Public Knowledge of and Attitudes to Science: Alternative Measures That May End the "Science War

Science, Technology, & Human ValuesBauer et al. / Ending the Science War

Public Knowledge of and Attitudesto Science: Alternative MeasuresThat May End the “Science War”

Martin W. BauerLondon School of Economics

and Political Sciences

Kristina PetkovaPepka Boyadjieva

Bulgarian Academy of Sciences

Research on the public understanding of science has measured knowledge as acquain-tance with scientific facts and methods and attitudes as evaluations of societal conse-quences of science and technology. The authors propose a set of alternative concepts andmeasures: knowledge of the workings of scientific institutions and attitudes to the natureof science. The viability, reliability, and validity of the new measures are demonstratedon British and Bulgarian data. The instrument consists of twenty items and takes ten tofifteen minutes to apply. The authors present new results on the public understanding ofscience for further investigation. Differences in the representation of science arereported between the British and Bulgarian young elite, between the elite and the publicin Bulgaria, between natural and social science students, and between beginners andadvanced students. The use of these measures will extend the scope of science indicatormeasures used by the European Commission and the NSF, help the assessment of thesocialization effort in university training, and may even contribute to the peace processin the “science wars.”

Over the past 25 years, various instruments have been developed to mea-sure public understanding of science and scientific literacy. For this purpose,the concepts are usually divided into three parts: interest in science, knowl-edge of science, and attitudes to science. Most studies followed the initial

30

AUTHORS’ NOTE: We thank Dr. J. Stockdale (London School of Economics), Dr. J. Leach(Imperial College London), and Dr. S. Miller (University College London) for their support incollecting the U.K. student data in their courses and two anonymous reviewers for their construc-tive comments. The research was supported by a grant from the SOROS Foundation No.RSS/HESP 634/1995.

Science, Technology, & Human Values, Vol. 25 No. 1, Winter 2000 30-52© 2000 Sage Publications Inc.

proposals of Withey (1959) and Miller (1983) that culminated in the NSF[**PLEASE SPELL OUT NSF**] science indicator surveys (Miller, Pifer,and Ressmeyer 1991). In this context, one may want to mention the U.K. sur-veys of 1988 and 1996 (Durant, Evans, and Thomas 1989; Bauer and Durant1999), the European Union (EU) indicator surveys of 1989 and 1992 (Bauer,Durant, and Evans 1993; Bauer and Schoon 1993; European Commission[EC] 1994; Miller, Pardo, and Niva 1997), and comparable surveys in France(Boy 1987), India (Raza et al. 1996), Brazil (CNPq 1987), South Korea (Kim,Carter, and Stamm 1995), Canada (Einsiedel 1993), Bulgaria (Petkova et al.1997), Sweden (Fjaestad 1996), and China (Zhang and Zhang 1993). Thismomentum produced a database that offers the prospect of systematic com-parisons of popular scientific cultures at the turn of the twenty-first century(Bauer 1995; Durant et al. 1999).

There is very little scholarly discussion of the measurements of publicunderstanding of science, and what there is oscillates between rejection ofmeasurement (e.g., Irwin and Wynne 1996) and the plowing of establishedfurrows. Occasional constructive commentaries call for either (1) more opendebates on surveys and measurement (Beveridge and Rudell 1988), (2) newoperationalizations of old concepts (Miller 1998), (3) new statistical proce-dures for analyzing existing data (Miller 1998), or (4) the exploration of dif-ferent concepts of knowledge and attitudes to science. This article seeks tomake a contribution to the last of these.

We propose alternative measures of knowledge and attitudes. Knowledgeof the scientific institution is the third dimension of the public understandingof science, complementing existing measures of factual and methodologicalknowledge. To measure attitudes to science, we propose to assess publicagreement and disagreement with an account of science that has been calledthe “ideology of science.” An opportunity has arisen to put to the test, in Brit-ain and in Bulgaria, newly constructed instruments, and we use both contextsto test the viability, reliability, and validity of these measures. Here we focuson methodological discussions and present tentative results.

Institutional Knowledge of Science

Hitherto, public knowledge of science has been measured on one or twodimensions: acquaintance with scientific facts and scientific method. Factualknowledge is measured with a “knowledge quiz” in which respondents state,for example, whether it is true or false that the Earth goes round the sun.Methodological knowledge measures are less developed, but they mayinclude items on probability reasoning or the logic of experiments. These two

Bauer et al. / Ending the Science War 31

dimensions are correlated and, for most purposes, combined into a singlemeasure. However, some people are better at methods, and others better atfacts. For international comparisons, the two dimensions have separate diag-nostic value.

Curiously, many public understandings of science researchers ignore theinstitutional side of science: while most measures that are used in researchassess public acquaintance with the facts and methods, they neglect scientificinstitutions. We believe that it is important to adapt the concepts of the publicunderstanding of science to this purpose, for institutional images engenderpublic trust or mistrust. Trust in science is an intangible guarantee of the free-dom of enquiry in modern society and a key resource in a changing historicalcontext. The relationship between trust in and familiarity with science iscomplicated, and it is only logical to investigate it through the same object:the scientific institution. Studies of public knowledge of the political systemare in many ways a model: how does knowledge of the institutions breed trustor mistrust in these institutions (Bauer 1996)? Analysis of 13,000 openresponses to the question, “What does it mean to study something scientifi-cally?” shows that in Europe, 13 percent of respondents refer spontaneouslyto the functions of science as a social institution (Bauer and Schoon 1993). Sothe image of science that people have may be more important than the factsand methods they know in building trust in science as an institution.

In this article, we are proposing a metric scale for measuring institutionalknowledge of science. Twelve items cover issues from teamwork, peerreview, funding, prestige, autonomy, science policy, and international com-petitiveness to a country’s science base. Appendix A shows the wording ofthe items. Respondents identify for each item whether “true,” “false,” or“don’t know” (K1 to K11) or whichever response alternative (see K12)applies. We distinguish between true and false responses and correct andincorrect answers. Some statements need to be identified as false to give acorrect answer, while for others, the “true” response is correct. Thus, oneneeds to designate the correct answer for each item in advance.

The description of science as a social process is controversial. Anydescription of an institution is contentious because its future is at stake. How-ever, self-thematization is a necessary part of institutional processes and isoften delegated to specialist philosophers, historians, or sociologists (Luh-mann 1993). They produce descriptions that are a mixture of a “mission state-ment” about what ought to be and of what is actually the case. The differencebetween ethos and positive reality is often unclear in social reality. Even sci-entists need moral guidance, and many descriptions of science are thereforemore moralistic than realistic. To avoid the difficulties of competing ideals, at

32 Science, Technology, & Human Values

least temporarily, one resorts to some authority. For factual and methodologi-cal knowledge, we happily rely on scientific textbooks and expertise to deter-mine the correct answers in our questionnaires. For knowledge of scientificinstitutions, we rely on the observations of sociology and policy analysis(e.g., Crawford, Shinn, and Sorlin 1993; Coates 1995; Cole and Cole 1973;Hagstrom 1965; Glaser 1964; Price 1972; Price and Beaver 1966; Ravetz1971; Ziman 1995).

In contrast to factual and methodological knowledge of science, the insti-tutional reality of science varies with historical context. For each item, weestablish the correct answer for the specific context in which the data are col-lected. This requires calibration of the instrument. For each item, one deter-mines the correct answer (1) by assessing the evidence from policy analystsin the field and/or (2) from sociological studies on the issue. Our exampleshows that it is viable to determine empirically the correct response for eachof the items.

Attitudes to Science

Hitherto, the measurement of attitudes to science followed Withey (1959)and Miller (1983): it involved assessing the public evaluation of the societaloutcomes of science and technology.1 The technical discussions on attitudesto science evolved around two issues. First, researchers distinguish betweengeneral and specific attitudes. General items state claims such as “scientificdevelopments make our lives more healthy.” Specific items state claims suchas “new developments in computing lead to more unemployment,” allowingpeople to differentiate their evaluation according to scientific activities.Respondents indicate their agreement with a number of claims on a 4-pointLikert scale with a “don’t know” option. Sometimes respondents have anambivalent middle option (neither/nor). Results show that general and spe-cific attitudes are only weakly related (see Evans and Durant 1995) and thatresponses to general items are framed by the specific associations that cometo respondents’minds, which are unknown without further inquiry (Daamen,Verplanken, and Midden 1991). This suggests that even general attitudes arespecific responses. On the other hand, general attitude measures serve asmarkers in public debates. Let us take the following item: “on balance, sci-ence and technology are for the better of humanity.” The sharp decline ofpositive responses to this question between the 1960s and the 1980s fuelled adebate on technophobia in Germany (Jaufmann, Kistler, and Jansch 1989).This marker function of general attitude items in public debate is an


important matter for further research. The second issue concerns the wordingof items. Multivariate analyses consistently group items that, to express apositive attitude, need to be disagreed with and vice versa (see Evans andDurant 1995). Positively worded items are prone to acquiescent responsebias (Schuman and Presser 1981), and negative wording has more diagnosticvalue (see Bauer, Durant, and Evans 1993). To disagree requires thinking andproduces more variance and consistency. Therefore, a scale of five negativelyworded items is better than a large battery of positively worded items.

We suggest an alternative object of attitudes: the “nature of science” itselfinstead the consequences of science. Specialists argue over two problems:how does scientific knowledge develop (the problem of accumulation andcognitive change), and how does scientific knowledge differ from otherforms of knowledge (the demarcation problem)? In measuring public atti-tudes, we are not evaluating the merits of these arguments. Our question isempirical: where does the public stand in the controversy over the nature ofmodern science? Periodically, specialist discussions receive wider publicattention, albeit with delay. This allows us to distinguish empirically a con-tinuum between a more sedimented orthodox position and a more recentposition on the nature of science.

In a recent book, Ziman (1995) orders and evaluates claims on the natureof science that form a coherent ideology and serve to absolve scientists fromsocial responsibility. We take five of these claims and add three other contro-versial claims. For each of them, we ask the public to agree or disagree on a5-point scale. Appendix B gives the wording for all eight items of our attitudescale. A respondent’s agreement with the item expresses an “orthodox” view,and disagreement with most of them expresses a “skeptical” view of the epis-temic dignity of science. With our data, we test empirically whether thedimension “orthodox-skeptical” is psychometrically viable, reliable, andvalid for describing the public’s view of the nature of science. From the spe-cialist point of view, this may appear to be a crude reduction of complexity.However, bringing public opinion to bear on these issues, while far fromdecisive, may contribute to the debate.

Empirical Basis and Viability of Measures

The knowledge and attitude scales described were used in the UnitedKingdom and in Bulgaria in 1996. The questions are arranged into a self-completion questionnaire that is easily applicable and requires about ten min-utes to complete. Appendixes A and B show the instructions for each scaleand the wording of the items. The attitude items should precede the


knowledge items, as studies have shown that knowledge questions affectinterest or attitude responses but not vice versa (Gaskell, Wright, andO’Muircheartaigh 1993). The questionnaire also includes basic demo-graphic information such as the respondents’ age, sex, country, and field ofstudy to allow comparison of results across different subgroups ofrespondents.

Table 1 shows the empirical basis of this study, which distinguishesbetween the young elite and the general public. The young elite sample (n =366) consists of university students from both the United Kingdom and Bul-garia interviewed between May 1996 and January 1997 and includes under-graduate and graduate students as well as social science and natural sciencestudents. The general public sample is a two-stage national representativesurvey conducted in Bulgaria with a response rate of 80 percent (1,100 calls;n = 878; face-to-face interviews) from April 1996. The survey included ourscales as part of a larger study of the public understanding of science in a Bul-garia in transition (Petkova et al. 1997). No equivalent sample for the U.K.context is available. With this database, we are able to test the characteristicsof our measures in two contexts: among the young elite of two countries andamong the general public in one country.


Table 1. The Empirical Basis

Sample Type United Kingdom Bulgaria Sum

Young elite University students University students(N = 272): LSE (168), (N = 94): PlovdivUCL (78), and IC (28) University (57), survey

respondents withuniversity education (37) N = 366

General public None Representative sampleof the Bulgarian public(N = 878) N = 878

Sum N = 272 N = 972 N = 1,244

NOTE: The British elite sample includes undergraduate (n = 58) and graduate students(n = 110) in social sciences from the London School of Economics (LSE) and under-graduate students in natural science from University College London (UCL; n = 76) andImperial College London (IC; n = 28). The Bulgarian elite includes undergraduates insocial sciences and humanities at Plovdiv University (n = 57) and a section (n = 37) fromthe national survey: people younger than age thirty and with a university education,which means in the Bulgarian context most likely a technical or natural scientificeducation.

Reliability

Reliability refers to whether measures are internally consistent. Weexplore reliability as dimensionality and consistency of the scalar items. Canwe confidently assume that each of the items indicates a single underlyingconcept? We answer this question by conducting a factor analysis of our data.One-dimensionality is a requirement for an additive scale.

Consistency of Institutional Knowledge

Earlier studies showed that the dimensionality of knowledge varies acrosscontexts and deserves careful exploration in different contexts (Bauer,Durant, and Evans 1993; Bauer 1995). Appendix A shows the basic result foreach knowledge item. All knowledge items are coded 1 for a correct answerand 0 for an incorrect or a “don’t know” response. Our knowledge items aretwo-dimensional in both the elite and the general public context. Factor


Table 2. Factor Analysis of Knowledge Items for the EliteSample

Item Autonomy Function

K3 Mutual criticism .702K8 Need to worry .660K12 Industry funding .605K7 Not self-determined .543K4 Social Security funding .512K2 Teamwork .623K9 Top priority investment –.361 .601K1 International contacts .526 .547K5 Peer review .529K10 Research planning .419K6 Recognition as reward .381K11 Natural science student body .418

Factor 1 Factor 2

Eigenvalue 2.69 1.64Explained variance (%) 22 14

NOTE:N = 366; factor loadings < .30 are deleted.The sampling adequacy is satisfactory(KMO = .69), SPSS WINDOWS principle component analysis with an oblique factorrotation is used, and the structure matrix is presented.

analysis reveals two factors explaining one-third of the variance in both sam-ples. Tables 2 and 3 show the factor solutions.

Comparing the solutions shows that the factors are reversed, while thegrouping of the items remains the same. This may indicate the differentialimportance of these two dimensions for the elite and for the general public;the first factor is the more important one in each sample. An interpretation ofthe grouping of the items points to their “truth value.” One factor combines allitems for which a “false” response is correct; the other factor combines allitems for which a “true” response is correct. This may indicate that thedimensionality is a consequence of the wording of items. A second interpre-tation of this grouping is substantive. Awareness of scientific teamwork, of itsinternational network, and of the high priority of science in industrializedcountries makes one factor. We call this knowledge of the functions of sci-ence (henceforth function). Awareness of scientists’ mutual criticism, of thelimits of their research autonomy, and of their insecure social position makeswhat we call the knowledge of the limits of autonomy (henceforth autonomy).

The hierarchy of the factors and their intercorrelation indicates differentrepresentations of science. For the young elite, autonomy is primary,


Table 3. Factor Analysis of Knowledge Items for the GeneralPublic

Item Function Autonomy

K2 Teamwork .646K1 International contacts .639K9 Top priority investment .631K10 Research planning .569K5 Peer review .560K6 Recognition as reward .518K8 Need to worry .693K7 Not self-determined .584K3 Mutual criticism .575K4 Social Security funding .509K11 Natural science student body .320

Factor 1 Factor 2

Eigenvalue 2.61 1.26Explained variance (%) 24 12

NOTE:N = 878; factor loadings < .30 are deleted.The sampling adequacy is good (KMO= .79), SPSS WINDOWS principle component analysis with an oblique factor rotation isused, and the structure matrix is presented.

explaining 22 percent of the variance, and the two factors are independentdimensions of their representation of science. Function is not correlated withautonomy (r = .08,ns). By contrast, for the general public, function is pri-mary, explaining 24 percent of the variance, and the two dimensions are cor-related in the public representation of science. Knowledge of the functioningof science relates to awareness of its limited autonomy. This feature of therepresentation of science deserves further exploration among differentpopulations.

Hence, the analysis of the knowledge items suggests two dimensions, afour-item scale measuring autonomy and a six-item scale measuring func-tion. Table 4 shows the reliability of the two subscales. The strength of a fac-tor correlates with the consistency of its items. The larger the factor, the betterthe reliability of the items that load the factor. Consistent with the above, wefind a reversed hierarchy of the two scales for the public and for the youngelite. Function is a better measure for the general public’s knowledge of sci-ence (alpha = .66), while autonomy is better for the young elite (alpha = .60).

Consistency of Attitudes to Science

Appendix B shows the basic results for the attitude items. First, we need toinspect the response distributions of the attitude items that differ between theelite and public samples. The general public tends to agree and the elite tendsto disagree with the statements. An exception is item 8, the “outcome ofexperiments,” in which both groups tend to disagree. Attitude items are proneto acquiescent response bias: that is, respondents tend to agree with whateverclaim, particularly when confronted with an unfamiliar issue. However, our


Table 4. Explained Variance in Factor Analysis and Consis-tency of Knowledge Subscales

Explained Variance (%) Cronbach Alpha

Subscale Elite Public Elite Public

Function (six items) 14 24 .50 .66Autonomy (four items) 22 12 .60 .45

NOTE: For both samples, the six-item scale for function and the four-item scale forautonomy is used as identified by the factor analysis of the general public sample (seeTable 3).

responses show only a small bias. Item 8 has the lowest “completely agree”score of 1.1 percent for the elite and 8.5 percent for the public sample. Weconsider this the upper limit of a consistent response bias in our data. Ourrespondents are discriminate in their responses, rather than responding with aconsistent agreement.

The attitude items are one-dimensional for both the elite and the generalpublic. A single factor explains 40 percent of the elite variance and 33 percentof the public variance, as shown in Table 5. The three items—“all science isgood” (5), “science is rational and objective” (4), and “some day scientificknowledge will present the true picture of the world” (7)—interpret the scalebest in both samples. “Scientific inquiry can know no limits” (6) is an impor-tant marker for the general public and less so for the young elite. The reliabil-ity of the attitude scale is high: Cronbach’s alpha = .78 for the elite and .69 forthe general public. Further analysis shows that the consistency of the attitudemeasure improves to alpha = .73 when the three items with the lowest loadingare excluded. However, the gain is small and until further evidence, we canconsider the items as a one-dimensional attitude scale and internallyconsistent.


Table 5. Factor Loadings of Attitude Items

Item Elite Sample General Public

ATT5 All science is good science .748 .668ATT4 Rational and objective .716 .747ATT7 Will provide true picture .690 .681ATT3 Science cannot be blamed .649 .623ATT6 Science has no limits .576 .672ATT2 Science is policy neutral .655 (.395)ATT1 Knowledge accumulation .511 (.426)ATT8 Agreement of experiments .452 (.179)Eigenvalue 3.19 2.67Explained variance (%) 40 33N 366 878Alpha .78 .69

NOTE: Item att8 (“all scientists normally agree on the outcomes of a particular experi-ment”) has a much lower factor loading in the Bulgarian sample than in the elite sample;it is likely to measure a different dimension of attitudes in the general public. One maywant to consider excluding the item from an additive scale. The sampling adequacy forconducting factor analysis of the attitude items is excellent (KMO = .85 for the elite sam-ple and .80 for the public).

Validation

Validation refers to the assessment of whether a scale measures what itwas constructed to measure. Most validation procedures make use of someexternal criteria as a frame of comparison. We are exploring the validity ofour knowledge and attitude measures in three different ways. First, how pow-erful are the scales for differentiating between subsamples? Second, a mean-ingful and surprising relationship between knowledge and attitudes is indica-tive of valid measures. Finally, a multiple-regression model for attitudesshows meaningful structure in our data. The validation of a new instrument isalways balancing (1) the reproduction of results that are concurrent with pre-vious ones and (2) the production of unexpected results that can be consid-ered as new insights (Bartholomew 1996).

The Power to Differentiate

We consider agreement with our attitude statements as expressing theorthodox, more idealist view of science, while disagreement expresses theskeptical, more realist view of science, which is consistent with the recentspecialist discourse on the nature of scientific activity. Figure 1 shows the dis-tribution of attitudes. In their views on modern science, Bulgarian students(median = 19; mean = 18.65;SD= 4.87; range = 8-35) are more traditionalthan their English colleagues (median = 27; mean = 26.70;SD= 4.62; range =12-40). The difference is striking in both median and mean with similarvariance.

On the other hand, regarding knowledge, British students are less aware offunction (mean = 2.96; median = 3.0;SD= 1.48;n= 280) than their Bulgariancolleagues (mean = 4.13; median = 4.0;SD= 1.24;n= 94). However, the Brit-ish are more alert to autonomy (mean = 2.70; median = 3.0;SD= 1.13) thanthe Bulgarians (mean = 1.94; median = 2.0;SD= .93). Whether you answerthe questionnaire in Bulgaria or in the United Kingdom makes a difference:the two contexts and our scales clearly covary.

Among the Bulgarian public, education and age make a difference toknowledge. The young and the more educated are more knowledgeable. Thisis not surprising: we expect young people to be better educated than their par-ent generation in both countries. The Bulgarian elite is more alert to auton-omy (median 2.00;n= 94) than the wider public (median = 1.00;n= 878) andmore aware of the function of science (mean = 4.13;SD= 1.24) than the pub-lic (mean = 3.52;SD= 1.78). “Don’t know” responses on knowledge ques-tions are significant in themselves (Bauer 1996). Our data confirm the known


patterns: Bulgarian women,2 the more elderly, and the less educated are morelikely to declare their ignorance of scientific knowledge. The young elite ismore skeptical about modern science (median = 19; mean = 18.65;SD= 4.86;range = 8-35) than the general public, who tend to hold more orthodox atti-tudes (median = 17; mean = 17.05;SD= 4.90; range = 8-35). We think this isconsistent with common sense, according to which “familiarity breeds con-tempt”; students are closer to science than the public and therefore morelikely to have a realistic evaluation of its activities.

Our British sample includes natural and social science students; the lattergroup includes both undergraduates and postgraduates. According to com-mon sense, we expect that undergraduates are more traditional than post-graduates as exposure to science breeds more skeptical views. British socialscience students (mean = 27.47;SD= 3.71;n = 58) are more skeptical thantheir natural science colleagues (mean = 24.73;SD= 4.86;n = 104). How-ever, the variance of the latter is larger, which indicates a wider range ofviews. In social science, the education level does not influence the views:undergraduates (mean = 27.47;SD= 3.71;n = 58) are as skeptical as post-


SCAN TO COME

Figure 1. Distribution of summated attitudes to science among the Bulgarianand British young elite in percentages of the sample for 1996-1997.

NOTE: The left-hand side indicates a more orthodox and idealistic appreciation of sci-ence, and the right-hand side indicates a realistic appreciation of modern science(n-Bulgaria = 94; n-Britain = 272).

graduates (mean = 27.99;SD= 4.27;n = 110). However, the variance of thelatter is larger, indicating a polarizing of attitudes with progressingeducation.

Our scales clearly covary with the country, age, sex, and level of educationof respondents; their education in natural or social sciences; and their under-graduate or postgraduate status. The new instruments clearly have the powerto discriminate these different contexts. Our results show that the public ismore traditional than the elite, younger students are more traditional thanolder ones, and natural science students are more traditional than social sci-ence students. The young are more knowledgeable than the old, and whetheryou are a member of the elite or of the general public, being in Britain or inBulgaria alerts you to different facets of the institution of science. This hasface validity. In particular, social science students are closer to the specialistdiscourse on science than the public. The latter’s views more likely reflectpast debates. We must also consider the postwar history of Britain and Bul-garia. Bulgaria was until 1989 a communist country, where scientism wasstate ideology. Marxist scientism viewed science as the only privilegedknowledge about the world, to which students had to subscribe to make prog-ress in the system. Bulgarian universities did not change fundamentally in theperiod between these political changes and the collection of our data. The sci-entistic ideology continued to be the curriculum for the young elite. By con-trast, in Britain, the nature of science has been controversial for some time,and no single doctrine is canonical at universities. Our observed differencesbetween the British and the Bulgarian young elite are consistent with whatone expects under these different circumstances, and they are the focus of astudy comparing the two cultures of popular science (Petkova et al. 1997).

The Relationship betweenKnowledge and Attitudes

The relationship between knowledge and attitudes to science is controver-sial and contradictory. There are strong commonsense expectations amongscientists and policy makers that “to know it is to like it”: more knowledge ofscience leads to more support for science. This expectation is consistent witha traditional view of science: as the rational pursuit of objective “truth” aboutthe world that we cannot do other that like[**WORD MISSING?**] .

Again we can present a differentiated picture. The correlation between thetwo knowledge scales of function and autonomy is different for the elitegroup and the general public. Among the elite, the correlation is weak butnegative: those who know about the functions are likely to be less aware of


autonomy and vice versa (r = –.18,p< .001). In the Bulgarian public, the twotypes of knowledge are positively correlated (r = .23,p < .001): those whoknow more about function are also more aware of autonomy. This is an inter-esting difference in the representation of science between the young elite andthe wider public that warrants further investigation.

Among the elite, awareness of the limits of autonomy is related to moreskeptical attitudes to science (r = .38; p < .001), while among the public,insights into autonomy are not related to their attitude to science (r = .08,ns).However, knowledge of function correlates with more traditional attitudes toscience (r = –.31;p< .001) among the elite and the public (r = –.33;p< .001).Insights into the autonomy of science engender a skeptical view of modernscience but only among the elite.

Differences in the representation of science among educated and less-educated publics have been demonstrated in earlier studies (Boy 1987). Elitegroups seem to distinguish more clearly between fact and value. The patternof correlation between knowledge and attitudes among our elite and publicsample shows partly concurrent validity but also suggests more differentiatedrelationships that are worth further investigation.

Predicting Attitudes to Science

Finally, we test our measures with two multiple-regression models. In theelite model, we include the two knowledge scales and three dummy vari-ables: sex (1 = male), level of studies (1 = undergraduate), and country (1 =United Kingdom). In the public model, we include the two knowledge scales,sex as dummy (1 = male), and age and level of education as ordinal variables.Table 6 shows the results with attitude as the criterion variable. Different pic-tures emerge for the elite group and for the public. A high score on the attitudescale indicates a skeptical view of science and a low score a more traditionalview. Hence, a positive correlation indicates that skeptical attitudes are thedirect outcome. Bivariate correlation shows that the elite young men are morealert to autonomy than women, and undergraduates are better at functionknowledge than postgraduates (however, this may be due to the fact that thepostgraduates are all in the social sciences).

The elite model explains 42 percent of the attitude variance. British stu-dents are considerably more skeptical about modern science than Bulgarianstudents (r-part = .37). A British student will, on average, have an attitudescore that is 10 percent, or 4 points, higher than a Bulgarian student. Functionknowledge goes with traditional attitudes; alertness to autonomy goes withskeptical attitudes, all else being equal. Male students view science more


traditionally than female students, as do undergraduates compared to post-graduates. Among the Bulgarian public, higher education is related to a moretraditional attitude to science (r = –.12;p< .001). Controlling for knowledge,this relation with education disappears. Knowledge explains 18 percent ofthe attitude variance. Function knowledge increases traditional attitudes(r-part = –.42), while alertness to autonomy increases skeptical attitudes toscience (r-part = .13), controlling for sex, age, and education level. Educationlevel does not make any independent contribution. This confirms the resultsalready known.

Summary and Conclusion

Research on the public understanding of science has to date considered“knowledge” to be the acquaintance with scientific facts and methodologyand “attitudes” to be the evaluation of societal consequences of science andtechnology. In this article, we propose alternative concepts and measures ofknowledge and attitudes for research on the public understanding of science.We measure knowledge of science as the acquaintance with the empiricalfacts about scientific institutions, and we measure attitudes to science byagreement or disagreement with statements that are consistent with the ideol-ogy of science identified in the specialist literature. This allows us to assess


Table 6. Multiple Regression Analysis of Attitude to Science

Variable Beta Part-r p Dummy Variable

Elite modela

Country .46 .37 < .001 (1 = United Kingdom)Autonomy .18 .16 < .001Sex –.15 –.14 < .001 (1 = male)Function –.15 –.13 < .01Graduate –.13 –.12 < .01 (1 = undergraduate)

General public modelb

Function –.45 –.42 < .001Autonomy .14 .13 < .001

NOTE: The criterion variable in both models is attitude to science. We use the SPSSWINDOWS stepwise procedure.Adjusted R-square indicates the overall strength of themodel in terms of explained variance; the part correlation indicates the relative impor-tance of the predictor.a. Multiple R = .66. Adjusted R-square = .43. F = 54.69. p < .001. N = 366.b. Not significant are sex, age, and level of education. Multiple R = .42. AdjustedR-square = .18. F = 94.98. p < .001. N = 878.

institutional knowledge and attitudes on a continuum from traditional-idealistic to realistic-skeptical. We have discussed the viability, reliability,and validity of these measures on the basis of British and Bulgarian data. It isour main purpose to present and discuss new concepts of public understand-ing of science and to make new measures available for researchers to useshould an opportunity arise for a large-scale survey of the public.

Measuring institutional knowledge of science raises specific difficulties,and a word of warning is needed. The institutional facts of science are contin-gent to time and space. Correct answers to questions on “institutional facts”in one context may be incorrect answers in another context. Therefore, theknowledge items proposed here require careful calibration within the spe-cific country context that is surveyed. This is in contrast to the well-knownquiz of scientific facts and methods in which researchers can assume univer-sally correct answers to apply. We have demonstrated within the British andBulgarian contexts that this calibration of the measures is both necessary andpossible. Our multivariate analysis shows that institutional knowledge of sci-ence varies on two dimensions: the function of science is indicated by sixconsistent items, and the autonomy of scientists is indicated by four consis-tent items. The former is more salient and reliable among the general public;the latter is more salient and reliable among the educated young elite. Ourattitude scale with eight items is one-dimensional and reliable for all practicalpurposes.

From a new instrument, we expect a mixture of concurrent validity andnew results. The latter is important for avoiding the tautology of new mea-sures needing to replicate what is already known. The power to yield surpris-ing results is an important feature of a new instrument. Our validation showsthat the scales differentiate meaningfully between the educated elite of Brit-ain and Bulgaria, between the educated elite and the general public in Bul-garia, between natural and social science students, and between beginnersand more advanced students. Interesting patterns of correlation between thetwo dimensions of knowledge, between knowledge and attitudes, andbetween gender, age, education, and attitudes suggest further investigationsinto the representation of science among various publics. Our results clearlymap differences in the representation of science as an institution among theeducated elite and the general public. In summary, we validated empiricallythe assumption of a hypothetical continuum from idealistic-traditional torealist-skeptical attitudes toward modern science and its relation to twodimensions of knowledge of the scientific institution.

We foresee at least three immediate applications for these new measures.First, it may be useful to bring public opinion to bear on the recent contro-versy among natural scientists and humanists that has become known as the


“science wars.” Public opinion will not solve the sociological and philosophi-cal issues involved but may make a contribution to the peace process in thisdrole de guerre. Second, science-indicator studies both in Europe and NorthAmerica may extend the scope of existing measures to include knowledge ofand attitudes to the scientific institution at little additional cost. And finally,researchers may monitor and compare the development of students’ imagesof science as they progress in their studies. Our measure is easily applicablewithin the university context both to monitor and evaluate socializationefforts and effects. We can only encourage colleagues to apply this newinstrument, as we will do in the future.

Appendix AInstitutional Knowledge Scale:

Wording and Response Distribution

All figures are percentages; E = elite, P = public, C = correct answer;N-elite = 366;N-public = 878. Items K1 to K10 are equally correct in both contexts in 1996; K11 andK12 are context specific. K11 uses the wording “in this country” in Bulgaria, whichmakes a “true” response the correct answer; for the United Kingdom, a “false”response is the correct answer. For K12, in Bulgaria, the government is the mainsource of research funding, which makes “government” the correct answer; in theUnited Kingdom, “industry” is the correct answer. K12 is not used in the general pub-lic survey.

Instruction: “Please read the following statements carefully and give your answerby ticking the true/false or don’t know response option.”

Don’tItem Wording Group True False Know

K1: Scientists do research and discuss their workwith colleagues in other countries.

CE 73.5 7.0 19.5P 62.1 6.7 31.2

K2: Scientific research is mostly teamwork.C

E 71.7 13.1 15.2P 71.8 9.9 18.3

K3: Scientists do not criticize each other’s work.C

E 4.5 85.5 9.6P 22.9 45.0 32.1


K4: Industrialized countries spend more onscientific research than on social security.

CE 23.3 33.2 43.6P 43.3 18.5 38.3

K5: Scientific results are checked by scientificexperts before they are made public.

CE 41.7 32.1 26.2P 48.1 13.0 38.9

K6: The reward of scientific research is recognitionrather than money.

CE 60.2 21.4 18.4P 57.4 25.5 17.1

K7: Scientists decide for themselves what researchthey conduct.

CE 33.4 47.1 19.5P 53.6 20.6 25.8

K8: Scientists are in a good position and do nothave to worry about funding of their research.

CE 4.5 84.5 11.0P 33.6 45.0 21.4

K9: For the industrialized countries investment inscience is top priority.

CE 33.4 41.2 25.4P 65.8 5.8 28.4

K10: The planning and steering of research isnowadays a matter of government policy toensure international competitiveness.

CE 44.9 21.1 34.0P 46.4 21.6 32.0

K11: In European countries (in this country) morethan half of all students study to become naturalscientists or engineers.

C-Bul C-UKE 18.4 34.0 47.6P 68.2 0 31.8


Don’tItem Wording Group True False Know

(continued)

Don’tItem Wording Group Government Industry Foundations Know

K12: Who do you think spendsmost on scientific researchin this country?

C-Bul C-UKE 13.1 49.5 12.6 24.9

Appendix BResponse Distribution on Attitude Item for Both Samples

E = elite sample and P = public sample (N-elite = 366;N-public = 878). Instruc-tion: “This is a short questionnaire about the different ideas people hold about thecharacter of the scientific enterprise. Please indicate whether you agree or disagreewith the following statements by ticking the appropriate answer.”

Completely Neither/ CompletelyItem Wording Agree Agree Nor Disagree Disagree

(1) (2) (3) (4) (5)

Att1: Scientific knowledgeaccumulatescontinuously.E 37.7 43.3 12.0 5.6 1.3P 57.9 19.4 16.7 4.1 1.9

Att2: Science is policyneutral.E 8.0 10.2 22.5 40.6 18.7P 35.6 14.9 25.4 16.5 7.5

Att3: Science cannotbe blamed for itsmisapplication.E 19.0 18.4 24.3 30.7 7.5P 44.0 21.0 24.5 7.1 3.5

Att4: Science is rationaland objective.E 15.8 29.4 24.1 23.3 7.5P 42.6 24.7 26.2 4.7 1.8

Att5: All science isgood science.E 12.6 8.6 19.3 34.8 24.9P 52.6 20.2 18.3 4.4 4.4


Appendix A Continued

Att6: Scientific inquirycan know no limits.E 20.6 25.4 19.3 26.2 8.6P 56.5 19.6 17.3 4.6 2.1

Att7: Some day scientificknowledge will presentthe true pictureof the world.E 10.4 15.8 26.7 27.3 19.8P 48.1 25.4 19.7 4.0 2.8

Att8: All scientists normallyagree on the outcomeof a particularexperiment.E 1.1 5.1 15.0 49.2 29.7P 8.5 12.3 35.3 27.0 16.9

Notes

1. Scientists often stress the difference between science and technology (e.g., Wolpert 1992).But from the public point of view, it is difficult to see a difference. Science is felt in everyday lifethrough its applications to health care of in consumer products[**WORD MISSING?**] . Sci-ence policy focuses on research and development and brings science, technology, and industryinto an ever-closer complex that is best characterized as “technoscience.” By both origin and byethos, science is not reducible to instrumental rationality. It may be desirable to preserve this dis-tinction in other contexts.

2. In the elite sample, a regression analysis of “don’t know” responses as criterion anddegree, sex, age, and country as predictors, sex is the only variable in the equation explaining just2 percent of the variance (r = .14,F = 7.11,p = .008).

References

Bartholomew, D. 1996.The statistical approach to social measurement. San Diego, CA: Aca-demic Press.

Bauer, Martin. 1995. Industrial and post-industrial public understanding of science. Paper pre-sented to the Chinese Association of Science and Technology, International Conference onPublic Understanding of Science, October, Beijing.

. 1996. Socio-demographic correlates of DK-responses in knowledge surveys: Self-attributed ignorance of science.Social Science Information35:39-68.


Completely Neither/ CompletelyItem Wording Agree Agree Nor Disagree Disagree

(1) (2) (3) (4) (5)

Bauer, Martin, and John Durant. 1999. Trends in public understanding in Britain 1988-1996.Technical report, LSE/Science Museum, London.

Bauer, Martin, John Durant, and Geoffrey Evans. 1993. European public perceptions of science.International Journal of Public Opinion Research6 (2): 164-86.

Bauer, Martin, and Ingrid Schoon. 1993. Mapping variety in public understanding of science.Public Understanding of Science2:141-55.

Beveridge, A. A., and F. Rudell. 1988. And evaluation of “public attitudes towards science andtechnology” inScience Indicators. Public Opinion Quarterly52:374-85.

Boy, Daniel. 1987. Today’s myth of science: How sciences are perceived, categorized and evalu-ated in contemporary society. Unpublished manuscript.

CNPq. 1987.O que of brazileiro pensa da ciencia e da tecnologia?2d ed. Brazilia and Rio deJaneiro, Brazil: CNPq.

Coates, D., ed. 1995.Economic and industrial performance in Europe. Aldershot, UK: Elgar.Cole, Jonathan, and Stephen Cole. 1973.Social stratification in science. Chicago: University of

Chicago Press.Crawford, Elspeth, Terry Shinn, and S. Sorlin, eds. 1993.Denationalization of science: The con-

text of international scientific practice. Sociology of Science Yearbook 1992. Dordrecht, theNetherlands: Kluwer.

Daamen, Danker,*FIRST NAME? Verplanken, and Cees Midden. 1991. Cognitive structuresin the perception of modern technology.Science, Technology, & Human Values15:202-25.

Durant, John, Martin W. Bauer, George Gaskell, Cees Midden, Miltos Liakopoulos, and L. Scholten.1999. Industrial and post-industrial public understanding of science. InBetweenunderstand-ing and trust: The public, science and technology, edited by M. Dierkes and C. vonGrote.Reading, UK: Harwood.

Durant, John, Geoffrey Evans, and Geoffrey Thomas. 1989. The public understanding of sci-ence.Nature340:11-14.

Einsiedel, Edna. 1993. Canadian public understanding of science.International Journal of Pub-lic Opinion Research6:35-44.

European Commission (EC). 1994.The European report on science and technology indicators1994. Brussels: European Commission.

Evans, Geoffrey, and John Durant. 1995. The relationship between knowledge and attitudes inpublic understanding of science.Public Understanding of Science5:57-74.

Fjaested, Bjorn. 1996.Public perception of science, biotechnology, and a new university. Oster-sund: Mid-Sweden University.

Gaskell, George, Dan Wright, and Colm O’Muircheartaigh. 1993. Measuring scientific interest:The effect of knowledge questions on interest ratings.Public Understanding of Science2:39-58.

Glaser, B. 1964.Organizational scientists: Men in professional careers. Indianapolis, IN: BobbsMerrill.

Hagstrom, Warron. 1965.The scientific community. New York: Basic Books.Irwin, Alan, and Brian Wynne, eds. 1996.Misunderstanding science? The public reconstruction

of science and technology. Cambridge, UK: Cambridge University Press.Jaufmann, D., E. Kistler, and G. Jansch. 1989.Jugend und Technik. Frankfurt: Campus.Kim, Hak-Soo, R. F. Carter, and K. R. Stamm. 1995. Developing a standard model of measuring

the public understanding of science and technology.Journal of Science and Technology Pol-icy Korea7:2.

Luhmann, Niklas. 1993.Die Wissenschaft der Gesellschaft. Frankfurt: Suhrkamp.Miller, Jon D. 1983. Scientific literacy: A conceptual and empirical review.Daedalus, Spring,

29-48.


Miller, Jon D. 1998. The measurement of civic scientific literacy.Public Understanding of Sci-ence7:203-23.

Miller, Jon D., Rafael Pardo, and Fujio Niva. 1997.Public perceptions of science and technol-ogy: A comparative study of the European Union, the United States and Japan. Madrid: Fun-dacion BBV.

Miller, Jon D., Linda Pifer, and T. J. Ressmeyer. 1991.Public attitudes toward science and tech-nology 1979-1990. Integrated Codebook, version 1.1. Chicago: Academy of Sciences.

Petkova, Kristina, T. Boyadjiev, Galin Gornev, Ivan Tchalakov, and Martin W. Bauer. 1997. Sci-entific institutions in a society of transition: Strategies of modernisation. Project report to theSoros Foundation, Sofia.

Price, Derek de Solla. 1972. Money and influence: The link of science to public policy.Daedalus103:97-113.

Price, Derek de Solla, and D. Beaver. 1966. Collaboration in an invisible college.American Psy-chologist21:1011-18.

Ravetz, Jerry. 1971.Scientific knowledge and its social problems. Oxford, UK: Clarendon.Raza, G., S. Singh, B. Dutt, and J. Chander. 1996.Confluence of science and people’s knowledge

at the sangam. New Delhi: NISTED.Schuman, H., and S. Presser. 1981.Questions and answers in attitude surveys. New York: Aca-

demic Press.Withey, S. B. 1959. Public opinion about science and scientists.Public Opinion Quarterly

23:382-88.Wolpert, Lewis. 1992.The unnatural nature of science. London: Faber.Zhang, Z., and J. Zhang. 1993. A survey of public scientific literacy in China.Public Under-

standing of Science2:21-38.Ziman, John. 1995.Of one mind: The collectivisation of science. Washington, D.C.: American

Institute of Physics Press.

Martin W. Bauer (Ph.D.) is a Research Fellow at the Science Museum London (since1991) and Associate Professor in Social Psychology and Research Methodology at theLondon School of Economics (since 1995). Research includes (a) the public understand-ing of science as indicated by public opinion surveys, focus groups, and media coverage;and (b)the functions of public resistance, particularly against biotechnology. BooksincludeResistance to New Technology—Nuclear Power, Information Technology, Bio-technology(Cambridge University Press, 1995),Biotechnology in the Public Sphere(Science Museum, London, 1998), andQualitative Researching with Text, Image andSound(Sage, in press). His research papers have appeared inNature, Science, PublicUnderstanding of Science, Social Science Information, Journal for the Theory of SocialBehavior, Systems Practice, Cosmos and Culture, International Journal of Public Opin-ion Research, andAlliage.

Kristina Petkova (Ph.D.) has been an associate professor since 1989 at the BulgarianAcademy of Sciences, Institute of Sociology and was awarded the Fulbright Scholarship(1991-1992). She is a member of the Society for Social Studies of Science and the Asso-ciation of Experimental Social Psychology. She has been published inEuropean Journalof Social Psychology, Public Understanding of Science, andPolitical Psychology. Shehas also published three books in Bulgarian.


Pepka Boyadjieva graduated philosophy with a DSc in sociology. Pepka is a professor atthe Bulgarian Academy of Sciences, Institute of Sociology. Pepka has been published inPublic Understanding of Scienceand has four books published in Bulgarian in additionto being awarded an Andrew Mellon Fellowship 1999 (currently in Edinburg).


Public Knowledge of and Attitudes to Science: Alternative Measures That May End the "Science War

Documents