Path: > Syllabus > Topics and Readings > Readings Menu > Simple Correlational Analysis
Research Applications

Simple Correlational Analysis

"Determinants of Abortion Policy in the Developed Nations,"
Policy Studies Journal, (Summer, 1979), 711-781.

Marilyn J. Field, Syracuse University

In the more developed nations of the world, the last two centuries have brought dramatic changes in the lives of individuals and the scope of public policy. Among the most interesting changes are those involving fertility. The use of modern as well as primitive techniques of birth control has grown tremendously. Where most families once expected at least six or seven children, today they plan for two or three or sometimes none.

The expanded use of birth control and the development of family planning propaganda over the last two centuries have been accompanied first by a conservative and then, since about World War I, by a largely liberal expansion of government policy concerning contraception and abortion.[1] Few nations now retain substantial restrictions on the sale of contraceptives or the provision of birth control information although a number of nations continue to limit legal access to abortion.

What accounts for the continuing variation across nations in policies concerning abortion? This article reports some research findings relevant to this question, at least as it pertains to the more developed states. The findings also speak to a question that has concerned many social scientists over the last couple of decades: do political structures and values have much independent impact on public decision-making?

This last question has surfaced in many places--in the literature on community power, in studies of Individual and organizational decision processes, and, most provocatively, in the early comparative analyses of public policy which appeared in the 1960's.[2] From these varied streams comes a common proposition: the impact on policy outcomes of political institutions and public officials has been greatly exaggerated and the influence of social and economic characteristics of polities neglected.

Later comparative policy studies have found political factors to be more significant than the first research studies indicated.[3] These studies have also recommended that attention be paid to qualitative policies as well as the more frequently studied expenditures policies; that more use be made of longitudinal as well as cross-sectional data; and that a broader range of jurisdictions and variables be examined. Since abortion policies are largely qualitative ones aimed at directing individual behavior through regulations and incentives and since moderately good policy data are available across a varied group of countries and a considerable range of time, abortion policies constitute a good area in which to extend comparative policy research.

Moreover, abortion policies are worth studying because they are substantively important. They pose fundamental questions for societies concerning the definition of personhood, the value of life, and the competition between individual desires for self- determination and fulfillment and community or elite objectives regarding morality, economic development and national security. These questions have occasioned extreme controversy, producing multiple conflicts of values among social groups. If ideological differences are ever systematically linked to policy choices, we might expect to find statistical evidence of such links for abortion policies.

HYPOTHESES. The following hypotheses will be examined in this paper:

1. The greater the proportion of Catholics in a country, the more conservative will be the abortion policies.
2. The greater the strength of leftist, non-Communist parties under democratic regimes, the more liberal the policies on abortion.
3. When controlled for proportion Catholic, the association between leftism and policy liberalism will diminish.
4. Under democratic regimes, liberal policies will more frequently have been adopted by Socialist parties than by other parties.
5. Within Communist regimes, the role of Catholicism will be negligible.
6. Within Communist regimes, abortion policies will be more liberal than within non-Communist regimes.
7. Totalitarian or authoritarian regimes will differ in their policy history from democratic regimes.
8. The association between abortion policy and socioeconomic factors will be less strong than the relationship between abortion policy and political factors.
Although certain of the above hypotheses may seem obvious to those with some knowledge of the controversies over abortion, the record of social science research and of comparative policy research, in particular, is full of obvious hypotheses which have not held up under empirical testing.[4]
METHODOLOGY. Tests of the above hypotheses are based primarily on an analysis of quantitative data from 29 nations involving independent variables prominent in the literature on birth control and public l)policy and supplemented by qualitative materials. (The larger study from which this article derives presents a history of abortion policy development and reports on policy interrelations and diffusion as well as on the social, economic and political correlates of birth control policy variation and change, some of which are reviewed here.)[5]

Case Selection. The concept of development underlying the selection of cases focuses primarily on population well-being and modernity rather than on economic or political system moderntiy per se. Twenty-nine nations qualify as developed on the basis of the following standards: (1) population over 85 percent literate by 1960; (2) female life expectancy over seventy years at age zero by mid- or late-1960's; and (3) crude birthrate below 25 births per 1,000 population in 1965 or nearest year for which Information exists. Colonies and other entitles lacking political independence have been excluded as have been most very small nations and nations for which Information on the selection criteria was unavailable. (Nations selected are listed in the next section of this paper.)

Abortion Policy. More than on any other basis, legal access to abortion a determined by the reasons (known in the literature as indications) permitted by law. The legal justifications for abortion may be differentiated according to whether they provide:

1. no acceptable reasons or indications;
2. life-saving Indications only;
3. life- and health-preserving Indications;
4. medical/social indications--meaning that social circumstances affecting a woman's health may be taken into consideration but not used as sole grounds for abortion;
5. purely social indications beyond a woman's age and number of previous births; and
6. request of the woman as a sufficient indication.
This list of permissible reasons for abortion forms a reasonably straight forward policy continuum running from most to least restrictive, from conservative to liberal.
Nations that allow the same indications for abortion have been further differentiated, first, by any post-conception time limits on abortion and, second, by any major bureaucratic Interventions in the doctor-patient decision process (e.g., approval required from a second physician or a regional abortion commission). Such intervention increases the opportunity for disagreement and delay in a situation where prompt action is legally and medically important and implies that conservatism is desirable or otherwise a woman and physician's agreement would be enough.
The 1972 ranking of the 29 nations from most to least restrictive is as follows: (1) Ireland; (2) Spain, Belgium, Netherlands; (3) Italy; (4) France. New Zealand; (5) Argentina, Austria; (6) West Germany, Switzerland; (7) Canada; (8) Uruguay; (9) Australia; (10) United States; (11) Iceland; (12) Romania; (13) Bulgaria; (14) Norway; (15) Sweden; (16) Czechoslovakia, Denmark; (17) Finland; (18) United Kingdom; (19) Japan. Poland; (20) Hungary, U.S.S.R.; (21) East Germany. The 1962 rankings will not be given here but annotated lists of both sets of rankings as well as sources of data are presented elsewhere.[6]

In addition to the rankings, the data analysis uses a policy change variable computed by subtracting the 1962 ranks from the 1972 ranks.

Ideally the intensity and variability of abortion law enforcement should enter into a cross-national policy analysis. Unfortunately, lack of adequate comparative data on aspects of abortion policy implementation and law enforcement necessitate an emphasis on legislation, judicial decisions and major executive degrees. Even laxly enforced restrictions appear, however, to have had an impact by tending to (1) Increase the price of competent abortion services thereby limiting the access of lower income women to safe abortions; (2) increase maternal morbidity and mortality rates; and (3) inhibit an improvement in medical techniques and the exchange of medical knowledge among care providers and consumers.

Independent Variables. The measures of socioeconomic development on which the evaluationof hypotheses is based are fairly conventional, but due to space limitations, only data on energy consumption per 1,000 population and GNP per capita will be presented. (See Table 1 for a note on data sources.)

The treatment of Catholicism as a political variable is based on a conceptualization of politics as action, organization and belief more-or-less deliberately aimed at or concerned with influencing public policy. Although religious groups engage in many not-deliberately-political acts, they also seek to shape policy in a variety of ways. Since Catholics are only the most prominent but not the sole religious opponents of abortion, a more elaborate and accurate portrayal of the policy impact of religious groups would be possible if adequate multi-nation data were available on the strength of anti-abortion Protestant or Jewish groups. Also, using percent Catholic as a measure of Catholic strength does not reflect the variation across nations in proportions of believers and nominal Catholics. However, alternative measures such as priests and seminarians per 1,000 population show patterns similar to the percent Catholic measure except, interestingly, In the cases of Austria, Uruguay and Argentina, nations with somewhat more liberal policy histories and fewer priests and seminarians than would be expected given their proportion of Catholics.

Data Analysis. The statistical analysis relies primarily on simple correlations supplemented by some multivariate statistics. Applying these interval level statistical techniques to an ordinal dependent variable and interval independent variables is consistent with the approaches suggested by Labovits and Tufte.[7]

The evaluation of hypotheses in this paper depends in large on the cross-sectional analyses for two time points:1972 and 1962. At some points, data on policy change (in some cases going back to 1920) is also presented.

RESEARCH FINDINGS. The structure of the following discussion derives from an interesting contrast in the statistical findings presented in Table 1. The contrast is between (I) the 22 non-Communist nations where--as predicted--Catholicism stands out as a dramatic conservative influence on abortion policy and (2) the seven Communist nations where Catholicism shows a alight positive association with abortion policy. The relationships involving Catholicism are discussed initially for the non-Communist nations alone with the impact of regime type on policy-making considered subsequently.
TABLE 1: Simple Correlations between Political and Socio-Economic Variables and
(1) Liberalness of Abortion Policies in 1962 and 1972 and (2) Amount of Abortion Policy Liberalization
for All Nations, Non-Communist Nations, and Communist Nations[a]

All Nations


Correlations with Abortion Policies, in separate years--1962 and 1972

% Pop. Roman Catholic
% Socialists Elected[b]
Energy Consump. per Cap.
GNP per Capita

Correlations with Changes in Abortion Policies, from 1962 to 1972[c]

% Pop. Roman Catholic
% Socialists Elected[b]
Energy Consump. per Cap
GNP per Capita

a. Data Sources for dependent variables are described and discussed in Marilyn Field Clark, "The Comparative Politics of Birth Control," PhD Dissertation, Department of Political Science, University of Michigan, 1976. Data on independent variables is from Michael Taylor and Charles Hudson, World Handbook of Political and Social Indicators (Ann Arbor: Inter-University Consortium for Political Research, 1972).
b. Two cases of missing data.
c. For this analysis, change in the dependent variable (1972 minus 1962 policy ranks) is correlated with static values of the independent variable. If there was unusual change during this period in the relative position of a nation on an independent variable, it would not be reflected here. On the whole, however, for the variables for which comparisons are possible, major change during the decade from 1962 to 1972 appears to be the exception rather than the rule.

Catholicism and Socialist Party Strength, For the non-Communist countries in 1972, size of the Roman Catholic population accounts for nearly two-thirds of the (total) variation across nations in liberalness of abortion policies. Moreover, even a small Catholic population appears historically to have exerted a remarkably conservative influence on policy. For example, all countries with fewer than five percent Catholics have permitted medical/social indications for abortion for at least 17 years but no nation with more than ten percent Catholics had legalized these indications before 1973. (Britain, about eight percent Catholic, legalized abortion under most circumstances in 1967.)

Our hypothesis that greater Socialist party strength will correlate with liberal birth control policies is also supported by the figures in Table 1 (r .51 for 1972 abortion policies). Given the history of Catholic hostility toward leftists, it is not surprising that, when controlled for Catholicism, the association between policy and Socialist party strength drops from .51 to . 13.

Is there reason to attribute some developmental or Independent role to Socialist parties rather than to label the Socialist/ policy correlations spurious? Historical data about the types of political parties which have presided over legislative policy changes during the period 1920-74 suggest that the latter interpretation is too conservative. More radical abortion policies (i.e. those which permit at least medical/social indications) have in seven out of ten instances been adopted by Socialist governments whereas the four changes limited to pure health grounds have all occurred under governments not controlled by Socialists.

In the mid-1970's, three countries with substantial Catholic populations liberalized their abortion policies. Two, Austria and Germany, had relatively new Socialist dominated governments. In the third, France, Socialists lost the election preceding law change by a small margin following a campaign in which abortion law reform was a prominent, Socialist-supported issue. The liberalization in France was more limited than that in Germany or Austria. In 1978 in Italy, leftist political parties organized a coalition that overcame the opposition of the minority Christian Democratic government to policy liberalization.

During the period 1920-74, two other democratic nations adopted liberal abortion policies under governments not dominated by Socialists. Iceland was the first democratic nation to permit other than purely medical reasons for abortion. The circumstances of change in Iceland are not clear, but the government in power in 1934 was a hard-to-classify coalition, apparently left of center with Socialist representation but greater agrarian party strength. Also, Iceland with Its minuscule Catholic population has no history of Catholic/Socialist conflict.

In Japan, another nation with a tiny Catholic population, the major liberalization of the abortion law was essentially the work of a short-lived post-war Socialist government. According to a study of the decision, the American occupation forces were reluctant to allow change under that government but did not object when the succeeding non-Socialist government adopted similar legislation. 8

To summarize, the evidence points to the Catholic Church as the dominant influence on abortion policy decisions in non-Communist nations. Leftist political parties, though inhibited by Catholicism, appear on some occasions to make an independent contribution to the content and timing of policy liberalizations.

Catholicism, Regime Type and Policy. Communist nations are, as predicted, more liberal than non-Communist nations with a median policy rank of 19 for the former and 6.5 for the latter. Also, as predicted, the size of the Catholic population does not appear to be a major factor in abortion policymaking for Communist nations. The 1972 correlations between Catholicism and abortion policy, reported in Table 1, tend to be weakly positive for the seven Communist nations rather than strongly negative as for non-Communist countries.

No pattern of religious influence appears associated with the timing of Communist policy actions either. The last Eastern European nation to liberalize Its abortion law was East Germany (ten percent Catholic) whereas the most Catholic nation, Poland, was one of the first (1956) to change its policy. Some Communist regimes have recently restricted the social grounds for abortion, but the religious composition of the population seems to be unrelated to these decisions.

In addition to affecting the relationship between Catholicism and abortion policy, the presence of non-democratic regimes--in this case, non-Communist as well as Communist--appears to be associated with deviations from the general trend over the last 50 years toward policy liberalization. Nearly two-thirds of the liberal changes have occurred under democratic regimes, but four-fifths of the conservative changes have taken place under non-democratic regimes.

Recent conservative actions by Romania and, to a lesser degree, Bulgaria, Czechoslovakia, and Hungary recall the move by the Soviet Union in the 1920's restricting legal access to abortion. The non-Communist nations with authoritarian regimes show a somewhat similar pattern of conservative change. The German, Italian, French, Japanese and Spanish fascist or Militarist governments of the second quarter of this century all passed or strengthened existing conservative laws concerning birth control. More recently, the increasingly authoritarian regimes of Argentina and Uruguay appear to be reversing an earlier trend toward neutral or permissive birth control policies. In contrast, among democratic regimes, only one nation--Ireland in 1935--has strengthened conservative policies in the past 50 years.

Just as it appear to matter whether a regime is democratic or not, it also appears to matter whether democratic states permit an active policy role for institutions which lack representativeness in the conventional electoral sense. In several nations, relatively unrepresentative bodies--courts, national commissions, hereditary legislative chambers--have played key roles in abortion policy liberalization. Moreover, decision processes Involving abortion have also found some non-Communist governments and party elites (for instance, in England and Canada) atypically reluctant to play leadership roles, thus leaving room for backbench initiative and procedural unorthodoxies such as free votes.

These patterns support T. Alexander Smith's classification of abortion policy as one of a distinctive set of "emotive symbolic" policies such as women's rights, capital punishment, school prayer and homosexuality. ~ Such policies deal with deeply held values that relate more to "way of life" rather than economic issues. And, although the point has been strongly argued, it does appear that this type of difference in policy content does affect the nature of the policy process. In the case of abortion, we find supporters of liberal change avoiding conventional party political processes and instead exploiting the institutional framework of government by taking advantage of unorthodox parliamentary procedures and by probing for sympathy and response In bodies not directly dependent upon relatively small margins of electoral victory. To the extent that these instruments and institutions successfully deflect the charged emotions associated with this highly controversial issue from governments and parties, they may minimize the destabilizing and polarizing effects that such an issue may have.

Recent Policy Change. The correlations in Table 1 indicate that were about as conservative, compared to other nations, in 1972 as in 1962. They were also less likely to have liberalized their policies than other countries. Since 1972, however, liberalizing nations (e~g.. France, Sweden. West Germany,have averaged about 50 percent Catholic compared to 13 percent for thee preceding ten year periodl What appears to have happened is that the most well-off nations--whether Catholic or not--have

become more receptive to liberalization with less well-off nations (e.g., Argentina, Spain. Ireland) remaining relatively conservative, The figures in Table 1 presage this development, showing also more liberalization among the better-off nations. This pattern differs from that of the 1930s through 1950s when the liberalizing nations of Scandanavia and Eastern Europe were not among the leading industrialized nations.

By way of summary, we may note that even though socioeconomic factors have become more visible quantitatively, they do not dominate cross-sectional statistical explanations of abortion policy variation. For the 22 non-Communist countries, the percent of the population Roman Catholic accounts for nearly 65 percent of variation In 1972 abortion policies with the most important soclo-economic variable, per capita energy consumption, adding only one further percent to explained variation. (The standardized partial regression coefficients are -.84 and -.10 respectively, supporting the multiple correlation findings.) Across all 29 nations, regime type and Catholicism together account for about i~0 percent of variation in abortion policies with energy consumption again adding little to the analysis. (Respective standardized partial regression coefficients for the three variables are .54, -.50 and .06, A more elaborate analysis using interactive terms, additional independent variables, and path analysis is consistent with the results reported here.)

CONCLUSIONS. What can we conclude concerning the questions which introduced this study: how are variations in abortion policies among the developed nations to be explained; and, In particular, do political structures and values matter? Our findings suggest that political institutions and ideologies do affect the nature and timing of policy decisions on abortion, contributing distinctively to policy differences across nations, at least in the short-run. The increasing visibility of economic factors in the last few years suggest the importance over the long run of economic forces affecting fertility related practices, values and policies.
Overall, Hugh Heclo's comment on the development of unemployment and pension policies in Britain and Sweden seems largely applicable here:
There is no valid either/or choice--politlcal versus socioeconomic variables--in understanding the growth of modem social policy.. . The grand choice between economic and political explanations turns out to be little more than a difference in analytic levels, a difference between socioeconomic preconditions and the political creation and adjustment of concrete policies. 10
The above quote implies end this research confirms the worth of multiple analytic strategies. The impact of Catholicism, the partially independent role of Socialist parties, the contribution of non-electoral institutions, and the varying visibility of socioeconomic factors could probably not have been caught by a single method of study. In further abortion policy research, the addition of mass and elite opinion data would be helpful, particularly given Godwin and Shepherd's recent call for the mediating roles of political structures to be more systematically analyzed. ~ In sum, policy study should be served by varied perspectives--a broad perspective across nations, a long perspective across time, and a deep perspective into the policy processes of individual nations.


*The major financial support for this research was provided by the American Association of University Women, the U.S. Public Health Service and the Rackham School of Graduate Studies, University of Michigan. The author is indebted to Robert Putnam, Richard Hofferbert, Susan Borker, Charles Levine, and Ken Auerbach for their advice and critique.

1. For a discussion of these developments and for documentation of the historical bases for the hypotheses presented in this paper, see Marilyn Field Clark, "The Comparative Politics of Birth Control," Ph.D. dissertation, Department of Political Science, University of Michigan, 1976, and a 1977 paper with the same title presented at the Midwest Political Science Association meeting, Chicago, April 21.

2. See, for instance, Peter Bachrach and Morton Baratz, "Two Faces of Power," APSR, September, 1963, pp. 947-53; Matthew Crenson, The Un-Politics of Air Pollution (Baltimore: Johns Hopkins Press, 1971); James March and Herbert Simon, Organizations (Rew York: John Wiley & Sons, 1958); Charles Lindhlom, 'The Science of Muddling Through," PAR, Spring, 1959, pp. 79-68; Thomas Dye, Politics, Economics and the Public (Chicago: Rand McNally, 1966).

3. These studies include Virginia Gray, "Models of Comparative State Politics," AlPS, May, 1976, pp. 235-56; R. Kenneth Godwin and W. Bruce Shephard, "Political Processes and Public Expenditures," APSR, December, 1976, pp. 1127-35; Sarah M. McCally, "The State Political Party and the Policymaking Process," APSR, March, 1973, PP. 55-72; B. Guy Peters, "Economic and Political Effects on Development of Social Expenditures in France, Sweden and the United Kingdom, MJPS, May, 1972, Pp. 225-38.

4. See, for instance, Claire Selitir, et al., Research Methods in Social Relations, 3rd ed. (Hew York: Holt Rinehart and Winston, 1976), ch. 1 and Thomas Dye, Understanding Public Policy, 2nd ed. (Englewood Cliffs, New Jersey: Prentice-hall, 1975) ch. 12.

5. Field Clark, 1976 and 1977 and "Models of Policy Determination," paper presented at the Midwest Political Science Association meeting, Chicago, April 21, 1978.

6. Field Clark, 1976, ch. 2 and Appendix B.

7. Sanford Lahovitz, "The Assignment of Numbers to Mank Order Categories," ASR, June, 1970, pp. 515-24 and Edward Tutte, "Improving Data Analysis in Political Science," World Politics, July, 1969, pp. 641-54.

8. Deborah Oakley, "The Development of Population Planning in Japan, 1945-52, and American Participation," Ph.D. dissertation, Department of Population Planning, University of Michigan, 1977.

9. Alexander Smith, The Comparative Policy Process (Santa Barbara: ABC-Clio, Inc., 1975), ch. 4.

10Hugh Heclo, Modern Social Politics in Britain and Sweden (New Haven: Yale University Press, 1974), p. 288.

11. Godwin and Shephard, 1976.