**5.4 Empirical evidence on differentiation effects**

### **5.4.1 Short-term effects of differentiation on educational achievement**

The above makes it clear that the idea and practice of differentiation depend critically on proving sorting students improves their educational achievement, compared to nondifferentiated education. Thus, the United States and other countries where differentiation exists have seen a large number of research studies on that question. Nevertheless, their results are somewhat ambiguous and do not allow a clear conclusion on the effects of differentiation. In general, it can be stated that few studies have concluded that differentiation has better overall effects than undifferentiated education. Metaanalyses of large numbers of studies have shown that differentiation either has no *overall* effects (Slavin, 1990) or its effects are negative (Hoffer, 1992).

The key empirical question we are facing here is: what achievement would students from differentiated (homogeneous) groups have if they were educated in undifferentiated

In other words, differentiation probably has its winners and losers while it is a zero-sum

Finally, the third and most recent approach analyzes each individual group in heterogeneous schools (i.e., students in academic versus vocational programs etc.) and compares it with the types of students at non-differentiated schools (e.g., Hoffer, 1992; Kerckhoff, 1986). Even this methodological procedure has its issues, especially those related to the way we match students at differentiated schools with students at non-differentiated ones. Thus, results are somewhat ambiguous here as well. However, the results of this type of studies only vary in the *level* of differentiation effects identified. For instance, Hoffer (1992) found that differentiation has a slightly positive effect on students in academic tracks and a strongly negative effect on students in non-academic tracks. According to Kerckhoff (1986), students in academic tracks gain more than one could expect while students enrolled in non-academic tracks lose due to differentiation. Betts & Shkolnik (2000a) identified differentiation effects as well yet they were much smaller than expected by previous studies. The current state of knowledge on the academic effects of differentiation can be summarized in the following way. A great many empirical studies practically rule out the possibility that differentiated education generally improves overall results and helps both high-track and low-track students. If differentiation is given the benefit of doubt, then it has zero effects (as for the overall results as well as the results of particular groups). However, recent empirical evidence rather seems to demonstrate the fact that while differentiation has zero overall effects on average, it provides great academic improvement to high-track students. **Thus, differentiation seems to increase differences in educational achievement which in turn** 

On the other hand, research has also demonstrated that the effects of differentiation depend on the specific ways it is practiced. For instance Gamoran (1992b) opined that students in academic programs are less advantaged at schools where grouping is flexible, rather than permanent. He also found out that more inclusive schools (those with higher proportions of students in academic programs) do better in overall educational attainment. Other researchers have concluded that the effects of differentiation vary by study courses. For example, Slavin's (1990: 480) review of existing research suggests that heterogeneous (non-

Since the proponents of differentiation mainly argue that this organizational design increases the effectiveness of the education process, researchers have focused on rather short-term educational effects, i.e. what students win and lose in terms of their academic ability and knowledge by enrolling in a certain group (see above). Apart from that, researchers have analyzed more long-term educational effects as well as those that are not

It has been demonstrated that students enrolled in academic programs have higher aspirations for further studies (e.g., Vanfossen et al., 1987), are more likely to enroll in tertiary education programs (e.g., Thomas et al., 1979) and have better odds of actually attaining tertiary education (e.g., Alexander et al., 1987). Students of academic programs are less likely to drop out of school (Gamoran & Mare, 1989). Of course, all those effects have

differentiated) education may have *positive* effects for social science courses.

**increases education inequalities** (Riordan, 2003: 189).

**5.4.2 Other effects of differentiation** 

directly related to education.

game.

(heterogeneous) groups? Thus, we want to measure the effects of homogeneous versus heterogeneous education. Before dealing with the effects of differentiation on each individual group, we will briefly mention the methodological obstacles of measuring differentiation effects. The inconsistency of results is likely caused by divergent methodologies. However, a methodological discussion is also highly relevant in analyzing education inequalities at the system level (between schools), and therefore, we will briefly deal with that discussion first.

In principle, there are three different approaches to measuring the effects of differentiation (Betts & Shkolnik, 2000a; Slavin, 1990;)23. In the first approach, research compares students of different education programs (academic, general and vocational) *within* a school and the growth of their educational achievement over time (Gamoran & Mare, 1989). This kind of studies tracks students over time and determines whether the academic achievement of students in different education tracks grows at different speeds. Since students in different tracks are expected to have different characteristics, their achievement is statistically controlled for socioeconomic background, preexisting knowledge and achievement, IQ and many other variables. Even when influences other than differentiation itself are "filtered out" statistically, an overwhelming majority of studies reach a clear conclusion that taking part in academically more demanding tracks accelerates educational achievement, while enrolment in less demanding and prestigious tracks impedes achievement.

The methodology of this type of research has been criticized for several reasons. First and foremost, in spite of the high number of different control variables entered into the regression model, we are unable to control for many relevant variables that play a role and are responsible for the fact that the groups under investigation (i.e., different education tracks) possess highly different characteristics that cannot be measured persuasively (e.g., the level of self-esteem or motivation). Slavin (1990: 489) points out that no *statistical* controls are sufficient or adequate when there are substantial differences between the groups' initial knowledge and skills; instead, this situation will always cause underestimation of predicted achievement of academically successful students, while that of the less academically successful students will be overestimated24.

The second type of research studies (see review by Slavin, 1990) tries to remedy the abovementioned methodological problem by analyzing the *average* educational achievement of schools practicing differentiated versus non-differentiated education. Those studies have usually reached the conclusion that differentiation has little or no effects on overall educational achievement. The methodological issues with this approach are clear: while zero *overall* differentiation effects are determined, we fail to understand the *distribution* of those effects. Yet it is very likely that differentiation has varying effects on students of different tracks; the zero overall effect obscures the fact that students in more academic tracks gain from differentiation and those in less demanding tracks are the victims of it (Hallinan, 1990).

<sup>23</sup> Here we speak of quantitative measurement of differentiation effects. Apart from that, there is qualitative research, especially ethnographic studies (see Gamoran & Berends, 1987 for a review), which

rather focuses on differentiation-related mechanisms and processes (see next section). 24 Here, Slavin implicitly touches the fact that in comparing two or more qualitatively different groups, this difference cannot be easily eliminated statistically (quantitatively).

(heterogeneous) groups? Thus, we want to measure the effects of homogeneous versus heterogeneous education. Before dealing with the effects of differentiation on each individual group, we will briefly mention the methodological obstacles of measuring differentiation effects. The inconsistency of results is likely caused by divergent methodologies. However, a methodological discussion is also highly relevant in analyzing education inequalities at the system level (between schools), and therefore, we will briefly

In principle, there are three different approaches to measuring the effects of differentiation (Betts & Shkolnik, 2000a; Slavin, 1990;)23. In the first approach, research compares students of different education programs (academic, general and vocational) *within* a school and the growth of their educational achievement over time (Gamoran & Mare, 1989). This kind of studies tracks students over time and determines whether the academic achievement of students in different education tracks grows at different speeds. Since students in different tracks are expected to have different characteristics, their achievement is statistically controlled for socioeconomic background, preexisting knowledge and achievement, IQ and many other variables. Even when influences other than differentiation itself are "filtered out" statistically, an overwhelming majority of studies reach a clear conclusion that taking part in academically more demanding tracks accelerates educational achievement, while

The methodology of this type of research has been criticized for several reasons. First and foremost, in spite of the high number of different control variables entered into the regression model, we are unable to control for many relevant variables that play a role and are responsible for the fact that the groups under investigation (i.e., different education tracks) possess highly different characteristics that cannot be measured persuasively (e.g., the level of self-esteem or motivation). Slavin (1990: 489) points out that no *statistical* controls are sufficient or adequate when there are substantial differences between the groups' initial knowledge and skills; instead, this situation will always cause underestimation of predicted achievement of academically successful students, while that of the less academically

The second type of research studies (see review by Slavin, 1990) tries to remedy the abovementioned methodological problem by analyzing the *average* educational achievement of schools practicing differentiated versus non-differentiated education. Those studies have usually reached the conclusion that differentiation has little or no effects on overall educational achievement. The methodological issues with this approach are clear: while zero *overall* differentiation effects are determined, we fail to understand the *distribution* of those effects. Yet it is very likely that differentiation has varying effects on students of different tracks; the zero overall effect obscures the fact that students in more academic tracks gain from differentiation and those in less demanding tracks are the victims of it (Hallinan, 1990).

23 Here we speak of quantitative measurement of differentiation effects. Apart from that, there is qualitative research, especially ethnographic studies (see Gamoran & Berends, 1987 for a review), which rather focuses on differentiation-related mechanisms and processes (see next section). 24 Here, Slavin implicitly touches the fact that in comparing two or more qualitatively different groups,

this difference cannot be easily eliminated statistically (quantitatively).

enrolment in less demanding and prestigious tracks impedes achievement.

deal with that discussion first.

successful students will be overestimated24.

In other words, differentiation probably has its winners and losers while it is a zero-sum game.

Finally, the third and most recent approach analyzes each individual group in heterogeneous schools (i.e., students in academic versus vocational programs etc.) and compares it with the types of students at non-differentiated schools (e.g., Hoffer, 1992; Kerckhoff, 1986). Even this methodological procedure has its issues, especially those related to the way we match students at differentiated schools with students at non-differentiated ones. Thus, results are somewhat ambiguous here as well. However, the results of this type of studies only vary in the *level* of differentiation effects identified. For instance, Hoffer (1992) found that differentiation has a slightly positive effect on students in academic tracks and a strongly negative effect on students in non-academic tracks. According to Kerckhoff (1986), students in academic tracks gain more than one could expect while students enrolled in non-academic tracks lose due to differentiation. Betts & Shkolnik (2000a) identified differentiation effects as well yet they were much smaller than expected by previous studies.

The current state of knowledge on the academic effects of differentiation can be summarized in the following way. A great many empirical studies practically rule out the possibility that differentiated education generally improves overall results and helps both high-track and low-track students. If differentiation is given the benefit of doubt, then it has zero effects (as for the overall results as well as the results of particular groups). However, recent empirical evidence rather seems to demonstrate the fact that while differentiation has zero overall effects on average, it provides great academic improvement to high-track students. **Thus, differentiation seems to increase differences in educational achievement which in turn increases education inequalities** (Riordan, 2003: 189).

On the other hand, research has also demonstrated that the effects of differentiation depend on the specific ways it is practiced. For instance Gamoran (1992b) opined that students in academic programs are less advantaged at schools where grouping is flexible, rather than permanent. He also found out that more inclusive schools (those with higher proportions of students in academic programs) do better in overall educational attainment. Other researchers have concluded that the effects of differentiation vary by study courses. For example, Slavin's (1990: 480) review of existing research suggests that heterogeneous (nondifferentiated) education may have *positive* effects for social science courses.
