Appendix F - Indices contruction

View the most recent version.

Note on the construction of the language use indices

The indices used in this section were developed in the following manner. Whether it comes to the use of English or French, in all cases where people use only the language of the majority, they are assigned a value of 1, while those who use only the language of the minority are assigned a value of 5. In general, the large majority of questions consisted of a scale with five levels. For certain questions, there were only three possible categories: "French", "English and French", and "English". The median category therefore corresponds to a value of 3.

Two approaches were used to present the results covering each of the indices: one on language categories, another by average level of use of languages. The second approach consisted essentially of summing the values obtained from each of the questions and dividing this sum by the number of questions answered by the respondent. For the first type, the average scores obtained were regrouped into five categories. However, given that the results obtained consisted of a continuous scale from 1 to 5, we identified thresholds which allowed us to redistribute the values among the following five language categories: "Only the majority language", "Mainly the majority language", "Both language equally", "Mainly the minority language", "Only the minority language". The corresponding thresholds for these categories are as follows: "1 to 1.49", "1.50 to 2.49", "2.50 to 3.49", "3.50 to 4.49" and "4.50 to 5.0". Since no perfect solution exists to carry out such a distribution, this approach has the advantage of centering the range of values on either side of the median category and cutting down the values at the extremities of the scale.

Finally, the general language use index was drawn up not by using the average of values for all 23 variables, but by adding the value obtained for the question on languages used with friends to the average values obtained on each of the four following indices: languages at work, immediate contacts, institutions and media. Such an approach offers the advantage of not assigning too much importance to a domain made up of numerous questions to the detriment of another composed of fewer questions.

Lastly it should be noted that information covering language used at home was used in the same manner as that regarding friends. By using both the information on language spoken most often and language spoken regularly, the variable created is made up of the same five categories explained above.