Sunday, April 01, 2012

Effectiveness of LINC programs

It's rare to find a study looking at overall program effectiveness. But Citizenship and Immigration Canada has conducted just such a study of their Language Instruction for Newcomers to Canada (LINC) program, and the results are interesting. They look at a variety of elements including costs, intake type, and provision of daycare. What interested me most, though, was the language learning outcomes, as displayed in the graph below.



The vertical axis is gains by the LINC group, in terms of Canadian Language Benchmark levels, over and beyond gains made by a comparator group (i.e., if both groups had equal gains, the graph would show zero) of immigrants who had not received language training. The horizontal axis is the number of hours that the LINC students spent in class. It wasn't clear to me how they determined the number of hours for the control group, since they didn't go to class, but presumably they looked at the length of time it took for the LINC group to accumulate 500, 750, and 1000 hours and used those same intervals to test the comparator group. Also, since the two groups weren't exactly comparable, observable differences were controlled for using regression analysis.


At the first test point, only reading and writing gains were significantly larger in the LINC group, but as time went on, the LINC group significantly outpaced the controls in all skills.

According to the table below, the difference in apparent trajectories for the four skills seems to be most affected by how well the control group did outside of LINC courses. The LINC students appear to progress at a similar rate across all skills (0.16 benchmarks between listening, which increased the least and reading, which increased most), while the comparison group were much more uneven (0.45 benchmarks between reading, which increased the least, and speaking, which increased the most). It seems that [edit: outside of a LINC course] speaking is [edit: the skill that is] most readily improved outside of a LINC course. Note that the table's numbers have not been subjected to the regression analysis, which is why they deviate from the graph above [edit: and that they correspond only to the point on the graph labeled "1 to 499"].


Table 5‑3: Comparing mean difference scores (uncontrolled)
Current assessment – original assessment
Language AreaLINCComparisonDifference
Listening1.050.920.13
Speaking1.141.130.01
Reading1.210.680.53
Writing1.120.810.31


(Found through a Linked In group posting about this video)

No comments: