ernment. Between these extremes, one can postulate several intermediate policy positions, as demonstrated in the following scale.

Operational Definition. Protection of civil rights is scored as the pro position, which is consistent with our according positive scores to greater governmental activity in the issue area.

PRO-strong	5	Advocates a government policy of outlawing discrimination broadly across social life and providing for enforcement.
PRO-moderate	3	Advocates government policy of outlawing discrimination but qualified by time limits, areas of life, inadequate enforcement or restrictions on immigration.
PRO-weak	1	Advocates not the explicit outlawing of discrimination but its discouragement through incentive plans, alternative opportunities, and the like; opposes punitive state action against discriminators in favor of programs to encourage nondiscrimination.
NEUTRAL	0	Includes ambiguous or contradictory positions.
CON-weak	-1	Believes the protection of civil rights to be an inappropriate matter for government action; accepts discriminatory practices.
CON-moderate	-3	Accepts discrimination and advocates discriminatory policies in selected aspects of legislation.
CON-strong	-5	Advocates enactment of discriminatory legislation in broad areas of social life and establishment of penalties for noncompliance.

Coding Results. Tables 6.14a and 6.14b disclose that we have missing data for nearly half our parties, attesting to the difficulty in applying our operationalization of "protection of civil rights." Parties that supported civil rights were more likely to declare their position openly than those which did not. This selective advertisement of party sentiment probably accounts in part for the relatively high means for BV511 in the tables. Presumably, some of the parties not scored on this variable were tacitly opposed to the use of government action to discourage one part of society from discriminating against another. Note that our conceptualization of civil rights focuses on social discrimination rather than on political participation, with which the term "civil rights" has become linked in the contemporary American context. Coders were requested to identify a "target group" of social discrimination, which usually proved to be some minority in the society, for the purpose of coding this variable. If no prominent target group could be identified, which happened in some homogeneous societies, then the variable was judged inapplicable to party politics within the country. This undoubtedly accounts for another portion of parties that escaped coding on BV512. While there is no significant correlation between BV512 and AC512, the means for AC512 are among the lowest in the issue orientation cluster, reflecting considerable uncertainty underlying the coding. Users of these data should treat this variable cautiously.

go to page 72