Figure 2: Socioeconomic gaps in CIL and CT test scores
CIL
CT
Parental education
Parental occupation
70
70
60
60
50
50
40
40
30
30
20
20
the countries which participated in both the CIL and CT tests and regress CIL and CT test scores on our measures of socioeconomic status (separately for parental education and parental occupation), including country fixed effects in the specification of our model. Our interest lies in the estimated coefficient for the indicator of socioeconomic status. This corresponds to the difference in test scores between the high- status and low-status groups, net of country-specific time- invariant characteristics. The results indicate that, overall, the difference is greater for the CT test than for the CIL test. More precisely, the estimated gap in CT scores between the high-education and low-education groups is 45 points in the case of the CIL test and 52 points in the case of the CT test. These differences are statistically significant at conventional levels. Similarly, the estimated gap in scores between the high occupational status and the low occupational status groups is 40 points on the CIL test and 45 points on the CT test. Again, these differences, while not large in absolute terms, reach statistical significance.
Two main findings emerge clearly from Figure 2. First, regardless of what proxy for socioeconomic status is employed, and in line with expectations, students from more advantaged backgrounds perform better in both CIL and CT tests, compared with their peers from less advantaged backgrounds (i.e., the gaps in test scores, as represented by the dots in Figure 2, are statistically significantly greater than 0 in all the countries). Second, again irrespective of the proxy used for socioeconomic status, the gap in CT test scores tends to be larger than the one in CIL test score (exceptions are Korea and, to a lesser extent, France), although this pattern is more pronounced when looking at parental occupation. Having identified that there are consistent differences in socioeconomic gaps between CT and CIL test scores, the next step is to test whether such differences are overall statistically significant. To that end, we pool all the data from
4
Made with FlippingBook - Online catalogs