Subjective and Objective Measures of Product Knowledge Contrasted

ABSTRACT - The present paper contrasts subjective and objective measures of product knowledge. Findings from a study related to home computers reveal that the two set of measures are related, but also that they not are substitutable. Objective measures seem preferable when focusing on ability differences among consumers, while subJective measures should be preferred when focusing on motivational aspects of product knowledge.


Fred Selnes and Kjell Gr°nhaug (1986) ,"Subjective and Objective Measures of Product Knowledge Contrasted", in NA - Advances in Consumer Research Volume 13, eds. Richard J. Lutz, Provo, UT : Association for Consumer Research, Pages: 67-71.

Advances in Consumer Research Volume 13, 1986      Pages 67-71


Fred Selnes, Norwegian Fund for Market and Distribution Research, Oslo, Norway

Kjell Gr°nhaug, Norwegian School of Economics and Business Administration, Bergen, Norway


The present paper contrasts subjective and objective measures of product knowledge. Findings from a study related to home computers reveal that the two set of measures are related, but also that they not are substitutable. Objective measures seem preferable when focusing on ability differences among consumers, while subJective measures should be preferred when focusing on motivational aspects of product knowledge.


The purpose of the present paper is to contrast "objective" and subjective measures of product knowledge.

Product knowledge can be defined as product related information stored in memory, such as information about brands, products, attributes, evaluations, decision heuristics and usage situations (Marks and Olson 1981). The concept is considered being of crucial importance among consumer researchers, not at least due to the following:

- First, knowledgeable consumers are expected to possess superior ability in approaching new information. Due to more developed knowledge structures such consumers are assumed more able to interpret and integrate new information than are their less informed counterparts (Johnson and Russo 1984; Chase and Simon 1973).

- Second, product knowledge is assumed to impact the decision-heuristics applied in handling buying decisions. A few examples may clarify this point. Park and Lessig (1981) found informed consumers to make more use of functional attributes than did their less knowledgeable counterparts. Consumers low in product knowledge are assumed to be more inclined to seek external information than are consumers high in product knowledge (cf. Cox 1967; Newman 1977; Punj and Staelin (1983). An inverted U-shaped relationship between product knowledge and intensity of information search due to lower ability to handle and integrate among consumers low in product knowledge, have, however, been reported (Bettman and Park 1980).


Consumers develop product knowledge through search and use of information as well as through experience (Howard and Sheth 1969). Based on our above definition, the content of the knowledge domain may be described in terms of objects or brands known by the consumer, his or her knowledge about product attributes and usage situations, the ability to discriminate between product alternatives, and product evaluations.

In several studies the consumer's product class experience has been used as a proxy for his or her product knowledge (Bettman and Park 1980; Jacoby et al 1978; Katona and Mueller 1955; Newman and Staelin 1972). There are, however, several conceptual problems associated with the use of product experience as a measure of product knowledge.

- First, product knowledge may be developed through information search and use, and may thus be present without personal product experience.

- Second, increasing product experience may occur without subsequent increase in product knowledge. Bettman (1979) argues that development of product knowledge in general and decision heuristics in particular do not only depend on product experience as such, but also on how this experience is interpreted and causal inferences with regard to decision outcome made. An outcome as expected may lead to simplified decision heuristics, and thus increased experience may cause no change in product knowledge at all.

Subjective as well as objective measures have been used when studying consumers' product knowledge. Subjective measures are based on consumer's interpretation of what s/he knows, while objective measures are based on another person's evaluation of this knowledge. In consumer research there has been a tendency to prefer subjective measures (Brucks 1985; Park 1976), while there has been preference for objective measures among researchers from other disciplines such as social cognitive psychology (cf. Streufert and Streufert 1978). This has also been noted by Brucks (1985). Moreover Sujan (1985) has recently applied objective measures, and Brucks (1985) objective and subjective measures as well.

Whether subjective or objective measures are chosen should depend on the purpose of the research (Mitchell 1982). In our view objective measures are preferable when the research objective is related to the consumer's ability to encode new information or to discriminate and choose between product alternatives. Objective measures are related to the organization of the individual knowledge structure. Organization of information and knowledge structures is closely connected to the ability to filter out irrelevant information (Taylor and Crocker 1980); to retrieve relevant information from memory (Sternthal and Craig 1982); to rehearse or encode new information (Fiske, Kinder and Larter 1980). Knowledgeable consumers, objectively assessed, should therefore be superior to less knowledgeable consumers in encoding product related information, and be better able to choose an alternative close to the ideal or preferred.

Subjective evaluation of ones product knowledge should have significant impact upon the motivation to conduct various behaviors. A consumer perceiving s/he knows everything that matters about a specific product class, and the various bits of knowledge are in balance, will feel confident and perceive the level of risk and cognitive conflict associated with the buying decision as low or modest (Cox 1967). A confident consumer is among other expected to acquire less information, to seek less advice from friends and more information from commercial sources. Their confidence make them feel better able to handle any attempt of commercial "manipulation".

One possibility is that subjective measures are perfectly correlated with objective measures. That is novices perceive themselves as novices and experts perceive themselves as experts. Another possibility is that the two types of measures are modestly or not correlated. (A novice is here a person with a low level of product knowledge, objectively assessed. Similarly an expert is person with a high level of product knowledge.)

If the subjective measures don't match the objective measures, this implies that some real novices perceive themselves as experts, and some real experts perceive themselves as novices. Novices that perceive themselves as experts overestimate their own knowledge. Experts that don't perceive themselves as experts underestimate their ability judge and choose among the various alternatives.

An intuitive research question is: What is the relationship between subjective and objective measures? In our opinion the relationship could - in the language of factor analysis - be either orthogonal, oblique os perfectly correlated. In case of the latter outcome, perfect correlation, choice of product knowledge measure, i.e. objective or subjective, would not matter at all. In such a situation the consumer interpret what s/he knows from information stored in memory. If the two measures are orthogonal, and thus unrelated, the researcher should be extremely careful in choosing between which measure to use. In the case of correlated, but not perfectly correlated measures, perceptual knowledge will depend on the interpretation of what is actually stored as well as other individual and situational factors.


Below is reported on the methodology underlying the present study designed to empirically explore the relationship between subjective and objective product knowledge measures.

Due to lack of relevant secondary data, primary research was conducted. A crossectional survey study was found appropriate to explore the relationship between the two sets of measures due to the requirement of keeping the intra-individual product knowledge constant.

Personal computers were chosen as reference-product in the present study due to expected variation in product knowledge. The factual knowledge related to this product, also makes it relatively easy to construct objective measures of product knowledge.

The sample consisted of 297 students attending a one-year introductory program in management. The students were very heterogenous with regard to previous education (including students with high-school only as well as college degrees in engineering, sociology and law) and work experience (ranging from no experience at all to several years of well-paid middle-management training).

Data collecting procedure: The participants were asked to complete a questionnaire addressing their familiarity with personal computers, not being allowed to talk to each other during the test. The data collection was conducted during the first week of the program


The following subjective measures were used:

(1) Level of subjective knowledge, SEV, measured by the respondents agreement of the degree to which s/he knew the relevant decision criteria. (The same operationalization has been reported by Park [1976].)

(2) Confidence, OEV, measured by asking how the respondents thought their closest friend would evaluate their familiarity with personal computers.

(3) Advicegiving, ADV, assessed by questioning whether the respondents had been asked for advice.

(4) Selfevaluation of brand knowledge, OBK, assessed by asking about success in a hypothetical computer purchase

Objective measures; Level of differentiation, discrimination, and integration are important dimensions to describe and map the individual knowledge structure (Streufert and Streufert 1978). Level of differentiation reflects the number of salient dimensions in the consumer's cognitive domain, level of discrimination is related to the number of categories among which the consumer is able to discriminate, and level of integration relates to the ability to associate different dimensions. In addition to cognitive structure,product knowledge is related to terminology or understanding of the product language. The following objective measures were used.

(1) Level of differentiation, two measures

- ATT - number of attributes

- BKN - number of salient brands

(2) Level of discrimination DSC

- inferred from the question used to assess ATT (A score, ranging from 1 to 4 was given for each characteristic. The average score across characteristics were computed for each individual [see Appendix A].)

(3) Level of integration, INT

- interpreted from a picture drawn bs the respondents.

(4) Terminological complexity, TER

- based on twenty 'what-is-meant-by' - questions reflecting various aspects of personal comPuters.

Appendix A gives a more detailed description of the various measures. The level of each dimension, that is level of discrimination (DSC), was determined from a table developed during the research process (see Appendix B). A qualitative analysis on a sample of questionnaires was the input in producing this table. The table show that respondents used quite different terminology to describe the same aspect of the product. For example, some students referred to the 'size' or the 'capacity' of a PC, whereas other were more advanced and referred to 'number of bits', 'RAM', 'ROM' and similar to describe the same aspect of a PC. Eight different aspects of personal computers emerged in this analysis. The average respondent, however, mentioned only a few of these aspects. Even if the table is biased toward the author's perception of the product domain, other 'experts' on personal computers found the table to be quite sensible.

To obtain the DSC measure, each attribute or characteristic the respondent mentioned was given a number from 1-4, reflecting the level of complexity as illustrated in the table. The weighted attributes were summed and averaged:


DSC: Level of discrimination

ATT: Number of attributes

Ci : Characteristic or attribute i

wj : Weigh assigned to attribute i; wj = 1,2,3 or 4

Level of integration (INT) was interpreted from the chart illustrating how the respondent believed the various characteristics belonged together. The charts were given scores from 0 to 6. Zero points were given to those respondents that did not answer the question. Points were awarded on the basis of how the respondent managed to give a wholistic picture of the product, that is illustrate how the different characteristics were connected. The most advanced charts had clearly divided between software and hardware related characteristics, and hardware was further divided into categories like processing capacity and 'cosmetic' characteristics like terminal, display, printer and so on.

Level of terminology (TER) was inferred from the 20 questions reflecting the respondent's understanding of words and expressions often used in the product language. Each answer was given a score from 0 to 2, where 0 indicates no understanding; 'some understanding'; and 2 a good understanding of the word or phrase in focus. The 20 questions were summed to produce TER. Cronbach's a (Nunnally, 1978) was computed to analyze the internal consistency of the scale:


where k is the number of items, ai2 is the variance of the measuring instrument item (each question), and a 2 is the variance of the sum of the k items. Alpha for the 20 items was 0.898, indicating a very strong internal consistency among the 20 items


Below are summarized the main findings:



Table 1 reports means, standard deviations and correlation coefficients between the variables used.

The average correlation within the objective measures (i.e. off-diagonal elements) is .467 and the corresponding standardized item - alpha* is .81. The high alpha indicates that the measures correspond with some underlying score. The square roof of alpha equivalates the correlation of the test with the true score (Nunnally 1978), i.e. .9 with the true score. [*The standardized item a is described in Hull and Nie (1981). The computational formula is given by:   EQUATION  where r is the average correlation between items. This is equal to standardizing each item, and thus making the various measures comparable.]

The average correlation within the subjective measures is .5, and the corresponding standardized item-alpha amounts co .8. The square roof of Alpha is .89, i.e. the correlation coefficient between the subjective measures and the true score.

The average correlation coefficient between the subjective and objective measures is .38 (p < .100). Thus a significant relationship is present between the two sets of measurements. Due to the somewhat higher average within correlation for the objective (.467) and subjective measures (.5), there is, however, some support for separating the two sets of measures. Due to different scales involved (three - five points, numerical counts) some of the relationships might have been understated.

A principal component factor analysis was conducted to infer the underlying dimensionality of the variables. Inspection of the table revealed that all variables are loading high on the first factor, which indicate some common score. The first factor explained almost 50% of the variance, the second 13%. Number of attributes (ATT) and level of discrimination (DSC) load high on both the first and the second factor. The eigenvalue on the third factor dropped to 0.77. As we also expected two significant factors, we concluded only the two first factors in the remaining analysis.

By conducting an orthogonal rotation on the first two factors extracted, the following picture emerges:



Inspection of Table 2 reveals that the structural variables, except for number of salient brands, BEN, and the subjective variables together with TER (terminology) and BKN are clustered together. As discussed above, the two factors (subjective and objective) may not be orthogonal, but rather correlated. Oblique rotations were conducted.



Table 3 reports the results from an oblique rotation of the factor-Loading matrix. Delta equals -.6 and the correlation coefficient between the two factors is .37. Inspection of Table 3 reveals that all the subjective variables are loading on the first factor, while three of the five objective factors are loading high on the second factor. The two remaining objective variables, TER and BKN are loading on both factors, but most on the first one. In sum, the results to some extent support the assumption that product knowledge has subjective as well as objective aspects. Moreover, the two aspects are related.


Based on previous research two aspects of product knowledge were assumed at the outset of this paper. The reported findings support this assumption. The observed relationship between subjective and objective measures, does not, however, allow for straight forward recommendations, except for the obvious one that research purpose should guide the actual choice of measures. Based on the reported findings it can also be argued according to our previous discussion that objective measures are preferable when research is focusing on ability differences, while subjective measures should be preferred when preoccupied with motivational aspects of product knowledge.

The present piece of research including one product only, definitely has its limitations. However, the findings falsify the hypothesis that subjective and objective measures always can be substituted. The two aspects of knowledge are related, but the shape and degree of relationship will probably be influences by other individual variables, situational factors as well as by the product involved.

As objective and subjective measures of product knowledge are related but not substitutable, this implies that some consumers overestimate their own ability to buy the product, while others underestimate their ability. Future research should focus on these different types or constellations of knowledge and how the different types affect other consumer behavior variables. Most interesting would it be to investigate the relationship to information search intensity and direction, and also to the ability to select good alternatives in a product differentiated market.


Bettman, J.R. (1979), An Information Processing Theory of Consumer Choice, Reading, MA, Addison-Wesley.

Bettman, James R. and Park, C.W. (1980), "Effects of Prior Knowledge and Experience and Phase of the Choice Process on Consumer Decision Processes: A Protocol Analysis". Journal of Consumer Research, 8 (September), 197-207.

Brucks, M. (1985), "The Effects of Product Class Knowledge on Information Search Behavior". Journal of Consumer Research, Vol 12, No. 1 (June), 1-16.

Chase, U.G. and Simon, H.A. (1973), "Perception in Chess". Cognitive Psychology, 4 (January), 55-81.

Cox, D.F. (ed.) (1967), Risk Taking and Information Hand- ling in Consumer Behavior, Boston, Graduate School of Bus. Adm., Harvard University.

Fiske, S., Kinder, D., and Larter, M. (1983), "The Novice and the Expert: Knowledge based strategies in political cognition," Journal of Experimental Social Psychology, 19, 381-400.

Howard, J.A. and Sheth, J.N. (1969), The Theory of Buyer Behavior, New York, John Wiley & Sons, Inc.

Hull, C. Hadlei and Norman H. Nie (1981), SPSS: Update 7-9 McGraw-Hill Book Company.

Jacoby, J., R. Chestnut; and W. Fisher (1978), "A Behavioral Process Approach to Information Acquisition in Nondurable Purchasing", Journal of Marketing Research, XV (November), 532-544.

Johnson, E. and Russo, J.E. (1984), "Product Familiarity and Learning New Information". Journal of Consumer Research, 11, (July), 542-550.

Katona, G. and Mueller, E. (1955), "A Study of Purchase Decisions", in L.H. Clark (et.), Consumer Behavior: The Dynamics of Consumer Reactions, New York University Press, 30-87.

Marks, L.J. and Olson, J.C. (1981), "Toward A Cognitive Structure Conceptualization of Product Familiarity". Advances in Consumer Behavior, Ann Arbor, Association for Consumer Research, 145-150.

Mitchell, A.A. (1982), "Models of Memory: Implications for Measuring Knowledge Structures," in A. Mitchell (Ed.), Advances in Consumer Research, Vol IX, St.Louis 45-51.

Newman, J.W. (1977), "Consumer External Search: Amount and Determinants", in A.G. Woodside, J.N. Sheth and P.D. Bennett (eds.), Consumer and industrial Buying Behavior, New York, Elsevier North-Holland, Inc., 79-94.

Newman, J.W. and Staelin , R. (1972), "Prepurchase Information Seeking for New Cars and Major Household Appliances", Journal of Marketing Research, 9 (August), 249-257.

Nunnally, J.C. (1978), Psychometric Theory. New York; McGraw-Hill Book Company.

Park, C. (1976), "The Effect of Individual and Situational Related Factors on Consumer Selection of Judgement Models". Journal of Marketing Research, XIII (Hay), 144-151.

Park, C. and Lessing, V.P. (1981), "Familiarity and Its Impact on Consumer Decision Biases and Heuristics". Journal of Consumer Research, 8 (September), 223-230.

Punj, G.N. and Staelin, R. (1983), "A motel of Consumer Information Search Behavior for New Automobiles", Journal of Consumer Research, 9 (March), 366-380.

Streufert, S. and Streufert, S.C. (1978), Behavior in the Complex Environment, Washington D.C.: V.H. Winston & Sons.

Sujan, M. (1985), "Consumer Knowledge: Effects on Evaluation Processes Mediating Consumer Judgements", Journal of Consumer Research, Vol. 12, No. 1, (June), 31-46.

Taylor, S.E. and Crocker, J. (1981), "Schematic Bases of Social Information Processing". In E.T. Higgins, C.P. Hermann, and M.P. Zanna (Eds.), Social Cognition: The Ontario Symposium (Vol. t). Hillsdale, N.J.: Erlbaum.



Variable Operationalization

SEV     Below are given three descriptions (A, B, C) of product knowledge about personal computers. Read the alternative (A, B, C) and mark the alternative best describing your familiarity with personal computers.

A: I am quite unfamiliar with personal computers in the sense that I do not have any clear idea about which product characteristics are the important ones in providing me with maximum usage satisfaction.

B: I am somewhat familiar with personal computers in the sense that I have a somewhat clear idea about which product characteristics are important in providing me with maximum usage satisfaction.

C: I am quite familiar with personal computers in the sense that I have a very clear idea about which product characteristics are important in providing me with maximum usage satisfaction.

OEV      "How do you think your closest friends evaluate your familiarity with personal computers?"

A: Quite unfamiliar

B: Somewhat familiar

C: Quite familiar

ADV      "Does it happen that you are asked for adviceabout personal computers?"

A: Never

B: Sometimes

C: Very often

OBK      "If you were going to buy a personal computer today with your current level of knowledge about the various alternatives, how well do you think you would succeed?"

A: Very poor

B: Poor

C: Neither poor nor well

D: Well

E: Very well

ATT       "Let us assume that you are going to buy a personal computer. What characteristics (attributes/dimensions) would you use to evaluate the different alternative based on your present knowledge about personal computers?"







BKN      "If you were going to buy a personal computer, which alternatives would you consider?"

1.                                                         4.                                                  

2.                                                         5.                                                  

3.                                                         6.                                                  

TER      "What is meant by"

1. BASIC ....................................

2. TERMINAL .................................

3. CPU ......................................




20. 200 K.....................................

INT      "We will illustrate this question by an example. The task is to group the characteristics you gave on the previous question the way you mean they naturally belong together. The characteristics should first be divided into main categories, then further into sub categories and sub-sub categories and so on. To illustrate this we can think of softdrinks, and we assume that the following characteristics is given: sweetness, taste, carbonic acid, caffeine, sugar, color, bottle, glass, plastic, cork and cocktail mix. These can be divided into two main groups: 1) Content and 2) canning. Content can be divided into "tastechemicals" and color; canning can be divided into bottle and cans. These can further be divided into subgroups, and we draw the following picture:


We may not be able to place all the characteristics. This is quite natural, and you should not try place a characteristic unless you believe it is a natural connection (e.g. "coctailmix").

On this question, there is no right answer. What we want to know is the way you believe these characteristics belong together.

Your task is to make a similar chart with the characteristics you provided on the previous question.

You are not to make any corrections on the previous question, but you may add new characteristics on this task if someone comes to mind. Use the next page to make the chart".




Fred Selnes, Norwegian Fund for Market and Distribution Research, Oslo, Norway
Kjell Gr°nhaug, Norwegian School of Economics and Business Administration, Bergen, Norway


NA - Advances in Consumer Research Volume 13 | 1986

Share Proceeding

Featured papers

See More


Predicting memory-based consumer choices from recall and preferences

Zhihao Zhang, University of California Berkeley, USA
Aniruddha Nrusimha, University of California Berkeley, USA
Andrew Kayser, University of California, San Francisco
Ming Hsu, University of California Berkeley, USA

Read More


Feature A Benefactor or A Victim? How Charity Appeals with Different Protagonist Foci Affect Donation Behavior

Bingqing (Miranda) Yin, University of Kansas, USA
Jin Seok Pyone, University of Kansas, USA

Read More


I'm Scared, Want to Listen? Fear's Influence on Self-Disclosure

Anupama Mukund Bharadwaj, University of Washington, USA
Lea Dunn, University of Washington, USA
Joey Hoegg, University of British Columbia, Canada

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.