In Canada, according to the 1989 Labour Market Activity Survey,more than 40 % of nonagricultural paid workers work in the private service sector. Only 10.7 % of these workers are unionized compared with 37.9 % in the goods sector and 58.6 % in the public sector.

A model of the determinants of individual propensity to unionize was designed including demographic, occupational, perceptual and attitudinal variables. The model was tested with a Gallup telephone survey of 495 nonunionized workers of the private service sector in Montreal and Toronto. The propensity of these workers to unionize was measured on a 0 to 100 point scale. Close to 40 % of the respondents surveyed would like to be unionized. Regressing the propensity to unionize on this scale with ordinary least squares gives a percentage change of propensity to unionize by unit change of the independent variables.

Contrary to expectations, even controlling for all other factors, it was found that women are less prone to unionize than men in this sector. Also contrary to expectations, workers in small establishments of the private service sector are more prone to unionize than those working in larger establishments. As expected, other demographic and occupational characteristics have no effect on propensity to unionize.

The Gallup survey confirmed the well known negative relation between job satisfaction and propensity to unionize, and the equally well known negative relations between a big union image and a perception of union as antidemocratic bodies, and propensity to unionize. A positive relation between the perceived instrumentality of unions and propensity to unionize was also confirmed. Even if workers in the private service sector see unions as big organizations and antidemocratic bodies they, like other workers, still see unions as instrumental in providing better working conditions.

One main contribution of this paper is the identification of a very strong positive relation between the cohesiveness of a worker's work group, the attitudes of the work group towards unions, and the individual propensity to unionize. The survey also confirmed an even larger effect of "significant others" with whom one lives and the community in which she or he lives and propensity to unionize. The socialization of attitudes towards unions and the influence of the work group's attitudes towards unions and its cohesiveness are the most important factors shaping one's propensity to unionize. A feeling of solidarity with workers in general is also instrumental in improving the propensity to unionize. On the contrary, a greater identification with the employer will decrease the desire to be unionized.

The paper also demonstrates the importance of the substitution effect of employment laws on the desire of private service workers to be unionized. Finally, the Gallup survey could not establish any link between perceived employer resistance to unionization and the propensity to unionize.

There are many strategic implications of those findings for the actors in the System of industrial relations. Among them, employers may be more succesfull in opposing unions with human resource policies aimed at enhancing job satisfaction than with direct opposition to union drives. Govemments, if aware of the substitution effect of employment laws on unions, will have to assess more closely the role that union may have to play in democratization of the workplace in the private service sector. Unions will have to pay more attention to women's needs in the private service sector. Workers in smaller firms should also be targeted because they are more prone to unionize. The union movement will also have to try to organize in communities where the socialization towards unions is more positive.