Copyright Sociological Research Online, 1996

Chanoch Jacobsen and Tamar Vanki (1996) 'Violating an Occupational Sex-Stereotype: Israeli Women Earning Engineering Degrees'
Sociological Research Online, vol. 1, no. 4, <http://www.socresonline.org.uk/1/4/3.html>

Abstract

The percentage of women engineering graduates in Israel has increased fourfold during the last two decades, but only a small percentage of Israeli women opt for these fields. We account for the current trend by a general theory of patterned deviance, viewing the recent increase of women's studying for engineering degrees as a case of nonconformity with a traditional norm. A simulation model of that theory reproduced 85.8% of the variance in the data on women engineering graduates between 1966 and 1987, indicating that the theory applies also in this case. The simulations show that it is becoming increasingly legitimate for women to study engineering and informal social control keeping women from enrolling in engineering has almost disappeared, but the internalized sex-stereotype still deters many women from taking such courses.

Keywords:

Engineering; Higher Education; Israeli Women; Simulation; System Dynamics

Introduction

1.1

Although women's participation in Israel's paid labor force has increased from 29.5% in 1960 to 38.5% in 1986, their representation in high-prestige occupations such as engineering is less than half that figure (Central Bureau of Statistics, 1987). In this, the situation in Israel is not unlike that in other modern societies, where gender- stereotypes of occupations still persist (Bridges and Bower, 1985; Fitzpatrick and Silverman, 1989; Kvande and Rasmussen, 1986). However, there are signs of a gradual change. In 1975, only 3.03% of all graduates at Israel's Institute of Technology had been women. By 1980 they had increased to 14.56%, reaching 20.98% in 1987 (Technion, 1987). It appears, therefore, that the sex-stereotype which has for so long defined engineering as a male profession may be slowly giving way to greater freedom in occupational choices for women.

1.2

Whatever the practical ramifications of this trend, a change in such a deep-seated norm merits the attention of sociologists, both as a phenomenon worthy of observation and as an indication of possible future developments. What triggered the change is a moot question, since in all likelihood a number different factors were involved. More important is what gave the recent trend its particular shape and how far can it be expected to continue? In years to come, will gender no longer be relevant to choosing engineering as a career? Or perhaps there will be a feminization of engineering, as has occurred in the teaching profession, so that the sex-stereotype has merely switched sides?

1.3

Answers to such questions require not merely information on the particular situation, but also an analysis of the social processes that are involved, particularly those having to do with the crescive change of informal norms. One such analysis has been made by Jacobsen and Bronson (1985) in their theory of patterned deviance. The theory describes how and explains why isolated and sporadic nonconformities grow into what they have called "patterned deviances", some of which later develop into legitimate new norms. This theory is relevant to our issue because the early manifestations of behavioural innovations are usually perceived as deviations from the established norm. Women who ventured into engineering when normative expectations spoke against such a choice were, technically speaking, not conforming with the traditional norm. Therefore a theory that purports to explain patterned deviance in general should cover this case also.

1.4

We begin with a summary of the theoretical argument and a general description of the simulation model and its practical application. These will be followed by a juxtaposition of a time-series of data on Israeli women who have earned engineering degrees with the simulated output from the model. If the model can reproduce the trends in the empirical data, we can claim empirical adequacy of the theory to this case. Inferences will then be drawn from the simulations to explain the current trends and predict probable future trends of women engineering students in Israel.

Theory and Model

2.1

Jacobsen and Bronson (1985) set out to explain the growth and spread of patterned deviances from accepted norms in modern societies. This phenomenon has been documented by Williams (1951), Beichelt et al. (1969), Mars (1974), and Ditton (1977) among others. The theory builds upon the concepts of Smelser's (1963) theory of collective behavior, integrating some of the classic insights about norm systems to explain the patterning process. Since we are concerned in this case with an informal norm, we shall not go into that part of theory that deals with formal norms and controls.

2.2

Modern societies are structurally conducive to a weakening of social controls of all kinds. The permeable, transient and heterogeneous structure of such societies gives rise to different, equally legitimate behaviour patterns. Consistent socialization is thus more difficult and autonomous control by internalized norms correspondingly rare (Sutherland and Cressey, 1960). In addition, the application of informal controls to non- conforming behavior is impeded by people's anonymity and transience in modern social structures (Simmel, 1950: pp.413 - 4; Wirth, 1938). Consequently, the more impediments there are to social control, the fewer people will be deterred from nonconformities if they feel pressure to do so.

2.3

Rapid technological or demographic changes create structural strain between traditional norms and current needs and wants of individuals. The strain lends partial legitimacy to violations of that norm (Berger and Luckman, 1967: p. 93), so that isolated and sporadic nonconformities will, over time, acquire an institutionalized pattern (Merton, 1957: pp. 318 - 9). Therefore, the more rapid and far- reaching the changes are, the greater will be the legitimacy of related norm violations.

2.4

The dynamic linkages between these postulates create two positive feedback loops. When weakened mechanisms of social control can no longer contain the violations' growing legitimacy, these feedbacks drive the norm system towards progressively more violators until either there are no more potential violators or a new informal norm has been crescively institutionalized, or both (Fig. 1).

Figure 1: Causal Loop Diagram

2.5

This theory has been repeatedly tested in a simulation model constructed specifically to simulate its postulated dynamics. In 13 out of 14 tests it was found empirically adequate to explain patterned deviance from a variety of formal as well as informal norms, from illegal immigration to drug abuse, and from unmarried cohabitation to smoking in company (Jacobsen and Bronson, 1985; 1989: pp. 531 - 6; 1995; Jacobsen and Hanneman, 1992). The case of women studying engineering may therefore be considered as one more test of the theory and model.

2.6

We have used the modeling technique known as System Dynamics (Forrester, 1961; Richardson and Pugh, 1981) to construct a model of this theory. The basic argument of System Dynamics is that, over time, the behaviour of any complex social system is shaped as much by its internal feedbacks as by exogenous events acting upon it. We therefore have to model the structure of such a system in order to explain its ongoing behaviour. The case for using this technique to test social theories, as well as the strategy we employed to do this have been detailed elsewhere (Jacobsen and Bronson, 1985; 1995). Perhaps most important for the present issue is that in this way we can test a theorized process by simulating the entire system's behaviour over time, rather than by cross-sectional analyses of selected variables.

2.7

In the model, each concept in the theory is defined as a numerical variable, each variable being a mathematical function of its directly antecedent variable(s) in the causal chain. The functions are normally either positive or negative S-shaped curves (Hamblin et al., 1973), except where the theoretical argument dictates a different relationship. While all variables as well as the relationships between them are thus mathematically defined, the functions are not precise, since they are nothing but logical deductions from the theory. The theory's empirical adequacy is therefore not tested by the accuracy of the separate functions (we know that they are probably not accurate), but rather by the behavior of the complete system over time as seen in the trends simulated by the model. If the model, once it has been initialized to define the initial situation, reproduces the trends as well as the variance in the empirical time-series with its corresponding variables, then the theory can be said to be empirically adequate to that particular case, and we can apply it to explain the data. If not, the theory does not fit.

2.8

The complete model contains 42 variables and relationships as postulated in the theory (the documented code is available from the senior author). But as we are dealing here with an informal norm, we have inactivated the section dealing with formal control, making the model considerably simpler. A flow-diagram depicting the section pertaining to informal norms is shown in Figure 2. This model should also fit the documented evidence about Israeli women earning engineering degrees.

Figure 2: Informal Norms Model

Data and Simulation

3.1

Official records from universities that confer engineering degrees show the number of men and women who have graduated from any of Israel's engineering schools and departments in the period 1966-1987 (Table 1). The figures were converted into percentages of all university graduates in Israel, for women and for men separately. The year 1978 showed a sharp peak, due to a wave of immigration in the early 1970s from the USSR, where women engineers are not so unusual. From the records we know that young women from this immigration wave caused the peak in the data when they graduated some years later. We therefore adjusted the number of graduates for 1978 to the average of the adjoining years. These are the data which we tried to reproduce with the model.

---

Table 1: Total Graduates and Engineering Graduates, Women and Men, 1966 - 1987

---

	Women			Men
Year	Total Graduates	Enginineering Graduates	% Engineering	Total Graduates	% Engineering
1966	931	5	0.537	1245	30.92
1967	1245	9	0.723
1968	1330	8	0.602
1969	1512	24	1.587
1970	1742	27	1.550
1971	1842	27	1.466
1972	2012	29	1.441
1973	2174	38	1.748
1974	2259	46	2.036
1975	2490	58	2.329	3059	42.56
1976	2548	54	2.119
1977	2596	62	2.388
1978	2643	92*	3.481
1979	2814	68	2.416
1980	3238	64	1.977	3564	34.32
1981	3358	75	2.233
1982	3455	73	2.119
1983	3688	73	1.979
1984	3851	82	2.129
1985	3957	83	2.098
1986	4256	120	2.820
1987	4339	105	2.420	4410	24.01

^*Adjusted to n=65

---

3.2

As it is unlikely that a woman will pose as a male for four years of engineering study, the model variable corresponding to the data is OVERT (overt violators). This variable was initialized to coincide with the first data point. The other initial values are estimates, based on what is generally known about social conditions in Israel at the time. As always, these estimates had to be consistent with the known parameters as well as with the internal logic of the theory, so that they are quite severely constrained.

3.3

The estimated initial values, and the range of values which the variables could have been given and still satisfactorily reproduce the data, are shown in Table 2. The percentage ranges for PRESUR (potential violators, ie. those who feel the urge to become engineers) and for OVERT are quite narrow because the estimates had to be fairly precise to fit the initial situation. But for the other variables the ranges are so wide that setting them a few percentage points higher or lower would have made little difference to the performance of the model. It is therefore not these particular estimates that reproduced the data, but rather the difference equations that are at the heart of the model and express its internal dynamics. These, we should remember, were not formulated for this case alone, but are part of the general model which had satisfactorily reproduced 13 other data sets (Jacobsen and Bronson, 1995).

Results And Discussion

4.1

In the best simulation run, the model reproduced 85.8% of the data variance with variable OVERT (Figure 3). This indicates that the recent increase of women who study engineering can also be explained by the theory of patterned deviance. The stress is on the word "also", because it is possible of course to construct a much simpler model to fit this particular data set. But the point is not to construct a model that can reproduce these data, nor even to formulate a new theory that explains the trends, but rather to explain the trends with an existing, previously tested, general theory. Such a theory will always be preferable from a scientific standpoint to any situation-specific economic or political explanation (Homans, 1967: p. 26), even if the latter may be very persuasive in particular cases (Office of Technology Assessment, 1988).

4.2

The general theory expressed in the model has been tested previously and found empirically adequate in 13 out of 14 different studies. Moreover, the one case where the model failed to reproduce the data, traffic violators in Israel, showed that the model is not so general nor so flexible that just any data set could be reproduced with it (Jacobsen and Bronson, 1995). For the present case, however, it may be asked how different would the data have to be for the model to fail? An answer can be found in the data point for 1978, when an exogenous factor had created a sudden peak. This one fluctuation increased the data variance so much that the model had difficulty in reproducing the trend. The theory predicts a monotonically rising curve, and consequently the model cannot reproduce wide oscillations in the data without exogenous input, which would require separate and independent data. With that qualification, it does provide an explanation for the trends shown in Figure 3.

Figure 3: Plotted Output

4.3

The simulation that gave the best fit showed, among other things, that the percentage of the population who consider it legitimate for a woman to study engineering has increased from our initial estimate of 40% in 1966, to 58% in 1987 (Table 3). When we initialized legitimacy at only 11% (the lowest initial value to reproduce the data satisfactorily), it grew by the same 18% to 29%. In other words, even with an extremely conservative estimate of initial legitimacy, the simulation shows that it is rapidly becoming legitimate to defy the traditional occupational sex-stereotype and study to become an engineer. On the other hand, the asymptotic trend of the data in Figure 3 indicates that it is tapering off. What is the reason for that?

---

Table 3: Simulation Output for Selected Variables (%)

---

Year	Autonomous Control (AUTCON)	Informal Control (INFCON)	Legitimacy (LEGIT)	Overt Violators (OVERT)
1966	73.03	10.57	40.0	0.53 7
1967	69.59	8.50	40.2	0.827
1968	64.33	5.94	42.5	1.068
1969	59.04	3.95	44.5	1.273
1970	54.67	2.60	46.2	1.448
1971	51.09	1.66	47.7	1.596
1972	47.37	1.16	49.3	1.722
1973	44.22	0.82	50.7	1.829
1974	41.62	0.59	52.0	1.921
1975	39.47	0.48	53.0	1.998
1976	37.70	0.38	53.9	2.064
1977	36.23	0.31	54.6	2.119
1978	34.92	0.25	55.2	2.166
1979	33.69	0.23	55.8	2.205
1980	32.65	0.22	56.2	2.239
1981	31.78	0.21	56.6	2.268
1982	31.05	0.20	56.9	2.292
1983	30.44	0.19	57.2	2.313
1984	29.92	0.19	57.4	2.331
1985	29.48	0.18	57.6	2.345
1986	29.12	0.18	57.8	2.358
1987	28.81	0.18	58.0	2.369

---

4.4

Turning again to Table 3, we see that informal social control, initialized by setting structural impediments at 80%, dwindled from 10.57% to 0.18%. Even with impediments initialized as low as 1%, informal control still dropped to 0.8%. We know, therefore, that the tapering off of the ascending trend is not due to increasing informal social control. Autonomous control, on the other hand, though also decreasing rapidly, was still relatively strong (29%) the end of the simulated period. When INSOC (the percentage socialized to the traditional norm) was initialized at 60% (the best run), autonomous control had dropped some 44.2% by 1987. When we initialized INSOC at only 27% (the lowest value to reproduce the data), autonomous control was still 18% at the end of the run.

4.5

It appears, therefore, that by 1987 it was neither a lack of legitimacy nor informal social control that kept so many young women in Israel from studying engineering. Rather, it was that relatively few of them had overcome the internalized traditional norm sufficiently to want to try engineering as a profession. It seems reasonable to argue that this absence of a more widespread desire to be engineers is at least partly due to the kind of socialization which most Israeli women have been receiving in their formative years. Although the sex-stereotype of the male engineer has definitely weakened over the last decades and fewer young women now feel pressured to conform to it than in the past, it still exists. Some more time will have to pass before the present pioneering cohorts of women engineers can serve as effective role models for teenagers and younger women who are now planning their futures.

Acknowledgments

The authors thank Albert Goldberg for his critical comments on an earlier version of this article.

Appendix: Acronyms, Variables and their Definitions

AUTCON (Autonomous Control)

Percentage wanting to violate the norm, but are deterred by internalized sanctions.

COMPLY (Compliers)

Percentage who comply with the norm.

CRESIV

Probablity of a newly institutionalized informal norm.

CRISIS

Probability of acute threats from outside the norm system.

DELCTO

The delay involved for a complier to become a violator.

DIFFER (Difference)

The percentage difference between expected violators and actual violators computed by the model.

EXPECT

Expected Violators: PRESUR * NODETR.

EXTFOR (Exogenous Forces)

Changes that cause norm violations.

IMPEDS (Impediments)

Percentage undeterred by informal control.

INFCON (Informal Control)

Percentage wanting to violate the norm, but deterred by informal pressures.

LEGIT (Legitimacy of Violations)

Percentage who deem the norm unreasonable or unacceptable.

NODETR (Nondeterrence)

Total percentage wanting to violate the norm who are not deterred from violation.

NORAMB (Norm Ambiguity)

Percentage who do not know how to comply.

OVERT (Overt Violators)

Percentage who overtly violate the norm.

PERMIS (Permissiveness)

Percentage feeling constrained from reacting to overt violations.

PRESUR (Potential Violators)

Percentage who want to violate the norm.

QUITOV (Reform Rate)

Yearly percentage of OVERT who cease violation.

SOCIAL (Socialization)

Population percentage socialized to the norm.

TRYOV

Yearly percentage of compliers who try overt violation.

References

BEICHELT, B., BIERVERT, J., DAVITER, J., SCHMOLDERS, C. & STRUMPELL, B. (1969) Steuernorm und Steuerwirklichkeit. Koln and Opladen: Westdeutscher Verlag.

BERGER, P. & LUCKMAN, T. (1967) The Social Construction of Reality. New York: Anchor Books.

BRIDGES, J. S., & BOWER, M. S. (1985) 'The Effect of Perceived Job Availability for Women in College: Women's Attitudes Towards Prestigious Male-Dominated Occupations', Psychology of Women Quarterly, vol. 9, pp. 265 - 77.

CENTRAL BUREAU OF STATISTICS (1987) Statistical Abstract of Israel, Vol. 38. Jerusalem: Prime Minister's Office.

DITTON, J. (1977) Part-Time Crime: An Ethnography of Fiddling and Pilferage. London: Macmillan.

FITZPATRICK, J. L., & SILVERMAN, T. (1989) 'Women's Selection of Careers in Engineering: Do Traditional Differences Still Exist?', Journal of Vocational Behavior, vol. 43, pp. 266 - 78.

FORRESTER, J. W. (1961) Industrial Dynamics. Cambridge, MA: Wright-Allen.

HAMBLIN, R. L., BROOKE JACOBSEN, R. & MILLER, J. J. (1973) A Mathematical Theory of Social Change. New York: Wiley.

HOMANS, G. C. (1967) The Nature of Social Science. New York: Harcourt, Brace & World.

JACOBSEN, C. & BRONSON, R. (1985) Simulating Violators. Baltimore, MD: Operations Research Society of America.

JACOBSEN, C. & BRONSON, R. (1989) 'Thank You For Not Smoking: Simulating the Growth of a New Social Norm' in M. C. Jackson, P. Keys, and S. A. Cropper (editors) Operational Research and the Social Sciences. New York: Plenum.

JACOBSEN, C. & BRONSON, R. (1995) 'Computer Simulations and Empirical Testing of Sociological Theories', Sociological Methods and Research, vol. 23, no. 4, pp. 475 - 506.

JACOBSEN, C. & HANNEMAN, R. A. (1992) 'Illegal Drugs: Past, Present, and Possible Futures', Journal of Drug Issues, vol. 22, no. 1, pp. 105 - 120.

KVANDE, E. and RASMUSSEN, B. (1986) Women Challenge Technology. Oslo: Norwegian Institute of Technology.

MARS, G. (1974) 'Dock Pilferage: A Case Study in Occupational Theft' in P. Rock and M. McIntosh (editors) Deviance and Social Control. London: Tavistock Publications.

MERTON, R. K. (1957) Social Theory and Social Structure. Glencoe: Free Press.

OFFICE OF TECHNOLOGY ASSESSMENT (1988) Educating Scientists and Engineers: Grade School to Grad School. Washington, D.C.: U.S. Government Printing Office.

RICHARDSON, G. P. & PUGH, A. L. (1981) Introduction to System Dynamics Modeling with Dynamo. Cambridge, MA: M.I.T. Press.

SIMMEL, K. (1950) The Sociology of Georg Simmel. Translated by K. H. Wolff. New York: Free Press.

SMELSER, N. J. (1963) A Theory of Collective Behavior. New York: Free Press.

SUTHERLAND, E. H., & CRESSEY, D. R. (1960) Principles of Criminology. Chicago: Academic Press.

TECHNION (1987) Official Graduate Records, 1966- 1987. Haifa: Israel Institute of Technology.

WILLIAMS, R. M. Jr. (1951) American Society. New York: Knopf.

WIRTH, L. (1938) 'Urbanism as a Way of Life', American Journal of Sociology, vol. 44, pp. 3 - 24.

Copyright Sociological Research Online, 1996