin 75 minutes.

Manipulation of Independent Variables

Level of encoding. With the LOP framework as a guide,level of encoding was induced by using two orientingtasks during the procedures. First, the sensory-orientedinstructions given before exposure to the medium em-phasized that subjects should focus on the appearanceof each version of the medium when evaluating it. Se-mantic-oriented instructions emphasized that subjectsshould focus on the content of the material contained inthe medium. The "normal" processing condition em-phasized that subjects should examine the medium as theynormally would when watching/reading it.

Second, level of encoding was reinforced through thespecific items used to evaluate each part of the medium.Subjects in the sensory condition completed items onmedia attributes such as clutter, liveliness, professionalappearance, drabness, and attractiveness. In the seman-tic condition the items provided for evaluations of attri-butes such as the blend of local and national news, ob-jectivity, confusion in the stories, and how boring thecontent was. Subjects saw the evaluation items beforeexposure to the material to be evaluated and completedthem for each part of the medium immediately after thatpart was presented. Thus, by completing the items im-mediately after each part, subjects were being reinforced

in the level of encoding to use when examining the nextpart.

In the normal processing conditions subjects respond-ed to different lifestyle items after each part. These itemswere not linked to the media and pertained to such char-acteristics as attitudes toward risk-taking, self-confi-dence, and the desire for money.

Media conditions. The two media representing exter-nal pacing and self-pacing were developed to be as se-mantically equivalent as possible. The television mate-rial was developed first. Local news programs from threecities outside the test area were taped "off the air." Tapesof eight TV commercials for several brands not availablein the test area were obtained. Each news show, editedto 72-minute length, contained samples of traditional news,weather, and sports segments. Five 30-second commer-cials then were edited into each segment at various pointsto arrive at three distinct 10-minute segments of pro-gramming and commercials. Thus, a total of 15 com-mereials appeared during the entire 30-minute tape. Someof the individual commercials were repeated during theentire 30-minute tape.

Included within the repeated commercials was a testmessage used for one of the recognition measures. Thismessage, created specifically for the study, promoted adevice available from "your gas and electric company"that shuts off a water heater daily during peak demandperiods. The content of the test message was entirelyverbal with the words rolling by on the screen while beingread by a male announcer. This message was repeatedfour times during the 30-minute tape. Product classesrepresented in the other commercials included gasoholand several food products.

The newspaper version of the media variable wasadapted directly from the TV version. The news portionsof each show were transcribed and typeset (with picturesand headlines added) into three distinct newspapers, eachcontaining five to six 82 x H" pages of news, weather,and sports. The verbal content of each newspaper storywas identical to that in the TV version.

Fifteen full-page print ads for the same brands usedin the TV condition were incorporated into the news-paper version. Each ad was repeated the same numberof times and appeared after the same news story or otherad as in the TV version. Though the content of many ofthe ads differed somewhat between TV and print ver-sions, the content of the test message was identical inthe two media conditions. The three newspapers werebound together in a single booklet but retained separateidentities within the booklet.

Age. Different levels of age were achieved throughthe inclusion of the two age groups. The comparison ofage-cohort groups is by far the most common approachused in research on the aged. Obviously, it does not al-low random assignment on the age variable and thereforedoes not control for other differences between age groups(e.g., education, occupation). Unfortunately, analysis ofcovariance and matched samples do not overcome these

CONSUMER LEARNING DEFICIENCIES IN THE ELDERLY 59

control problems (see Kausler 1982; Salthouse 1982).For example, individuals in different age groups with,say, 12th-grade education still cannot be equated on thisvariable. They remain different because of changes inthe content of formal education and the recency of itscompletion. Our approach was to obtain age-cohort groupscontaining noninstitutionalized adults responsible formaking purchases on their own.

Dependent Variables

Memory for various aspects of the presented materialwas tested through a series of unexpected recall and rec-ognition measures. Separate measures were taken for thenews and advertising contents of the material. In thisway evidence about the potential uniqueness of the ef-fects to advertising could be obtained.

Unaided recall. Free recall was tested by asking sub-jects to indicate the main ideas for as many news storiesand advertisements as they could remember. Correct re-call for a news story was determined by whether a spe-cific story was identified in sufficient detail to indicatememory. For example, reference to "sports" was notdeemed correct recall; however, reference to a "hockeygame" (a specific story on one of the stations) was con-sidered correct. Recall for a specific ad was consideredcorrect if at least the product class was mentioned. Twoseparate measures of unaided recall were calculated.

1. News story recall—the number of specific news storiescorrectly recalled.

2. Advertising recall—the number of specific ads correctlyrecalled.

Aided recall. To determine whether the provision ofan extemal cue reduces the memory decrement of theelderly, a measure of aided brand name recall was taken.The advertised product classes were given to each sub-ject, who was asked to fill in the brand name. Minorspelling errors were ignored. Aided recall was repre-sented by the number of correct brand names provided.

Recognition. Tests for recognition memory were takenfor both news stories and advertisements to determinewhether a person could remember exactly what had beensaid or printed. News story recognition was assessed bygiving subjects a series of statements making claims aboutwhat was said in six of the news stories. Half of theclaims were true. Subjects responded to each claim byindicating whether it was true or false. The number ofcorrect responses (out of six) served as the measure ofnews claim recognition accuracy. Similarly, a series ofeight statements (half true) was used to assess claim rec-ognition for the test advertisement (water heater device).As an example of the claim recognition measures, oneitem for test-message claim recognition was: "The res-idential load management advertisement says 'Call youroil company today for more information.'" A correct re-sponse would be "false" because the ad stated: 'Call yourGas and Electric Company today for more information."

Finally, brand name recognition for the seven other

advertised products was assessed. For each product classsubjects were asked to indicate which of four brand namesrepresented the brand advertised.

Analysis

Analysis of variance (ANOVA) was used to assess mainand interaction effects on each dependent variable. Bothclaim recognition measures were analyzed by using d'from signal detection theory (Green and Swets 1966).The brand recognition measure was corrected for guess-ing effects by a correction factor proposed by Link (1982).

RESULTS

Manipulation Checks

The LOP framework for studying memory has beencriticized because of the lack of a separate measure ofencoding level. This situation presents difficulties when,as in our research, LOP is the source of a key experi-mental manipulation. Specifically, the absence of a sep-arate, well-accepted measure of encoding level pre-cludes the use of a standard procedure for establishingthe success of a manipulation of encoding level. Prob-lems with manipulation checks of encoding level arecompounded in studies of the elderly. The failure of amanipulation check to confirm a semantic level of en-coding in the elderly may not indicate the failure of themanipulation itself. If, as suggested by the processingdeficit explanation, the elderly are unable to encode atdeeper levels, it would be theoretically impossible for aseparate, direct measure of encoding level to .confirm theeffectiveness of a semantic orienting task. As a result,many of the studies using the orienting-task paradigmhave not included manipulation checks. Instead, the re-searchers have depended on encoding inductions that arestrong in a face-valid sense. We attempted to use a com-bination of strong inductions and validating evidence.

Self-report measures of encoding level were incor-porated into the procedures to serve as a partial manip-ulation check. Separate measures were developed forsensory and semantic levels and for each media condi-tion. After purification, the sensory and semantic scalesfor the print condition consisted of five items and fouritems with internal consistency coefficients of .90 and.74, respectively. Sensory and semantic scales for theTV condition consisted of seven items and six items withinternal consistency coefficients of .82 and .62, respec-tively. Each subject completed both sensory and seman-tic scales. The items in each scale asked subjects to in-dicate in agree-disagree terms the extent to which theythought about specific appearance (sensory) or content(semantic) features when viewing or reading the infor-mation.

Even if the elderly have a processing deficiency, thesescales should confirm successful sensory-semantic in-ductions in the younger subjects and at least a sensoryinduction in the elderly. Table 2, the cell means for eachscale in each experimental condition, approximates such

60 JOURNAL OF AAARKETING RESEARCH, FEBRUARY 1987

Table 2CELL MEANS FOR SEAAANTIC AND SENSORY AAANIPULATION CHECK SCALES

Semantic scale vatuesPrint

medium (4 items)

TV medium (6 items)

Sensory scale valuesPrint

medium (5 items)TV medium (7 items)

N

14.96(n = 23)

22.36(n = 25)

17.6125.88

Elderly

SM

14.27(n = 22)

22.24(n = 21)

17.1824.38

SN

14.24(n = 21)

25.46(n = 24)

21.0029.92

N

16.74(n = 23)

20.95(1 = 21)

18.4824.67

Younger adults

SM

15.91(n = 22)

23.25(n = 20)

18.0026.10

SN

10.54(n = 24)

20.43(n = 21)

22.5427.90

N, no encoding instructions; SM, semantic instructions; SN, sensory instructions.

a pattem. In the print condition the mean score for youngeradults on the semantic scale is significantly higher undersemantic instructions than under sensory instructions (t= 4.08, p < .01). Similarly, the mean score on the sen-sory scale is significantly higher under sensory instruc-tions than under semantic instructions (t = 2.78, p <.01). A reasonably similar pattem of results is seen foryounger subjects in the TV condition. The semantic scoreis higher under semantic than under sensory conditions(t = 2.16, p < .04). The sensory score is higher undersensory than under semantic instructions, though not sig-nificantly so.

For elderly subjects the sensory scores are signifi-cantly higher under sensory than under semantic instruc-tions in both print and TV conditions {t = 2.42, p <.02, and t = 2.07, p < .05, respectively). However,semantic scores do not have the appropriate pattem ineither media condition.

The overall pattem of scores on the self-report mea-sures of encoding could be interpreted as initial evidenceof a processing deficiency in the elderly. However, itdoes not eliminate the possibility of a manipulation fail-ure of the semantic instructions when given to the el-derly. Younger subjects may have understood the en-coding instructions given to them under both levels,whereas the elderly may have understood them only un-der sensory conditions and failed to process semanticallynot because of an inability to do so but because they didnot realize they should. We tum to indirect evidence toassess this possibility.

The processing deficiency hypothesis does not suggestthat the elderly cannot think at a semantic level, onlythat they are less able than the young to encode newinformation at a semantic level. Self-generated thoughtsshould be more semantic while one is attempting to en-code at a semantic level than while one is attempting toencode at a sensory level. Thus, if the elderly understoodthe semantic instructions, self-generated thoughts shouldbe more semantic than they were under sensory instruc-tions.

Immediately after exposure to and evaluation of thethree media parts, subjects completed a thought-listing

Table 3PERCENTAGE OF THOUGHTS CODED AS SEMANTIC

PrintTelevision

N

6722

Elderly

SM

7765

SA'

5242

Younger

N SM

76 8070 79

adults

SN

3472

N, no encoding instructions; SM, semantic instructions; SN, sen-sory instructions.

procedure for the test message that was the last elementof the information presented. They were given the head-line of the message ("Are you spending too much forhot water?") and asked to list the thoughts they had aboutthis message at the time of exposure. This thought-list-ing procedure has been used successfully by Lutz andMacKenzie (1981). If these self-generated thoughts aremore semantic under semantic than sensory instructionsfor both elderly and young subjects, we have indirectevidence that the elderly understood the semantic in-structions.

Three independent judges blind to the treatment con-dition of each subject coded each thought as semantic ornonsemantic. Interjudge reliability was .90. Table 3 showsthe percentage of total thoughts coded as semantic foreach of the 12 cells. In all four comparisons betweensemantic and sensory manipulations, the percentage ofsemantic thoughts is highest in the semantic condition.Thus, we have indirect evidence that the semantic in-structions resulted in at least a semantic orientation inboth elderly and young subjects.'

'A subject's statement was classified as semantic if it focused onthe information contained in the advertisement (e.g., "this service seemslike a good idea"). A limitation of the cognitive response data is thatthey are only a reflection of thinking about the test message. How-ever, subjects in the encoding conditions were instructed to think aboutall the material in the same way, so it seems reasonable to infer thatthe way they thought about the test message is the way they thoughtabout all the material. In the no encoding instruction condition, thisinference cannot be made because subjects received no special in-structions.

CONSUMER LEARNING DEFICIENCIES IN THE ELDERLY 61

Background Variables

Subjects were measured on several background vari-ables to identify potential mediating variables not oth-erwise accounted for. Major differences between elderlyand young subjects were found on two variables. First,not surprisingly, young subjects had completed more yearsof formal education than elderly subjects. However, anymediating effects of this variable may have been sub-stantially offset by the fact that elderly subjects reportedspending almost twice as many hours watching televi-sion and reading newspapers as did young subjects. Fi-nally, the age groups did not differ significantly on ahealth index (a = .92) based on eyesight, hearing, andmobility—important physical factors relating to the abil-ity to see and hear information and engage in consumerbehavior.

Memory Effects

In this section we report the key substantive findingsabout memory effects of the independent variables. Theorder of presentation of the dependent variables corre-sponds to the order of memory processes to which theyare relevant. Recognition results are reported first be-cause they reflect only encoding differences. Recall re-sults then are discussed because, to the extent they differfrom recognition fmdings, they reflect effects on re-trieval. Cell means for the array of recognition and recallmeasures are reported in Tables 4 and 5, respectively.

Recognition. Consistent throughout each of the rec-ognition measures is a significant effect of age, the na-ture of which reveals a significant leaming decrement inelderly subjects. Furthermore, the consumer-orientedmeasures (brand recognition and claim recognition forthe test message) show meaningful interaction effects.For each measure a significant media x age interaction(brand F = 5.07, p < .02; claim F = 4.74, p < .03)occurs in which the leaming decrement of the elderly isgreater in the television condition. The interesting aspect

Table 4CELL MEANS FOR RECOGNITION MEASURES

Etderly Younger adultsN SM SN N SM SN

News claim recognition(maximum = 24)Print 14,68Television 13,50

14,20 14,0014,31 14.71

16,8817,05

17,3216,85

Test message claim recognition(maximum = 32)Print 21,68 20,65 20,87Television 21,85 23,38 22,93

Corrected brand name recognition(maximum = 7}Prim ,81 1,08 1,05Television 1,63 1,87 1,40

25,05 25,5928,76 28,35

3,364,66

3,984,94

15,8716,90

23,0927,65

2,125,48

N, no encoding instructions; SM, semantic instructions; SN, sen-sory instructions.

Table 5CELL MEANS FOR RECALL MEASURES

Recall for newsPrintTelevision

N

stories3,744,35

Recall for advertisementsPrint 1,21Television 1,60

Brand name recallPrint ,09Television ,84

Elderly

SM

2,602,38

1,50,88

,14,90

SN

2,673,64

,731,21

,48,71

78

23

2

Younger

N

,86,86

,14,00

,87,48

SM

6,277,40

2,362,95

1,272.45

adults

SN

4,967,45

1,613,30

,332,90

N, no encoding instructions; SM, semantic instructions; SN, sen-sory instructions.

of this interaction effect is that television increases leam-ing over print to a greater extent for young subjects thanfor the elderly.

Differential encoding-level effects across age groupswould provide evidence for encoding failures on the partof the elderly. The presence of the media manipulationsuggests that such effects should be assessed for eachmedium. For each recognition measure, contrasts of meansreveal that in the print medium semantic processingsignificantly improves (p < .05) memory over sensoryprocessing for the young but not the elderly. In the tele-vision condition, processing level had no effect on rec-ognition. Thus, evidence for a processing deficiency isfound in the print condition. However, the fact that tele-vision generally boosted recognition in comparison withprint suggests the basic principle of LOP (i.e., deepersemantic processing leads to improved memory) does nothold for television.

Aided brand recall. The pattem of results for aidedrecall of brand name conforms precisely to that for brandname recognition. Contrasts again show superior mem-ory for the young in the print condition when semanticprocessing is induced, but not for the elderly. Further-more, the boost in memory from television is much greaterfor the young.

Unaided recall. In free recall of advertisements andnews stories, elderly subjects generally show inferiormemory in comparison with young subjects. This find-ing is consistent with the recognition and aided recallfindings.

For advertising recall, interesting media effects asso-ciated with encoding level and age are found. Semanticprocessing improves recall under the print condition buthas a somewhat detrimental effect under television. Thisresult deviates somewhat from preceding findings in thatthis pattem occurs in both age groups. However, in con-formance with preceding findings, television boosts re-call for the younger subjects to a greater extent than itdoes for elderly subjects. For news story recall only thethree main effects are significant (all at p < .01).

62 JOURNAL OF /MRKETING RESEARCH, FEBRUARY 1987

DISCUSSION

Our findings lead to several important conclusions aboutinformation processing differences betweeh elderly con-sumers and younger consumers. First, elderly consumersclearly have a general leaming deficit. Elderly subjectsconsistently performed at a lower level on memory testsacross all conditions of leaming. Second, the source ofthis leaming deficit appears to be associated with the en-coding stage of information processing, that is, the pointat which information is placed into memory. This con-clusion is warranted because measures that aid or bypassthe retrieval stage (aided and recognition recall) andmeasures that require retrieval (unaided recall) both in-dicate memory deficiencies in the elderly.

Third, the elderly consistently leam less than youngeradults and the size of the leaming decrement increasesunder certain conditions. Specifically, upon exposure toprint material, the memory performance of younger sub-jects increases when they are instructed to process at asemantic level. This effect does not occur for elderlysubjects. This pattem of results for both recognition andrecall again points to a processing deficiency on the partof the elderly when exposed to print material. They areless capable of deep, semantic-level processing and,consequently, do not realize the increase in leaming thatthe young do.

Another situation that increases the leaming decre-ment of the elderly occurs within the television mediacondition. Initially, the inclusion of this manipulation wasbased on the expectation that, though its extemally pacednature would affect memory negatively for both agegroups by making memory organization more difflcult,the decline in leeiming would be greater for the elderly.However, the observed increase in the leaming deficitof the elderly occurs for opposite reasons. The televisionexposure increased memory performance for youngersubjects and had little or no effect on that for elderlysubjects. Furthermore, sensory processing yieldedequivalent or, in some cases, superior performance incomparison with semantic processing. This pattem of ef-fects is somewhat more pronounced for consumer-ori-ented than for news materials. This surprising result hasimplications for understanding not only consumer infor-mation processing by the elderly, but also general mediaeffects on consumer information processing.

The appropriate source of a theoretical understandingof this result appears to be theories of encoding otherthan LOP. The focus on encoding is appropriate becausethe finding occurred in both recognition and recall tests.Strict adherence to LOP is inappropriate because its fun-damental principle of improved memory in semanticprocessing did not hold. Still, any theoretical explana-tion must be consistent with the findings of a study thatused LOP as a framework for studying memory. In thiscase it must be consistent with the finding that sensoryprocessing can result in equivalent or superior memoryrelative to semantic processing.

Given the preceding criteria, a candidate for under-

standing the positive effects of television on memory isthe theoretical notion of encoding distinctiveness, a re-cent modification of LOP (Jacoby and Craik 1979; Ja-coby, Craik, and Begg 1979). Encoding distinctivenessrefers to the tendency to discriminate one stimulus fromanother at encoding by focusing on the unique featuresof each stimulus. The result is greater contrastive valueto what is encoded and retention of more stimuli withina given stream.

In comparison with newspapers, television offers theviewer greater opportunity for distinctively encoding in-dividual items within the stream of presented material.Television makes use of more sensory-oriented stimuluscomponents—nonverbal sound, voices, color, pictures,and motion. Thus, instructions to encode at a sensorylevel would promote the use of encoding distinctivenessand result in greater memory for television material thanfor newspaper material, in which fewer sensory codesare available to distinguish between stimuli and semanticprocessing must be induced to improve memory. Nel-son, Reed, and Walling (1976) offer a similar explana-tion for the superiority of pictures over words in mem-ory, thus making the encoding distinctiveness explanationa feasible one for findings showing that pictures im-proved memory for younger age groups but not the el-derly (Winograd and Simon 1980). In our study the lesspronounced positive effect of television on memory fornews material is intuitively consistent with this expla-nation. News stories would tend to "run together" morein terms of their sensory features than would advertise-ments. In terms of understanding age effects on infor-mation processing, aged consumers appear to be less ca-pable of distinctively encoding stimuli within a streamand form memory traces qualitatively inferior to thoseof younger consumers.

The proposed explanation for the positive effects oftelevision is offered post hoc, not as a definitive expla-nation. Another explanation could be simply that TV hasgreater ability to motivate processing and retain attentionthan print, though such an explanation does not accountfor the effects of sensory encoding within the TV con-dition. Regardless of the explanation, the positive effectsof television on memory do not suggest a general pref-erence for television over print as an advertising me-dium. Print advertising allows the leaming of a greaterquantity of information within a single message. Ourmeasures did not tap the quantity of leaming from anyone message. Instead, they assessed leaming at theawareness level with respect to product class, brand, ora specific claim within a message.

More important to our study's purpose are the impli-cations of the results for marketing to elderly consumers.The overall fmdings suggest that elderly consumers haveencoding deficiencies when processing information fromeither print or television advertisements. Such deficien-cies imply difficulties in promotional effort aimed at theelderly. With encoding as the source of information pro-cessing difficulties, the elderly would not benefit frompromotional efforts to increase recognition. Thus, point-

CONSUMER LEARNING DEFICIENCIES IN THE ELDERLY 63

of-purchase displays, for example, would not seem to beuseful as promotional tactics to supplement advertisinginformation, though they might stand on their own as in-store influences on purchase behavior by the elderly.Brand differentiation efforts aimed at the elderly shoulduse sensory-oriented cues when presented in self-pacedmedia, because semantic-based cues are less likely to benoted. Though leaming deficits in the elderly relative tothe young increase with television, the impact of tele-vision in comparison with print is still positive for theelderly. This effect and the greater tendency of the el-derly to view television suggest that television should bea greater portion of the media mix for firms marketingto the elderly than for those marketing to younger con-sumers.

The conclusions and implications should be temperedsomewhat because of potential weaknesses in the re-ported research. First, evidence supporting the successof the semantic manipulation in the elderly is equivocal.The direct manipulation check did not indicate a seman-tic induction in the elderly group, though it did in theyoung group. An indirect check, based on cognitive re-sponse data, provided evidence supportive of a success-ful semantic induction in both age groups. Second, stud-ies of age differences in leaming through comparisonsof age-cohort groups necessarily lack complete controlover age-related extraneous variables (education, occu-pation, etc.). Traditional approaches to reducing ex-perimental error (matched groups, blocking, analysis ofcovariance) do not overcome inherent difficulties in de-termining equivalent operational levels of these extra-neous variables. Still, our findings afford important the-oretical insights into consumer information processingby the elderly which translate into meaningful marketingconsequences, all of which warrant further research at-tention.

REFERENCES

Baltes, Paul, Hayne Reese, and Lxwis Lipsitt (1980), "LifeSpan Developmental Psychology," Annual Review of Psy-chology. 31, 65-110.

Bettman, James R. (1979), An Information Processing Theoryof Consumer Choice. Reading, MA; Addison-Wesley Pub-lishing Company.

Burke, Deborali and Leah Light (1981), "Memory and Aging:The Role of Retrieval Processes," Psychological Bulletin,90 (November), 513-46.

Childers, Terry L. and Michael J. Houston (1984), "Condi-tions for a Picture Superiority Effect on Consumer Mem-ory," Journal of Consumer Research. 11 (September), 643—54.

Cole, Michael and Barbara Means (1981), Comparative Stud-ies of How People Think. Cambridge, MA: Harvard Uni-versity Press.

Craik, Fergus 1. M. (1977), "Age Differences in HumanMemory," in The Psychology of Aging, J. Birren and K.Schaie, eds. New York: Van Nostrand.

and Robert S. Lockhart (1972), "Levels of Processing:A Framework for Memory Research," Journal of Verbal

Learning and Verbal Behavior, 11 (December), 671-84.• and E. Simon (1980), "Age Differences in Memory:

The Roles of Attention and Depth of Processing," in NewDirections in Memory and Aging, L. Poon et al., eds.. Hills-dale, NJ: Lawrence Eribaum Associates.

• and Endel Tulving (1975), "Depth of Processing andthe Retention of Words in Episodic Memory," Journal ofExperimental Psychology: General, 104 (September), 268-94.

Eysenck, Michael W. (1974), "Age Differences in IncidentalLearning," Developmental Psychology, 10 (June), 936-41.

Green, D. M. and J. A. Swets (1966), Signal Detection The-ory and Psychophysics. New York: John Wiley & Sons, Inc.

Jacoby, Larry L. and F. I. M. Craik (1979), "Effects of Elab-oration of Processing at Encoding and Retrieval: Trace Dis-tinctiveness and Recovery of Initial Context," in Levels ofProcessing in Human Memory, L. S. Cermak and F. I. M.Craik, eds. Hillsdale, NJ: Lawrence Eribaum Associates.

, , and I. Begg (1979), "Effects of DecisionDifficulty on Recognition and Recall," Journal of VerbalLearning and Verbal Behavior, 18 (October), 585-600.

Kausler, Donald H. (1982), Experimental Psychology and Hu-man Aging. New York: John Wiley & Sons, Inc.

Link, S. W. (1982), "Correcting Response Measures forGuessing and Partial Information," Psychological Bulletin,92 (September), 469-86.

Lutz, Richard J. and Scou B. MacKenzie (1981), "Construc-tion of a Diagnostic Cognitive Response Model for Use inCommercial Pretesting," working paper. University of Cal-ifornia, Los Angeles.

Lynch, John G., Jr. and Thomas K. Srull (1982), "Memoryand Attentional Factors in Consumer Choice: Concepts andResearch Methods," Journal of Consumer Research, 9 (June),18-37.

Meadow, H. Lee, Stephen C. Cosmas, and Andy Plotkin (1981),"The Elderly Consumer: Past, Present, and Future," in Ad-vances in Consumer Research, Vol. 8, Kent B. Moruoe, ed.Ann Arbor, MI: Association for Consumer Research,742-7.

Nelson, Douglas L., Valerie S. Reed, and John R. Walling(1976), "The Pictorial Superiority Effect," Journal of Ex-perimental Psychology: Human Learning and Memory, 2(September), 523-8.

Phillips, Lynn and Brian Stemthal (1977), "Age Differencesin Information Processing: A Perspective on the Aged Con-sumer," Journal of Marketing Research, 14 (November),444-57.

Ross, Ivan (1982), "Information Processing and the OlderConsumer: Marketing and Public Policy Implications," inAdvances in Consumer Research, Vol. 9, Andrew Mitchell,ed. Ann Arbor, MI: Association for Consumer Research,31-9.

Salthouse, Timothy A. (19S2), Adult Cognition: An Experi-mental Psychology of Aging. New York: Springer-Verlag.

Visvabharathy, Ganesan (1982), "Product Specificity in Pub-lic Policy Toward the Elderly," in Advances in ConsumerResearch, Vol. 9, Andrew Mitchell, ed. Ann Arbor, MI:Association for Consumer Research, 23—6.

Winograd, Eugene and Elliott W. Simon (1980), "VisualMemory and Imagery in the Aged," in New Directions inMemory and Aging, L. W. Poon et al., eds. Hillsdale, NJ:Lawrence Eribaum Associates.

ZeithamI, Valerie A. (1982), "Consumer Response to In-StorePrice Information Environments," Journal of Consumer Re-search, 8 (March), 357-69.