The Political Opinions of Swedish Social Scientists*

Finnish Economic Papers – Volume 22 – Number 2 – Autumn 2009

THE POLITICAL OPINIONS OF SWEDISH SOCIAL SCIENTISTS*

Niclas BerggreN

The Ratio Institute,P.O. Box 3203, SE‑103 64 Stockholm, Sweden; e‑mail: [email protected]

HeNrik JordaHl

Research Institute of Industrial Economics (IFN), P. O. Box 55665, SE‑102 15 Stockholm, Sweden; e‑mail: [email protected]

and cHarlotta sterN

Department of Sociology, Stockholm University, SE‑106 91 Stockholm Sweden; e‑mail: [email protected]

We study the political opinions of Swedish social scientists in seven disciplines and find indications of a left‑right divide, with sociology and gender studies being the most left‑leaning disciplines, with business administration, economics, and law being the most right‑leaning ones, and with political science and economic history being located somewhere in between. This pattern is found when looking at party preferences, left‑right self‑identification, and positions on economic pol‑ icy issues. Overall, there is a slight dominance in sympathies for the right, al‑ though there are more academics to the left among those most involved in activi‑ ties with a potential to influence decision‑makers. (JEL: A11; A13; A14)

*The authorswish to thank Andreas Bergh, Anders University,the Swedish Institute for Social Research (SOFI) Björklund, Stefan Dahlberg, Gissur Erlingsson, Jeffrey at Stockholm University, the Department of Economics at Friedman, Nils Karlson, Daniel Klein, Iain McLean, Henrik the University of Helsinki, the Research Institute of Indus‑ Oscarsson, Bo Rothstein, Daniel Waldenström, two anony‑ trial Economics (IFN), and the Ratio Institute for valuable mous referees, and participants at the World Meeting of the comments and suggestions, and the Torsten and Ragnar Public Choice Societies in Amsterdam, as well as seminar SöderbergFoundations (Berggrenand Jordahl) and the participants at the Department of Government at Uppsala SwedishCouncil for Working Life and Social Research University, the Department of Political Science at Göteborg (Stern) for financial assistance.

75 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern 1. Introduction advisors) and also in indirect ways (as teachers and in the media).2 an increasing number of studies investigate the early studies of academics indicated aleft political views of academics. Most of them have and democratic (as opposed to republican) in- been conducted in the Us, where the results in- clination among Us academics (spaulding and dicate that social scientists predominantly sup- turner 1968, lipset and ladd 1972, ladd and port the democratic Party,although the degree lipset 1975, lipset 1982, Hamilton and Har- of dominance varies substantially between dis- gens 1993). this was especially the case for ciplines. We contribute to this literature with an humanist and social-science faculty, whereas analysis of swedish social scientists in business natural-science and business faculty tended to administration, economic history, economics, be less left-oriented. later studies give asimilar gender studies, law, political science, and soci- picture. rothman et al. (2005) find that liberals ology.toour knowledge, this is the first study and democrats outnumber conservativesand of its kind in a Nordic country. republicans by large margins, and the differ- We find evidence of a left-right divide be- ences are not limited to elite universities or to tween disciplines, with gender studies and soci- the social sciences and humanities. klein and ology most to the left and with economics and stern (2004, 2005a,b, 2006a,b,c, 2007a) find a business administration most to the right. We dominance of democrats among social scien- look at party sympathies, left-right self-identi- tists and, based on asurvey of members of the fication, and attitudes towards policyproposals scholarly associations in anthropology, econom- on economic issues and find the same left-right ics, history, legal and political philosophy, po- divide in each case. the differences between litical science, and sociology, that their policy disciplines remain also when controlling for preferences generally are such that partisan various background characteristics. the most identification are in line with left/right ideal evident policy consensus is that a large major- types. theyalso find that the diversity of policy ity of the social scientists oppose tariffs on views is greatest in economics and that those goods fromcountriesoutside the eU with an who deviate from left views are as likely to fit overwhelming resistance among economists. a libertarian profile as a conservative one.3 to overall, there is a slight dominance of sympa- ourknowledge,there is no longitudinal study of thies for the right.1 Not all academics lean left the political views of academics, but klein and relative to the political culture to which they stern (2009, p. 15) summarize existing cross- belong, not even in a country like sweden. discipline studies from different time periods in signs of a left orientation emerge, however, the United states. their main finding is that when looking at the potentially most influential “few professors in the social sciences and hu- academics in the public-policyarea. manities today are not on the left, and that there studying sweden contributes to the literature has been adecline since the 1960s in professors in providing information on the policyviews of social scientists in aNordic country,apart of 2 See e.g. Myrdal (1960), Carlson and Jonung (2006), the world with extensive welfare-state arrange- and Bergh and Erlingsson (2009). ments and a socially progressive tradition. the 3 Similar results are reported by Lee (1994), the Brook‑ swedish tradition of running politics according ings Institution (2001), and Tobin and Weinberg(2006). The results on party preferences are reinforced and to some ex‑ to rationalist and modernist precepts has given tent validated by voter‑registration studies – see e.g. Horow‑ ample room for social scientists to affect poli- itzand Lehrer (2002), Zinsmeister (2002), Cardiffand Klein cies in many direct ways (e.g. as politicians and (2005), and Klein and Western (2005). Cardiff and Klein e.g. find an average Democrat‑Republican ratio of 5:1 across universities and disciplines. However, Zipp and Fen‑ wick (2006, p. 320) dispute the claims of the previous lit‑ eratureand present results that indicate that “the American academy has not become a liberal hegemony; if anything, 1 The terms left and right areused in accordance with therehas been aslight trend to moderation –fromboth ends standard usage in Swedish discourse, which is not auto‑ of the political spectrum towardthe center.” Forareply,see matically comparable to usage in other political cultures. Klein and Stern (2007b).

76 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern who are not on the left, as indicated by repub- ements of ahealthyacademic environment. a lican voting, self-identified conservative lean- dominance of a certain perspective in a disci- ings, or policy views.” pline could be detrimental for an open inquiry arelated line of research, initiated by Brittan into issues, especially into those of policy rel- (1973) and kearl et al. (1979), is less focused evance. However, homogeneity could also be on studying party sympathies or self-character- interpreted as consensus, i.e. as an indication izations butrather highlights views on specific that scholars have tried various ideas in open policy issues, and some investigate the extent to discourse and come, through careful analysis, which academics (especially economists) are in to certain insights of high certainty. agreement on various issues. For instance, Full- We do not evaluate these alternative interpre- er and geide-stevenson (2003) and Whaples tations in this study but leave for others to ana- (2006) report a considerable amount of consen- lyze and discuss the positive and negative ef- sus among economists over various policy is- fects of our findings. our study can best be seen sues. there have also been such surveys of as descriptive and explorative and as a founda- economists in countries other than the Us. gen- tion for further, more explanatory work. to date, erally, the studies report a high degree of con- there is no unified general theory of political sensus within countries, butthere are, as shown preferences. Frequent explanations include so- by e.g. Frey et al. (1984), sometimes clear dif- cial background (campbell et al. 1960), group ferences between economists in different coun- membership (Mutz and Mondak 1997), as well tries.4 as macro- and microeconomic models (Fair there are twodominant rationales for the line 1978; elinder et al. 2008), and it would be in- of inquiry pursued in this paper. First, academ- teresting for future research to develop and test ics tend to be influential in any society. in par- theories in this area. ticular, the views of teachers and professors are We now turn to the swedish case. transmitted to students.5 other channels of influence open to and used by academics, as members of the elite with access to political 2. The Survey circles, are publications, the media, and arole as advisor (Hoffman-lange 2007, p. 918). only 2.1 Invitees and Respondents by knowing what academics actually believe and think is it possible to critically discuss their the basis for this study is aweb-based survey influence. second, heterogeneity, plurality and that wasconducted in december 2005–January non-conformism are often viewed as central el- 2006. e-mail invitations were sent to 4,301 swedish academics, which constitute “all” aca- 4 FormoreUSstudies, see e.g. Alston et al. (1992), demics who are doctoral students and univer- Blendon et al. (1997), Fuchs et al. (1998), Heckelman and sity employees doing research and/or teaching Whaples (2003), Colander (2005), and Whaples (2006). For at 25 colleges and universities in the disciplines non‑US studies, see e.g. Blockand Walker (1988), Ricketts and Shoesmith (1992), Anderson and Blandy (1992), Ander‑ business administration, economic history, eco- son et al. (1993), Nakhaie and Adam (2008), and De Ben‑ nomics, gender studies, law, political science, edictis and Di Maio (2009). and sociology.6 the choice of colleges and uni- 5 Frank et al. (1993) and Frank and Schulze (2000), versities was motivated by there being a pres- among others, claim that university teachers influence eco‑ nomics students by making them more selfish. This claim ence of at least one department that clearly rep- has been disputed by e.g. Frey and Meier (2005). Even if the latter arecorrect, it may be that other forms of influences prevail. Guimond (1997) finds that the main subject studied by students is a significant predictor of change in sociopo‑ 6 Note that we will use tests of statistical inference even litical attitudes for college students. There are also indica‑ though we invited the entirepopulation −not arandom tions that civic education under certain conditions affects sample −ofacademics in the seven disciplines. These tests teenagers towards greater acceptance of democracy and the could still be used for inference if all non‑responses were market economy (Slomczynski and Shabad 1998). Cf.Ladd randomly determined. Alternatively, they can be used to and Lipset (1975, pp. 306–311), Banks and Roker (1994) extrapolate to alargeruniverse of possible populations and Blyth (2002). (Blalock 1979).

77 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern resented at least one of the seven disciplines.7 and Uppsala University (533). Further details for as for the choice of disciplines, we restrict our- each institution in the study are available in ap- selves to the social sciences and do not include pendix B in the online supplement. e.g. the natural sciences or the humanities. the reason is that we are interested in disciplines 2.2 The Survey that are relevant to policy-making, in the sense When invitees visited theweb page with the that theydeal with socially relevant matters survey (www.policystudy.se, no longer opera- which frequently form the basis for political tional) they first received information and in- opinions and decisions. on the basis of this cri- structions. everything wasinswedish, so here terion of policyrelevance, we also exclude some we translate as necessary.after aspecification social sciences, most notably cultural geogra- of the purpose of the study,there followed a phy, psychology,and social anthropology.When specification of the length of time it would take we speak of social scientists, we refer to the tocomplete it (about 15 minutes), aguarantee seven disciplines included in this study. of anonymity,and information about the three the e-mail addresses were located from the researchers. there were 73 questions in all. web pages of the respective colleges and univer- Here, we utilize only a subset, primarily relat- sities, and only the academics with listed e-mail ing to politics and policy views. the questions addresses were contacted. tworeminding e-mails are specified belowwhen the results are pre- were sent out in line with schaefer and dillman sented.10 (1998), who find this to be a good way to in- crease the response rate. the number of replies was 1,512, which gives a total response rate of 3. Results and Analysis 35.2 percent.8 adetailed non-response analysis suggests that our respondents are representative 3.1 Preliminaries of the population on important indicators.9 the largest numbers of invitations were sent to stock- Before presenting the results, we wish to clarify holm University (625), lund University (535) a few matters. First, there are seven parties in the Riksdag (the swedish Parliament): the christian demo- 7 Thereare,intotal, 39 collegesand universities in crats, the Moderate Party,the center Party,and Sweden. 8 This is somewhat similar to response rates in preced‑ the liberal Party form a right bloc (in govern- ing studies (which, however,usually begin with arandom ment from 2006, but making up the parliamen- sample of invitees, whereas we invited “all” academics in tary opposition when the survey was conduct- the chosen disciplines in Sweden). Forexample,Fuller and ed), whilethe socialdemocraticParty, theleft Geide‑Stevenson (2003) had a response rate of 30.8 per‑ cent; Heckelman and Whaples (2003) got response rates of Party, and the green Party form a left bloc (with 34.4 percent, 33.2 percent and 29.6 percent; and Klein and the social democratic Party forming aminority Stern (2004, 2005a,b, 2006a,b,c,d, 2007a) report a response government with the support of the left Party rate of 30.9 percent. Generally,web surveys seem to have lower response rates than mail surveys –around 20 percent and the green party at the time when the survey compared to 40 percent (Sheehan and McMillan 1999). wasconducted, butmaking up the opposition 9 We find few differences between respondents and non‑ from 2006).11 at the time when the survey was respondents on variables with known values for the popula‑ conducted, two challenging parties gained par- tion: university or college affiliation, gender,academic position, and academic discipline. When it comes to vari‑ ticular attention: the June list and the Feminist ables with unknown values for the population we apply the initiative. Neither reached the 4 percent thresh- “willingness‑to‑respond” approach and investigate wheth‑ er early responses systematically differ from late responses that we received only after sending out one or two remind‑ 10 The full survey (in its original Swedish and in an Eng‑ ing emails (cf.Armstrong and Overton 1977). Differences lish translation), covering 14 pages, is available upon re‑ between early and late responders are almost completely quest. absent for the variables that we study (bloc preferences, 11 The Green Party has now (in 2009) formed an alliance policy views, self‑reported left‑right scale, and certain in‑ with the Social Democratic Party and the Left Party to the fluential activities). The details can be found in Appendix A effect that they will all be part of the government if they win in the online supplement. the 2010 election.

78 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern old in the september 2006 general election and revealed in table 1. it is based on the following hence do not hold seats in Parliament. For fur- question: ther details about the parties, see Box B1 in the online supplement. even though one doesn’t agree with a politi- second, we make frequent use of the left- cal party on each and every issue, one may right terminology,both when discussing party have stronger sympathies for acertain party, sympathies, policyopinions and (naturally) compared to others. Which party do you self-identification on the left-right scale. this have the strongest sympathies for? conforms to standard usage among political scientists and seems especially apt in aswedish there is a clear left dominance in sociology setting, where the scale is widely understood – with about five times as many who sympa- and used by voters to structure their political thize with the left as with the right bloc – as preferences.12 of course, left-right terminology well as in gender studies. there is a milder left entails simplification (see the discussion by character to political science and economic his- Mair 2007), and it should be pointed out that tory.there are three disciplines with aright there are differences between the parties within dominance: economics, law, and business ad- each bloc and that some are quite centrist on the ministration. in the former two, there are about scale. traditionally,perhaps mostly the liberal twice as manywho sympathize with the right as Party in the right bloc and the green Party in with the left bloc. in business administration, the left bloc have been oriented towards the the sympathy ratio rises to about three. the re- center.But in recent years, the bloc formations sult for all disciplines is included for compari- have been quite solid. the decision to include son. the right bloc attracts 1.3 times as many of the green Party in the left bloc has two bases: our academics as the left.14 We will return to it has supported, together with the left Party, these aggregate figures in more detail below. social democratic governments since its entry into Parliament; and it is placed to the left by table 1. Bloc preferences of swedish social scientists (per- voters on the left-right scale (grendstad 2003). cent). also, the terms left and right are to some extent understood differently in other political cul- left right ratio right bloc/ bloc bloc left bloc tures.13 left in sweden and left in the Us may mean quite different things – and for this rea- Business administration 19.8 60.8 3.1 son, comparisons of the shares of left- and economics 26.8 52.5 2.0 right-sympathizers are generally most meaning- law24.6 46.6 1.9 ful when made within a political culture. lastly, economic history 40.6 31.2 0.8 not all issues are easily described by the left- Political science 45.1 29.4 0.7 right scale. still, for our purposes the left-right gender studies 23.1 10.2 0.4 terminology is useful in providing approximate sociology 57.4 11.0 0.2 measures of ideological leanings. all 33.2 41.7 1.3 third, when results are reported for the differ- Notes: the sum of the share that supports the right bloc and ent disciplines, this everywhere refers to the dis- the share that supports the left bloc is not 100 percent since ciplines where respondents are currently active. thereare five reply alternativesthat do not belong to anyof these blocs: “other”, “None”, “do not know/do not want to answer”, the June list (Jl) and the Feminist initiative 3.2 Party Sympathies (Fi), twoparties that are not part of anyofthe blocs. the left bloc comprises the social democratic Party (s), the there are large differences between the disci- left Party (V), and the green Party (Mp). the right bloc comprises the christian democrats (kd), the Moderate plines when it comes to party sympathies, as Party (M), the center Party (c), and the liberal Party (Fp).

12 See Klingemann (1995), Bobbio (1996), Oscarsson (1998) and Grendstad (2003). 14 If eachdiscipline is given equal weight irrespective of 13 See e.g. Grendstad (2003) and Zechmeister (2006). its size the ratio is 0.93 (a mild left dominance).

79 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern one can relate these findings to the ratio be- tion, economics, and lawisassociated with a tween republicans and democrats in the Us in higher probability for supporting the right bloc. different disciplines. according to klein and these findings are robust with regard to the set stern (2005b), among sociologists and anthro- of control variables – from including age, gen- pologists, the ratio is 0.05 – that is, there are der and academic positions to including, in ad- more than 20 times as many democrats as re- dition, family income, and academic affilia- publicans. among political scientists, it is 0.2 tion. (more than five times as manydemocrats); and the biggest party among all our respondents among economists, it is 0.3 (three times as is the liberal Party (Fp), acenter-right party in many democrats). For social-science and hu- the right bloc. if one looks at the two political manities faculty in all theyestimate the ratio to blocs, the right bloc gets 41.7 percent and the be 0.12 (eight times as manydemocrats). there left bloc gets 33.2 percent, aclear dominance definitely seems to be more of aone-sided dom- (of 1.3:1) of the right. inance in the Us setting. the party-sympathy figures for the swedish looking specifically at sympathies for the social scientists participating in our survey can different parties in the various disciplines, as be compared to those of swedes in general in well as for all 1,512 respondents, our study re- december 2005–January 2006, as reported by veals that the liberal Party (Fp) is the largest three opinion-poll institutes. there are great dif- party in business administration, economic his- ferences between the social scientists surveyed tory,economics, and law. this is noteworthy, here and citizens in general. the social demo- since this party is relatively small in the elector- cratic Party is very small among our social sci- ate at large, where the more conservative Mod- entists compared to citizens in general, but also erate Party (M) is since long the largest party in the Moderate Party is substantially smaller.the the right bloc. this finding indicates that the party that is disproportionately large among our sympathies for the right bloc are relatively cen- social scientists is the liberal Party.For each trist among the academics in our study. the so- left academic in our study, there are 1.3 right cial democratic Party (s)isthe largest among academics, whereas the corresponding figure political scientists –the only discipline for for citizens in general is 1.1.15 which this holds, in spite of this party being the biggest in the electorate. the left Party (V) is the largest among sociologists (rendering the 3.3 Self‑Reported Left‑Right Scale left character of the discipline more distinct than if the social democrats had been the larg- We also asked the following question, with re- est party); and the Feminist initiative (Fi) great- ply alternativesfrom 1to10and with the option ly dominates gender studies. Fordetails, see “do not know/do not want to answer”: table c1 in appendix cin the online supplement. to see whether the differences in party sympathies observed across disciplines are due to 15 Astudy by Asp (2006) indicates that among Swedish compositional differences, we have regressed journalists (who, like academics, in many cases may be seen bloc sympathyondiscipline, age, gender,and as potentially influential intellectuals), the share of support for the left bloc among those who support one of the seven position in the academy. the logit results, and main parties is 67 percent, whereas the corresponding share a detailed motivation of the control variables, of support for the right bloc is 33 percent (in late 2005 and are presented in appendix din the online sup- based on the question “Which party do you like best to‑ plement. the results reveal that the bloc differ- day?”). This is a major difference compared to the academ‑ ics in our study, keeping in mind the slightly different party‑ ences between the disciplines are in qualitative sympathy question. With Asp’sway of counting,the share agreement with the findings presented above. of support for the left bloc among those of our social scien‑ Being a scholar in sociology is associated with tists who support one of the seven main parties is 44 per‑ cent, whereas the corresponding shareofsupport for the ahigher probability for supporting the left bloc, right bloc is 56 percent. Among journalists, the right/left whereas being a scholar in business administra- ratio is 0.5, and among our social scientists, it is 1.3.

80 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern Where on ascale from 1to10, where 1 3.4 Policy Views equals the most left-wing member of the current parliament and 10 equals the most right- in order to get amore granular picture of the wingmember of the current parliament, political views of our swedish social scientists, would you place yourself? we asked them about their views on anumber of proposed policyreforms. We focusonpolicy the differences between disciplines found for proposals on economic issues. the questions party sympathies are confirmed by distinct dif- are supposed to capture “important” policy is- ferences between the disciplines when it comes sues that have been publicly debated in sweden to self-identification on the left-right scale. the and for which no apparent consensus exists. averages are presented in table 2. again, three We had seven economic questions in the sur- groups appear.sociology and gender studies vey: have aleft orientation, economic history and political science are quite centrist (with a mild 1. What is your opinion of aproposal to left orientation), whereas law, economics, and raise or introduce tariffs on goods from business administration have a right orientation. non-eU countries in order to protect in terms of standard deviations, law and eco- swedish jobs? nomic history are the most heterogeneous 2. What is your opinion of aproposal to whereas gender studies and sociology are the abolish the right of labor unions to insti- most homogeneous.16 gate boycotts against companies that don’twant to sign collective bargaining agreements? table 2. left-right positioning in sevendisciplines (means 3. What is your opinion of aproposal to cut with standard deviations in parentheses). taxes? Discpline Left‑right position 4. What is your opinion of aproposal to downsize the public sector? gender studies 3.19** (1.72) 5. What is your opinion of a proposal to re- sociology 3.46** duce income differences in society? (1.79) 6. What is your opinion of aproposal to Economic history 4.75 have more medical care/hospitals run by (2.39) private companies? Political science 4.82 7. What is your opinion of aproposal to cut (2.09) down on social welfare benefits? law5.83** (2.44) economics 6.15** each question had the following reply alterna- (1.95) tives: Business administration 6.30** (2.01) Very bad proposal (1) rather bad proposal (2) all 5.35 (2.31) Neither a good nor a bad proposal (3) rather good proposal (4) ** * Notes: 0.01, 0.05 p-value using the Mann-Whitneytest Very good proposal (5) of equality of distribution across twoindependent samples using economic history as the comparison group, which is do not know/do not want to answer. why it is marked in bold. (the numbers 1–5 did not appear.) the responses from the different disciplines are shown in table 3. Here, we exclude the “do 16 When looking at the distribution of all responses, the not know/do not want to answer” replies to en- average is 5.4, and the median is 5. However,polarization can be seen here as well. For details, see Figure C1 in Ap‑ able amean and standard-deviation comparison. pendix C in the online supplement. since the underlying answers are ordinal, the

81 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern table 3. Views on economic policy per discipline (means with standard deviations in parentheses).

Business economic Economics gender lawPolitical sociology all administration history studies science 1.56** 1.49** 1.27 1.66** 1.66** 1.52** 1.77** 1.55 raise tariffs (0.73) (0.59) (0.55) (0.76) (0.79) (0.61) (0.80) (0.72) abolish union 3.03 2.46** 3.11 1.69** 2.69** 2.44** 2.09** 2.69 boycotts (1.41) (1.42) (1.39) (1.01) (1.46) (1.26) (1.29) (1.42) Undertaketax 3.74 2.98** 3.66 2.11** 3.53 2.82** 2.21** 3.23 cuts (1.13) (1.23) (1.03) (1.03) (1.25) (1.26) (1.10) (1.29) downsize 3.40 2.66** 3.37 1.74** 3.07* 2.57** 1.86** 2.89 public sector (1.22) (1.31) (1.07) (0.83) (1.27) (1.20) (0.96) (1.29) reduce income 3.06 3.69** 3.08 4.39** 3.34* 3.78** 4.27** 3.48 differences (1.15) (1.21) (1.15) (0.79) (1.19) (1.13) (0.96) (1.21) More private 3.66 2.95** 3.62 2.26** 3.40 3.03** 2.37** 3.25 healthcare (1.22) (1.47) (1.11) (0.99) (1.29) (1.24) (1.26) (1.32) cut welfare 3.08 2.33** 2.95 1.67** 2.75 2.33** 1.83** 2.61 benefits (1.12) (1.27) (1.08) (0.79) (1.31) (1.11) (0.99) (2.21) Notes: small numbers indicate opposition and large numbers indicate support. ** 0.01, * 0.05 p-value using the Mann- Whitney test of equality of distribution across two independent samples using economics as the comparison group, which is whyitismarked in bold. an ols regression with adummy-variable using economics as the reference category yields no differences in results. means have no exact interpretation and should the public sector than other social scientists. only be seen as a pedagogical device that saves theyare also more mixed than the others on space and enables quick comparisons of the dis- reducing income inequality. scholars in busi- ciplines. ness administration, economics, and lawfavor the value 3 (the reply “Neither a good nor a more private health care, whereas political sci- bad proposal”) can be interpreted as a neutral or entists and economic historians are neutral. mixed viewofthe policyproposals, whereas 1 scholars in sociology and gender studies op- and 2indicate various degrees of opposition and pose the proposition. With one exception (rais- 4and 5various degrees of support. overall it ingtariffs) gender studies is most to the left and seems fair to conclude that swedish social sci- has the smallest standard deviation. entists are not supportive of policy reforms the impression from the analysis of party aimed at raising or introducing protective tariffs sympathies recurs: Business administration, on goods from non-eU countries: all disciplines economics, and law lean more to the right, have mean responses well below 3, with econo- whereas economic history and political science mists being most strongly opposed. Nor do they are moderately on the left, and gender studies support abolishing union rights to boycott com- and sociology are more distinctly on the left.17 panies who do not want to sign collective bar- But in no case is there a very strong desire to gaining agreements – weak resistance domi- deviate from the status quo (except, perhaps, nates, except in business administration and when it comes to reducing income inequality, economics, where neutrality prevails. cutting which is strongly supported in gender studies welfare benefits is not really supported in any and in sociology). discipline. academics in business administration and economics are neutral, while academics in gender studies and sociology are quite 17 When we use the terms left and right in this section, strongly opposed to the proposal. on the other we rely on the most common positions of the two blocs. For proposals the pattern is more mixed. academics example,cutting taxes is considered aright position since the parties of the right bloc advocate this to a larger degree, in business administration and economics are while reducing income differences is considered a left posi‑ more supportive of tax cuts and of downsizing tion for an analogous reason.

82 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern

Figure 1. Views on economic policy (percent) of all respondents. Notes: 1: Very bad proposal; 2: rather bad proposal; 3: Neither a good nor a bad proposal; 4: rather good proposal; 5: Very good proposal.

the responses from all respondents are shown 3.5 Inﬂuential Activities in Figure 1. only on the first question is there an evident as has been argued above, a motivation for a consensus. alarge majority thinks that intro- study of this kind is the possible influence so- ducing or raising tariffs on goods from coun- cial scientists may exert over public discourse tries outside the eU in order to protect swedish and politics. to see the extent to which our re- jobs is arather or very bad proposal. this is spondents have had such opportunities, we also the issue where economists stand out the asked the following questions: most, with an overwhelming resistance to tariffs (see Figure 2). on the other questions, there is Haveyou participated in anysoU (swedish no similar consensus among the respondents. government official report) or any other gov- there is no general support for radical policy ernment commission? change in either aleft or aright direction. Howmanyarticles, chapters or books that lastly,itshould be noted that differences in can be classified as student literature or pop- replies mayreflect both different preferences ular science publications (publication that and different knowledge, butweare unable to inform the public about research within your differentiate between them and identify causal field) have you written? effects with our data.

83 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern

Figure 2. Views on economic policy (percent) of economists. Notes: 1: Very bad proposal; 2: rather bad proposal; 3: Neither a good nor a bad proposal; 4: rather good proposal; 5: Very good proposal.

Have you participated in anyradio or tV ernment commissions (which may have to do programs where you discussed or related to with its being a young discipline). Business ad- your research in a popular science way? ministration appears to be the least active subject in this regard. if we relate the mentioned activities to the Participation in government commissions is the ideological stance of the scholars, measured as most exclusive of these activities, 14 percent of their position on the left-right scale, we find no the respondents have participated once and 13 statistically significant differences between the percent have done so several times. radio and most active and the most passive group when it television are less exclusive,35percent have comes to participation in government commis- appeared at least once on these mass media. sions or to publishing teaching material and Publications are more common: 70 percent have popular science. Forradio and television the written at least one publication classified as mean left-right position is 4.98 for scholars who teaching material or popular science. there are have appeared at least three times compared to also differences between the subjects. table 4 5.63 for those who have not appeared asingle reveals that academics in political science are time.18 quite involved in all three activities. academics in gender studies also appear rather active, es- 18 The distributions in the two groups differ significantly pecially in radio and television but not in gov- according to the Mann‑Whitneytest.

84 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern table 4. influential activities (percent). activity levelof Business econo- Econo- gender lawPolitical socio- all partici- admini- mic mics studies science logy pation stration history government No 77.3** 79.7* 65.2 76.9 68.5 60.3 70.0 70.5 commissions low 11.1** 10.9* 14.1 12.8 16.4 19.1 15.2 14.0 High 8.2** 6.2* 18.5 10.3 13.0 19.6 12.7 13.0 teaching material No 29.0 23.4 38.0 12.8** 29.4 24.5* 25.3 28.4 and popular low47.2 54.7 38.4 46.2** 47.3 51.5* 52.3 47.2 science High 22.3 20.3 22.8 41.0** 23.3 22.1* 21.5 23.2 radio and No 64.6 59.4 60.5 35.9** 58.2 35.3** 44.3** 54.5 television low 23.2 26.6 21.4 25.6** 22.6 31.4** 30.4** 25.1 High 12.2 12.5 17.8 38.5** 19.2 32.8** 24.9** 20.0 Notes: Forgovernment commissions alow (high) levelofparticipation is defined as once (several times). Forteaching material and popular science alow (high)level of participationisdefined as between one and five articles and books (more than five articles and books). Forradio and television alow (high) levelofparticipation is defined as between one and three times (more than three times). the response category “do not know/donot want to answer” is not reported in the table. ** 0.01, * 0.05 p-value using the Mann-Whitneytest of equality of distribution across twoindependent samples using economics as the comparison group, which is why it is marked in bold.

there is however a connection between ac- universities with new and small university col- tive participation in government commissions leges. However, as reported in table d1 in the and sympathies for the social democratic Party, online appendix, the estimates of academic af- the party forming the government 1994−2006. filiation are not statistically significant. Further- the social democratic Party is the most sym- more, when replacing the affiliation variables in pathized-with party among the most active column (3) of table d1 by two dummy varia- scholars with sympathies being 13.3 percentage bles for academic quality, covering those uni- units higher than in the least active group. on versities that are on the 2006 academic rank- the one hand it seems as if the social demo- ing of World Universities (arWU) and on the cratic government has not engaged a represent- 2006 times Higher education World University ative sample of social scientists in government rankings list, we find that the estimated coef- commissions when it comes to sympathies. on ficients are, in neither case, statistically signifi- the other hand the overrepresentation of social cant. scientists who support the social democrats makes the composition of government commissions more in line with the party sympathies of 4. Concluding Remarks all swedish citizens.19 another waytostudy influence is to adjust academics tend to be influential in anysociety, for research quality by comparing old and large affecting decision-makers through lectures, various publications, the media, and in some cases directly as advisors. it is therefore of in- 19 Thereare of course many explanations for the degree terest to study the political persuasions of aca- of influenceactivities in the different disciplines, as well as forthe ideological character of those who undertakethem. demics. although such persuasions do not nec- Some explanations are demand‑ and others are supply‑in‑ essarily color their work, Myrdal (1969) and duced. Of the three types of activity, the most demand‑in‑ others have suggested that ideological sensi- duced is probably the first. Writing agovernment report or being on a government commission is the result of being bilities run deeper than many scholars realize or picked. may be willing to admit.

85 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern We find that swedish social scientists in sev- in the Us) is generally much larger than for the en disciplines can generally be classified as right bloc in sweden, and the differences be- leaning slightly right on the whole. it turns out tween the disciplines (in terms of how dominant that 51.5 percent support the right bloc com- the largest party or bloc is) are larger in the Us. pared to 41.0 percent for the left bloc (a right- the most homogeneous discipline appears to be left ratio of 1.3), to be compared with 1.1 in the gender studies, with the lowest standard devia- electorate in general. Behind these aggregate tions on most policy issues. shares, we find aleft-right divide. the most in summing up our results, we do not claim left-leaning discipline is sociology, where the that our findings can be used for causal infer- left bloc achieves more than 5 times more sup- ence. We report differences between the disci- port than the right bloc and where the left Par- plines and leave the question of why these dif- ty is also the biggest party.the second most ferences exist for future research. Nor do our left-leaning discipline is gender studies. the findings necessarily indicate aproblem. We do most right-leaning discipline is business admin- not say that a left dominance in sociology or a istration, where the right bloc achievesmore right dominance in business administration en- than three times more support than the left bloc tail discrimination of colleagues in the minority and where the liberal Party is the most pre- position. our study does not address such is- ferred. economics and laware the second- and sues. However, we hopethatthe results can third-most right-leaning disciplines. When it constitute an informed basis on which such is- comes to policyissues, the general picture – sues can be addressed. confirming the one stemming from party sympathies – is that scholars in business administration, economics, and lawtend to favorreforms References of the right whereas especially scholars in gender studies and sociology tend to favor reforms Alston, R. M., J. R. Kearl, and M.B. Vaughan (1992). “is there aconsensus among economists in the 1990s?” of the left – but the desire to change the status American Economic Review 82, 203–209. quo is weak overall. Anderson, M., and R. Blandy (1992). “What australian Unlike Us academics, who have been shown economics Professors think.” Australian Economic Re‑ view 25, 17–40. to tend towards the left in their political culture, Anderson, M., R. Blandy, and S. Carne (1993). “aca- swedish social scientists seem to tend more to demic economic opinion in east asia.” Australian Eco‑ the right overall. Here we have to keep in mind nomic Review 26, 5–19. Armstrong, J. S., and T. S. Overton (1977). “estimating that an american academic to the left may,in Nonresponse Bias in Mail surveys.” Journal of Market‑ actual views, be close to a swedish academic to ing Research 14, 396−402. the right. Asp, K. (2006). Journalistkårens partisympatier [the Party sympathies of Journalists],Working Paper No. 38, de- in a multiparty system there are also impor- partment of Journalism and Mass communication, göte- tant nuances within the two political camps. borg University. among our academics the liberal party is the Banks, M. H., and D. Roker (1994). “the Political so- cialization of Youth: exploring the influence of school preferred right-bloc party,incontrast to the experience.” Journal of Adolescence 17, 3–15. Moderate Party that is the preferred right-bloc Bergh, A., and G. Erlingsson (2009). “liberalization with- out retrenchment: Understanding the consensus on party in the electorate (and considered more swedish Welfare state reforms.” Scandinavian Political “right” than the liberals). likewise, we see that Studies 32, 71–93. the left Party is the preferred left party among Blalock, H. (1979). Social Statistics. New York: Macgraw- Hill. our academics, in contrast to the social demo- Blendon, R. J.,J.M.Benson, M. Brodie, R. Morin, D. E. crats, the preferred party in the electorate (and Altman, G. Gitterman, M. Brossard, and M. James considered more “right” than the left Party). (1997). ”Bridging the gap Between the Public’sand economists’ Views of the economy.” Journal of Eco‑ When looking at heterogeneity, there seems nomic Perspectives 11, 105–118. to be less dominance within the disciplines and Block, W., and M. A. Walker (1988). “entropy in the ca- smaller differences between the disciplines nadian economics Profession: sampling consensus on the Major issues.” Canadian Public Policy 14, 137– comparedwith the Us.the dominanceof dem- 150. ocrats (which seems to hold in every discipline Blyth, M. (2002). Great Transformations: Economic Ideas

86 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern

and Institutional Change in the Twentieth Century. cam- administrations and Faculties of 32 elite colleges and bridge: cambridge University Press. Universities. report, center for the study of Popular Bobbio, N. (1996). Left & Right: The Signiﬁcance of a Po‑ culture, sherman oaks, ca. litical Distinction. cambridge: Polity Press. Kearl, J. R., C. L. Pope, G. C. Whiting, and L. T. Wim- Brittan, S. (1973). Is There an Economic Consensus? lon- mer (1979). “a confusion of economists?” American don: Macmillan. Economic Review 69, 28–37. Brookings Institution (2001). National Survey on Govern‑ Klein, D. B., and C. Stern (2004). “democrats and repub- ment Endeavors. Washington, dc: the Brookings insti- licans in anthropology and sociology: Howdothey tutions. differ on Public Policy issues?.” The American Sociolo‑ Campbell, A., P. Converse, W. Miller,and D. Stokes gist 35, 79–86. (1960). The American Voter. New York: Wiley. Klein, D. B., and C. Stern (2005a). “Political diversity in Cardiff, C., and D. B. Klein (2005). “Faculty Partisan af- six disciplines.” Academic Questions 18, 40–52. filiation in all disciplines: aVoter registration study.” Klein, D. B., and C. Stern(2005b). “Professors and their Critical Review 17, 237–255. Politics: the PolicyViews of social scientists.” Critical Carlson, B., and L. Jonung (2006). “knut Wicksell, Review 17, 257–303. gustavcassel, eli Heckscher,Bertil ohlin and gunnar Klein, D. B., and C. Stern (2006a). “economists’ Policy Myrdal on the role of the economist in Public debate.” Views and Voting.” Public Choice 126, 331–42. Econ Journal Watch 3, 511–550. Klein, D. B., and C. Stern (2006b). “sociology and clas- Colander, D. (2005). “the Making of an economist sical liberalism.” The Independent Review: AJournal of redux.” Journal of Economic Perspectives 19, 175– Political Economy 11, 37–52. 198. Klein, D. B., and C. Stern (2006c). “Political scientists’ De Benedictis, L., and M. Di Maio (2009). economists’ Policy Views and Voting.” The Political Science Review‑ View about the economy: evidence from asurvey of er 35, 416–425. italian economists. Working Paper,dieF, University of Klein, D. B., and C. Stern (2007a). “is there a Free-Mar- Macerata. ket economist in the House? the Policy Views of amer- Elinder, M., H. Jordahl, and P. Poutvaara (2008). selfish ican economics association Members.” American Jour‑ and Prospective:theory and evidence of Pocketbook nal of Economics and Sociology 66, 309–334. Voting. cesifo Working Paper 2489. cesifo. Klein, D. B., and C. Stern (2007b). “letter to the editor: Fair, R. C. (1978). “the effect of economic events on reply to Zipp and Fenwick.” Public Opinion Quarterly Votes for President.” ReviewofEconomics andStatistics 71, 479–481. 60, 159–173. Klein, D. B., and C. Stern (2009). “By the Numbers: the Frank, R. H., T. Gilovich, and D. T. Regan (1993). “does ideological Profile of Professors.” in The Politically studying economics inhibit cooperation?” Journal of CorrectUniversity –Problems, Scope,and Reforms, Economic Perspectives 7, 159–171. 15–37. eds. r. Maranto, r. e. redding and F. Hess. Frank, B., G. G. Schulze (2000). “does economics Make Washington, dc: aei Press. citizens corrupt?” Journal of Economic Behavior and Klein, D. B., and A. Western (2005). “Voter registration Organization 43, 101–113. of Berkeley and stanford Faculty.” Academic Questions Frey, B. S., and S. Meier (2005). “selfish and indoctri- 18, 53–65. nated economists?” European Journal of Law and Eco‑ Klingemann, H.–D. (1995). “Party Positions and Voter ori- nomics 19, 165–171. entations.” in Citizens and the State,183–205. eds. Frey,B.S., W. W. Pommerehne, F. Schneider,and G. H.-d. klingemann and d. Fuchs. oxford: oxford Uni- Gilbert (1984). “consensus and dissension among versity Press. economists: an empirical inquiry.” American Econom‑ Ladd, E. C., Jr., and S. M. Lipset (1975). The Divided ic Review 74, 986–993. Academy: Professors & Politics. New York: Mcgraw- Fuchs, V. R., A. B. Krueger, and J. M. Poterba (1998). Hill. “economists’Viewsabout Parameters, Values, and Poli- Lee, D. R. (1994). “go to Harvard and turn left: the rise cies: survey results in labor and Public economics.” of socialist ideology in Higher education.” in The Cul‑ Journal of Economic Literature 36, 1387–1425. tural Context of Economics and Politics,15–26. eds. t. Fuller, D. A., and D. Geide-Stevenson (2003). “consensus W. Boxx and g. M. Quinlivan. lanham, Ma: Univer- among economists: revisited.” Journal of Economic sity Press of america. Education 34, 369–387. Lipset, S. M. (1982). “theacademic Mind at the top: the Grendstad, G. (2003). “reconsidering Nordic Party space.” Political Behavior and Values of Faculty elites.” Public Scandinavian Political Studies 26, 193–217. Opinion Quarterly 46, 143–168. Guimond, S. (1997). “attitude change during college: Lipset, S. M., and E. C. Ladd, Jr. (1972). “the Politics of Normative or informational social influence?” Social american sociologists.” American Journal of Sociology Psychology of Education 2, 237–261. 78, 67–104. Hamilton, R. F., and L. L. Hargens (1993). “the Politics Mair, P. (2007). “left-right orientations.” in The Oxford of the Professors: self-identifications, 1969-1984.” So‑ Handbook of Political Behavior, 206–222. eds. r. dal- cial Forces 71, 603–627. ton and H.-d. klingemann. oxford: oxford University Heckelman, J. C., and R. Whaples (2003). “are Public Press. choice scholars different?.” PS: Political Science and Mutz, D. C., and J. J. Mondak (1997). “dimensions of Politics 36, 797–799. sociotropic Behavior: group-Based Judgments of Fair- Hoffman-Lange, U. (2007). “Methods of elite research.” ness and Well-Being.”American Journal of Political Sci‑ in The Oxford Handbook of Political Behavior, 910–927. ence 41, 284−308. eds. r. dalton and H.-d. klingemann. oxford: oxford Myrdal, G. (1960). Beyond the WelfareState: Economic University Press. Planning and Its International Implications. New Ha- Horowitz, D., and E. Lehrer (2002). Political Bias in the ven, ct: Yale University Press.

87 Finnish Economic Papers 2/2009 – Niclas Berggren, Henrik Jordahl and Charlotta Stern

Myrdal, G. (1969). Objectivity in Social Research.New Slomczynski, K. M., and G. Shabad (1998). “can support York: Pantheon Books. for democracyand the Market Be learned in school? a Nakhaie, M. R., and B. Adam (2008). “Political affiliation Natural experiment in Post-communist Poland.” Politi‑ of canadian Professors.” Canadian Journal of Sociology cal Psychology 19, 749–779. 33, 873–898. Spaulding, C. B., and H. A. Turner (1968). “Political ori- Oscarsson, H. (1998). Den svenska partirymden: väljarnas entation and Field of specialization among college uppfattningar av konfliktstrukturen i partisystemet 1956– Professors.” Sociology of Education 41, 247–262. 1996 [The Swedish Party Space: Voters’ Perceptions of Tobin, G. A., and A. K. Weinberg. (2006). AProfile of the Conflict Structure in the Party System 1956‑1996]. American College Faculty. Volume I: Political Beliefs & Ph.d. dissertation, department of Political science, Behavior. san Francisco: the institute for Jewish & göteborg University. community research. Ricketts, M., and E. Shoesmith (1992). “British econom- Whaples, R. (2006). “do economists agree on anything? ic opinion: Positive science or Normative Judgment?” Yes!” The Economists’Voice 3, article 1. American Economic Review 82, 210–215. Zechmeister, E. (2006). “What’sleft and Who’sright? a Rothman, S., S. R. Lichter, and N. Nevitte (2005). “Poli- Q-method study of individual and contextual influ- tics and Professional advancement among college Fac- ences on the Meaning of ideological labels.” Political ulty.” The Forum 3, article 2. Behavior 28, 151–173. Schaefer, D. R., and A. A. Dillman (1998). “development Zinsmeister, K. (2002). “the shame of america’s one- of a standard e-Mail Methodology: results of an ex- Party campuses.” The American Enterprise 13, 18–25. periment.” Public Opinion Quarterly 62, 378–397. Zipp, J. F., and R. Fenwick (2006). “isthe academy a Sheehan, K., and S. McMillan (1999). “response Varia- liberal Hegemony? the Political orientation and edu- tion in e-mail surveys: an exploration.” Journal of Ad‑ cational Values of Professors.” Public Opinion Quar‑ vertising Research 39, 45–54. terly 70, 304-326.