Abstract
Contemporary studies of the politics of American professors compare their political preferences to those of the American public. That some professors exhibit more liberal attitudes than the public leads critics to ask whether this difference biases teaching and enforces political correctness that stifles the study of controversial topics. To provide an alternative substantive and methodological paradigm for future studies of academia – one that focuses on social contexts, mechanisms, and outcomes – this chapter briefly reviews Lazarsfeld and Thielens’s The Academic Mind: Social Scientists in a Time of Crisis. They studied the effects of McCarthyism on academia, documenting how contextual and personal variables influenced the professors’ apprehension, i.e., worry and caution. This chapter then applies multilevel statistical modeling to their pivotal three-variable contextual tables, showing how contemporary statistical methods can advance their analysis by close inspection of tabular data. Multilevel models can incorporate simultaneously the effects of numerous contextual and individual variables, providing measures of effects and appropriate tests of significance for the clustered data.
To organize our material [on the causes of teachers’ apprehension] we followed the well-established idea that all human experiences are determined by two broad groups of elements: the characteristics of the people themselves and those of the environment in which they live and work.
Paul F. Lazarsfeld and Wagner Thielens, Jr.(1958, 159–160)
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
To calculate gamma (γ) first define A as the number of concordant pairs of observations (++ or −−) and B as the number of discordant pairs of observations (+- or -+). Then, gamma \({\textrm{ = (A}} - {\textrm{B)/(A + B)}}\). For Table 3-3 (1958, 81): \({\textrm{A}} = 1,184 \times 423 = 500,832\); \({\textrm{B}} = 125 \times 719 = 89,875\); \({\textrm{A}} - {\textrm{B}} = {\textrm{410,957}}\); \({\textrm{A}} + {\textrm{B}} = {\textrm{590,707}}\); \(\gamma = 410,957/590,707 = .70\).
- 2.
Rough estimates of the average effects of these variables can be calculated by simply taking a simple average of the two conditional relationships. For the data of their Fig. 3-7 the roughly estimated average effect of involvement in an incident on apprehension, controlling for membership in a controversial organization \( = ((75\% - 56\% ) + (71\% - 36\% ))/2 = (19\% + 35\%)/2 = 27\%\). The roughly estimated average effect on apprehension of membership in a controversial organization, controlling for involvement in an incident \( = ((75\% - 71\% ) + (56\% - 36\% ))/2 = (4\% + 20\% )/2 = 12\%\). The roughly estimated interaction effect on apprehension of not being involved in an incident but belonging to a controversial organization is the difference between these differences divided by 2. It equals \((20\% - 4\% )/2 = (35\% - 19\% )/2 = 8\%\). These rough estimates do not take into consideration the different sample sizes and the limitations of the linear probability model.
- 3.
The direction of the effect between permissiveness and apprehension has perplexed some scholars (personal communication). Lazarsfeld and Thielens assumed that permissiveness led to apprehension (Fig. 7-13). Implicitly, they may have conceptualized permissiveness–conservatism as roughly analogous to the cluster of variables indicative of authoritarianism. The conservative pole of permissiveness is somewhat analogous to political and economic conservatism whereas the openness-to-ideas aspect of permissiveness suggests stronger commitments to anti-authoritarian democratic values. If authoritarianism is a variable of personality, then, given this analogy, it is logical to assume that apprehension is in part a manifestation of permissiveness, rather than the opposite. This leaves open the possibility of mutual effects.
- 4.
Although binary response variables are best analyzed using a logistic regression model, the linear probability regression model is easy to implement and can be applied to explore the data when the dichotomized response variable has near equal proportions in categories 0 or 1 as is the case here. Moreover, some sociologists state that proportions are easier to interpret than the logits and odds ratios of logistic regression (Cole 2001, Davis 2001, Hellevick 2009); sociologists have developed a core body of methods based on the linear probability model (Lazarsfeld 1961, 1971, Coleman 1964, Davis 1975, 1980); statisticians have provided some corrections that ameliorate its shortcomings (Fleiss 1981, Goldberger 1991); and the linear probability model can be readily transformed into a logistic regression model (Coleman 1981), as will be done later in this chapter. Thus, in modeling these data I will first apply a linear probability model that assumes that the proportion apprehensive varies continuously from 0 to 1 and that the error terms are asymptotically normally distributed, then I will apply the logit model that has much better statistical properties for analyzing dichotomous responses.
Bibliography
Akaike, H. (1974). A new look at statistical model identification. IEE Transact. Auto. Cont. AIC 19, 716–723.
Blumer, H. (1956). Sociological analysis and the variable. Am. Sociol. Rev. 21 (December), 683–690.
Cole, S. (Ed.) (2001). What’s wrong with sociology? New Brunswick: Transaction Publishers.
Coleman, J. S. (1964). Introduction to mathematical sociology. New York: Free Press.
Coleman, J. S. (1981). Longitudinal data analysis. New York: Basic Books.
Davis, J. A. (1975). Analyzing contingency tables with linear flow graphs: d systems. In D. R. Heise (Ed.), Sociological methodology 1976 (pp. 111–145). San Francisco: Jossey-Bass.
Davis, J. A. (1980). Contingency table analysis: Proportions and flow graphs. Qual. Quan. 14, 117–153.
Davis, J. A. (2001). What’s wrong with sociology?. In S. Cole (Ed.), What's wrong with sociology? (pp. 99–119). New Brunswick: Transaction Publishers.
Fleiss, J. L. (1981). Statistical methods for rates and proportions. New York: Wiley.
Gelman, A., and Hill, J. (2007). Data analysis using regression and multilevel/hierarchical models. New York: Cambridge University Press.
Goldberger, A. (1991). A course in econometrics. Cambridge: Harvard University Press.
Gross, N. and Simmons, S. (2007). The social and political views of American professors. Harvard Sociology Department, Working Paper, September 24.
Hellevick, O. (2009). Linear versus logistic regression when the dependent variable is a dichotomy. Qual. Quan. 43, 59–74.
Lazarsfeld, P. F., and Wagner Thielens, Jr. (1958). The academic mind. New York: Free Press.
Lazarsfeld, P. F. (1961). The algebra of dichotomous systems. In Solomon H. (Ed.), Studies in item analysis and prediction. Stanford: Stanford University Press. Reprinted in Continuities in the language of social research, edited by Paul F. Lazarsfeld, Ann K. Pasanella, and Morris Rosenberg, 193–207. New York: Free Press, 1972.
Lazarsfeld, P. F. (1971). Regression analysis with dichotomous attributes. In P. F. Lazarsfeld, A. K. Pasanella, and M. Rosenberg (Ed.), Continuities in the language of social research (pp. 208–214). New York: Free Press.
Littell, R. C., Milliken, G. A., Stroup, W. W., Wolfinger, R.D., and Schabenberger, O. (2006). SAS for mixed models, (2nd Ed.). Cary NC: SAS Institute.
Raudenbush, S W., and Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods, (2nd Ed.). Newbury Park, CA: Sage.
SAS Institute. (2005). The Glimmix procedure, November. Cary NC: SAS Institute.
Schwarz, G. (1978). Estimating the dimension of a model. Anna. Stat. 6, 461–464.
Singer, J. D., (1998). Using SAS PROC MIXED to fit multilevel models, hierarchical models, and individual growth models. J. Educ. Behav. Stat. 244, 323–355.
Singer, J. D., and Willett, J. B. (2003). Applied longitudinal data analysis. New York: Oxford University Press.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2010 Springer Science+Business Media B.V.
About this chapter
Cite this chapter
Smith, R.B. (2010). The Academic Mind Revisited: Contextual Analysis via Multilevel Modeling. In: Capecchi, V., Buscema, M., Contucci, P., D'Amore, B. (eds) Applications of Mathematics in Models, Artificial Neural Networks and Arts. Springer, Dordrecht. https://doi.org/10.1007/978-90-481-8581-8_9
Download citation
DOI: https://doi.org/10.1007/978-90-481-8581-8_9
Published:
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-8580-1
Online ISBN: 978-90-481-8581-8
eBook Packages: Humanities, Social Sciences and LawSocial Sciences (R0)