Keywords

7.1 Introduction

One of the most useful and reliable results from international large-scale assessment (ILSA) is the changes in student achievement across time within countries. While comparisons across countries can sometimes be problematic due to cultural differences and variations between the educational systems, comparisons within countries over time can be very useful for policy and practice. Knowledge of why achievement changes over time can be as informative and useful as making cross-national comparisons at any point in time. It is, however, difficult to provide evidence of the reasons behind changes in achievement. Most ILSAs are cross-sectional, and hence causal inferences based on these data are limited. However, there are so-called causal methodologies that provide more robust and reliable results and that increase the plausibility for causal inferences (Gustafsson & Nilsen, 2022; Nilsen et al., 2022). The present study utilizes such an approach to investigate plausible reasons behind the changes in science and mathematics achievement in the Nordic countries over the last three cycles (2011, 2015, and 2019) of the Trends in Mathematics and Science Study (TIMSS).

Several factors may have caused changes in student achievement. From previous research, we know that teachers are the most proximal to students, and key to student learning outcomes (Baumert et al., 2010; Charalambous & Praetorius, 2020; Darling-Hammond, 2000; Nilsen & Gustafsson, 2016). Especially, teacher practices are known to shape student learning, well-being, and motivation (Fauth et al., 2014; Senden et al., 2022). Teacher practices are part of a broad concept that may include teaching quality and homework practices. Teaching quality is defined as the aspects of teaching, that previous research has found, that promote learning, and refers to what teachers do in the classroom (Praetorius et al., 2018). Homework practices may refer to how often and how long (number of minutes) the homework assigned to students is, the type of homework assigned, and what the teachers do with the homework, such as discussing the homework in class or correcting the homework (Fernández-Alonso & Muñiz, 2021). Previous research on the effect of homework practices on student achievement is mixed; some research finds that homework promotes learning and others don’t (Fan et al., 2017; Gustafsson, 2013; Trautwein, 2007). This can often depend on the conceptualizations and methodologies used in the research.

The overarching aim of the present study is to investigate whether changes in teacher practices are related to changes in student science and mathematics achievement from 2011 to 2019 using TIMSS data.

7.2 Theoretical Framework

In this section we define the concepts included in this study based on previous research and describe what previous research has found in terms of relations between teacher practices and student learning outcomes.

This study defines teacher practices to encompass teaching quality, homework, and assessment practices, as detailed in Chap. 2. However, assessment practices are only included in one of the three cycles and are hence excluded from this chapter.

7.2.1 Teaching Quality

There are several frameworks describing teaching quality, and the concept itself has many names, such as instructional quality, teaching quality, and teacher practices. This is problematic and has hence been a topic of debate recently (Charalambous & Praetorius, 2020; Senden et al., 2022). The frameworks describing teaching quality are different depending on the type of data. For instance, frameworks describing teaching quality in studies using video observations of classrooms may differ from those using questionnaires (Senden et al., 2022). However, the key aspects of teaching quality are very similar. The present study uses the framework of the three basic dimensions (TBD) by Klieme and colleagues (Klieme et al., 2009; Praetorius et al., 2018), as this has been tested and validated in several studies in Europe (Senden et al., 2022, 2023).

Teaching quality refers to the aspects of teaching, more specifically the teachers’ behavior in the classroom, that in previous studies have been found to promote learning outcomes (Baumert et al., 2010; Fauth et al., 2014). The TBD framework includes three main aspects of teaching. Classroom management, referring to maintaining order and discipline in class and time on task, is the strongest predictor of achievement (Marder et al., 2023). TIMSS included this measure in the student questionnaire for the first time in 2019. Teacher support and clarity refers to both social-emotional support and support for learning, as well as clear and understandable teaching (Praetorius et al., 2018). This second aspect of teaching has been found to be a stronger predictor of student motivation than student achievement (Fauth et al., 2014; Nilsen et al., 2018). The third aspect of teaching quality is cognitive activation, which refers to challenging students to go beyond what they have learned, to promote reflection, reasoning, and critical thinking. In mathematics, problem-solving is an example of cognitive activation. In science, inquiry is a typical approach that would challenge students cognitively, for instance, by making students plan and interpret findings from an experiment (Teig et al., 2018, 2021). This aspect has been found to promote student learning, although findings are mixed as it is hard to measure (Baumert et al., 2010; Charalambous & Praetorius, 2020).

The TBD framework is a generic framework that works well across subject domains, and all three aspects have been found to be relevant in both mathematics and science (Fauth et al., 2014; Praetorius et al., 2018). However, cognitive activation is more content specific than the other two (Praetorius et al., 2018).

7.2.2 Interaction Between the Teaching and the Students—Limitations to Teaching Quality

The teaching that goes on in the classroom is not a one-way street; the quality of teaching depends not just on the teacher, but also on the class. The classroom composition matters. For instance, a classroom dominated by students who lack prior knowledge, who do not speak the language well, or who are hungry or sleepy, would limit high quality teaching (Kaarstein & Nilsen, 2021). Even if the teacher is competent and usually provides high quality teaching, such limitations would be challenging. Hence, considering such limitations when examining teaching quality is necessary.

7.2.3 Homework

Findings from research on the effects of homework are mixed (Fernández-Alonso & Muñiz, 2021; Gustafsson, 2013; Trautwein, 2007). This may be explained by several factors, such as the type of homework. For instance, in mathematics, teachers may provide homework that includes repetitions of tasks the students have done at school, or they may provide cognitively challenging tasks that would require reasoning or problem-solving. Further, the teacher may provide homework often, but could assign tasks that may be done quickly by students, or they could assign homework more seldom but that could take the students a long time to complete. Low-achieving students, and/or students from homes of low socioeconomic status (SES), usually need longer time to complete a homework task than high-SES students (Gustafsson, 2013). The findings may also depend on the operationalization of homework, the sampling design, the level of analyses, and the methods of analyses. For instance, using data with a cross-sectional design, operationalizing homework by the number of minutes students spend on homework, and analyzing the effect of homework on achievement at the student level, may produce negative regression coefficients. Low achieving students often use more time on homework, and hence, a longitudinal study would be needed to control prior achievement to enable reliable results. However, with data aggregated to the class-level, the impact of characteristics of individual students is reduced, and the influence of teacher-assigned homework may be expected to be more pronounced (Gustafsson, 2013).

7.3 Research Questions

The overarching aim of the present study is to investigate whether changes in teacher practices are related to changes in science and mathematics achievement from 2011 to 2019. Seeing how Norway changed the target grade in 2015 from grade four to grade five, the aim for Norway differs somewhat, and we investigate whether changes in teacher practices are related to changes in science and mathematics achievement over time (from 2011 to 2015) and/or changes in achievement from grade four to five. We address this aim more specifically through the following research questions:

  1. 1.

    How have teacher practices changed from 2011 to 2019?

  2. 2.

    Are changes in teachers’ practices related to changes in achievement? In Sweden, Denmark, and Finland the changes in achievement refer to changes from 2011 to 2019, while in Norway, the changes in achievement refer to changes from 2011 to 2019, and/or changes from grade four to five.

Achievements have increased in some countries during this time span and decreased for others. For the countries where achievements have increased, this could be related to an increase in one or several of the teachers’ practices. In other words, the increased achievements may be explained by better teaching practices over time. For this to happen, then the increased achievement must be related to increased teacher practices. The same is true for decreased achievements and teacher practices. If teacher practices have not changed, or if they are not related to achievement, they cannot explain any changes in achievement.

7.4 Methodology

7.4.1 Data and Sample

The data used in the present study is achievement and questionnaire data from representative samples of students from Sweden, Norway, Finland, and Denmark as well as their teachers. These data are retrieved from three cycles of TIMSS: 2011, 2015, and 2019. In Sweden, Finland, and Denmark, the students who participated were in grade four. In Norway, the target grade changed in 2015 from fourth to fifth grade, so that the present sample includes Norwegian fourth graders in 2011, and fifth graders in 2015 to 2019. The reason why the target grade changed from grade four to five in Norway, was because the Norwegian students in grade four were one year younger than students in grade four in the other Nordic countries. This is because Norwegian students start school the year they turn six years old, while students from the other Nordic countries start school the year they turn seven years old. Hence, to compare Norwegian results from TIMSS with that of other Nordic countries, Norway changed the target grade to grade five. This means that findings from Norway need a different interpretation. Part of the explanation for the increased achievement in Norway from 2011 to 2019, is that students were older and had one more year of schooling (Olsen & Björnsson, 2018). For instance, if findings show that teacher practices explain part of the changes in achievement, it means that teacher practices may explain changes in achievement across time and/or across grades. For descriptions of TIMSS data, including sampling, plausible values, and weights, please see Chap. 3.

7.4.2 Measures

Some of the measures in the TIMSS questionnaires have changed over time, especially teaching quality. To be able to compare measures across time, only items that have stayed the same over time are included.

Teacher practices

Several variables were used to measure the different aspects of teaching quality. Teacher support and clarity were measured by students’ responses to their perceptions of their teachers (e.g., I know what my teacher expects me to do, my teacher is easy to understand). Cognitive activation in mathematics was measured by teachers’ responses to how often they ask students to do various tasks (e.g., listen to me explain how to solve problems, work problems together in the whole class with direct guidance from me). The same question was asked for science (e.g., observe natural phenomena such as the weather or a plant growing and describe what you see, design or plan experiments or investigations). Classroom management was not introduced before 2019 and was thus excluded from the study.

Limitations to teaching were measured by teachers’ responses to the extent their teaching was limited by various challenges (e.g., students lacking prerequisite knowledge or skills, students suffering from lack of basic nutrition, students suffering from not enough sleep).

There were two measures for homework. The first is related to how teachers use the homework, and this was measured by teachers’ responses to how often they do the following with the homework: correct assignments and give feedback to students, discuss the homework in class, monitor whether or not the homework was completed. In addition, the teachers were asked how often (frequency) and how many minutes of homework they assign. To create an estimate of time spent on mathematics homework, an algorithm was created which multiplied the categories in the teacher questionnaire of frequency and time into an estimate of minutes spent on homework each week by each student (see Gustafsson, 2013, for a description of this procedure). In a parallel fashion an algorithm for estimating time spent on science homework was created.

A relatively large number of teachers reported that homework was never assigned, and particularly so for science. The proportion of teachers reporting any homework in mathematics was for Denmark 72%, Finland 86%, Norway 68%, and Sweden 65%. For science, the proportion of teachers reporting any homework was for Denmark less than 10%, Finland 88%, Norway 34%, and Sweden 33%. Thus, for science only, Finland assigns homework to such an extent that meaningful analyses can be conducted from the data. For mathematics, results must also be cautiously interpreted, given that only about two thirds of the students in Denmark, Norway and Sweden were assigned any homework.

7.4.3 The Analytical Approach

The method of analyses in this study resembles that of longitudinal growth models (Murnane & Willett, 2010) but adjusted to examine trend analyses.

Causality and causal language. The method of analyses implemented in this study utilizes merged data from three cycles. The method is far more robust than analyzing one cycle of an ILSA with cross-sectional data, or by analyzing data from each cycle separately, and then comparing results across time. The present method enhances the plausibility of causal inferences. That being said, the design is not truly longitudinal, as the same students are not followed over time. Yet, taking advantage of the representative samples and the trend-design of TIMSS—where achievements are scaled to enable comparisons across time—the reliability and validity of inferences is enhanced.

Still, this is not a randomized controlled trial, and hence, the inferences are not strictly causal. This means there is a need to examine the findings in light of previous research, and to consider omitted variables that may introduce bias. This further implies that causal language should not be used. However, this book is aimed at education policy, stakeholders, and researchers. To clarify advanced methodologies, terms like effect are used to describe the relationship between predictors and outcomes. Nevertheless, both in this chapter and the conclusions, it is emphasized that the inferences are not based on causal relationships.

The analytical approach. The approach in the present chapter is that of a structural equation model (SEM) with mediation, and we use the software Mplus for the analyses (Muthén & Muthén, 1998–2017). A dummy variable called Time is coded 0 for 2011, 1 for 2015, and 2 for 2019. In a null model we first examine changes in achievement over time as described in Fig. 7.1.

Fig. 7.1
figure 1

The effect of time on achievement using the null model. Note The regression coefficient is c

Full size image

The regression coefficient c reflects how achievement has changed over time and describes the slope of the effect of time on achievement.

The idea is to examine whether there are other factors that may “explain” changes in achievement, besides the passing of time. In this case, we wish to examine whether teacher practices may explain changes in achievement. We do this by examining whether teacher practices may mediate the effect of time on achievement as illustrated in Fig. 7.2.

Fig. 7.2
figure 2

Hypothesized mediation model in which aspects of teachers’ practices mediate the effect of time on achievement

Full size image

If an aspect of teacher practices (e.g., teaching quality) mediates the effect of time on achievement, this means that changes in teacher practices are related to changes in achievement, and it may indicate that changes in teacher practices explain changes in achievement over time.

If teacher practices have improved over time, the regression coefficient a would be positive, and if it is positively related to achievement, it means the regression coefficient b is positive. If both coefficients a and b are statistically significant, then the conditions for mediation are present. Whether or not the mediation is significant, is tested in Mplus through the command Model indirect. This so-called indirect effect is the main focus of the present study, as it tells us whether, and to what extent, teacher practices may mediate (or “explain”) changes in achievement.

Seeing how the aim is to explain changes in students’ achievement over time, and not differences in achievement between classes, the analyses are done at the student level. However, to take into account the hierarchical design of the data where students are nested within classes that are nested within schools, and to avoid under-estimation of standard errors, we use the option in Mplus called “TYPE = COMPLEX” where the data are clustered at the class-level. This way, the analyses take the hierarchical clustering of students into account. With regards to socioeconomic status (SES), it is a fixed effect, as it does change over time and is hence redundant in the analyses.

7.5 Results

In regard to measures of reliability and validity, the model fit was good (i.e., satisfying the cut-off points: the root of mean square error approximation [RMSEA] < 0.05, comparative fit index [CFI] > 0.95), and the factor loadings of the latent variable were between 0.51 and 0.92.

The null model. Before introducing the mediator and predictor of teacher practices, we examined the effect of time on achievement. The results reflect the slope of the relation between time and achievement. If the regression coefficient is positive, it means that achievement increased over time, and a large positive coefficient would reflect a large increase. Table 7.1 shows the results of the null model for science and mathematics.

Table 7.1 Regression coefficients for the effect of time on achievement
Full size table

The findings show an increase in science and mathematics achievement from 2011 to 2019 for Sweden and Norway. For Norway, part of the large increase was because the target grade changed from grade four to five in 2015 (Olsen & Björnsson, 2018). For Finland and Denmark, the achievements decreased in both subject domains. These findings are in line with the international TIMSS report (Mullis et al., 2020). However, Table 7.1 reflects regression coefficients for the slope of changes over two cycles. The actual changes in achievement are available in Chap. 1.

The full model. The results of the analyses examining relations between changes in teacher practices and changes in achievement are provided in Table 7.2 for science and Table 7.3 for mathematics. These results are depicted as text and symbols, the numbers are provided in Appendices 1 and 2.

Table 7.2 How changes in teacher practices are related to changes in science achievement over time
Full size table
Table 7.3 How changes in teacher practices are related to changes in mathematics achievement over time
Full size table

Science. In Table 7.2, the results are provided for the different teacher practices. In the first two rows, we explain in detail the first two aspects of teacher practice, clarity of instruction and cognitive activation, and thereafter we only comment on noteworthy findings. The first aspect of teacher practice, clarity of instruction (reported by students), has a positive and statistically significant relation to science achievement in Norway (small effect) and Finland (medium effect), while findings for Denmark and Sweden were not significant. Students’ perception of clarity of instruction decreased from 2011 to 2019 in all countries, except for Finland. This means that students perceived the instruction as less clear in 2019 than in 2011. To address the main focus of the present study, i.e., whether changes in teacher practices are related to changes in achievement, changes in the clarity of instruction were only significantly related to changes in achievement in Norway. In Norway, the slope of the relation between time and achievement was about 22 points, meaning that achievement increased by 22 points from 2011 to 2019 (where parts of this large increase in achievement is explained by the change of target grade from grade four to grade five). The indirect effect was negative and about one point. This may indicate that Norway could have increased their achievement by 23 points (rather than 22 points) had it not been for the decrease in clarity of instruction. This relation, between changes in clarity of instruction and changes in achievement, could indicate that decreased clarity of instruction may explain changes in achievement over time, and/or changes in achievement from grade four to grade five.

Cognitive activation in science, more specifically inquiry, as reported by teachers, had a significant and positive relation to science achievement in all countries, except for Denmark. However, it only changed significantly over time in Sweden and Norway. It decreased in Sweden and increased in Norway, meaning that teachers reported providing less inquiry-based teaching in 2019 than in 2011 in Sweden. In Norway, it could mean that teachers provided more inquiry-based teaching in 2019 than in 2011, and/or that teachers provided more inquiry-based teaching in grade five than in grade four. In both these countries these changes were significantly related to changes in achievement. The relation between time and achievement in Sweden was about five points, meaning that achievement increased by five points. The indirect effect was negative, and about one point. This may indicate that Sweden could have increased their science achievement by six points (rather than five points), had it not been for the decrease of cognitive activation. In Norway, the indirect effect was positive and about two points. This may indicate that Norway would only have had an increase of 20 points in their science achievement (rather than 22), had it not been for the increase in cognitive activation. This relation between changes in inquiry-based teaching and changes in achievement could indicate that the increase in cognitive activation may explain changes in achievement over time, and/or changes in achievement from grade four to grade five.

Table 7.2 further provides results on limitations to teaching, how homework is used in class, and the time and frequency of homework. For these, there are two noteworthy findings. First, limitations to teaching worsened in all countries, meaning that the teachers’ reported that the following aspects limited instruction more in 2019 than they did in 2011: more students lacking sleep, more students being hungry in school, less students with prior knowledge, and more students disturbing and being less interested in the instruction. This negative change is related to changes in achievement over time. In Sweden, limitations to teaching decreased achievement by seven points, in Norway by three points, in Finland by nine points, and in Denmark by seven points. This means that, for instance, Finland could have turned their decreased achievement, to increased achievement. For Norway, changes in limitations to teaching could explain either changes in achievement over time or over grades, or both. Second, the data show that very little science homework was assigned in grade four. Except for Finland, most students were not assigned any homework at all and for those who do get homework the assignments only require between three and 11 min of work each week. Even for Finland, where 88% of the students were assigned homework, the weekly average spent on science homework was eight minutes, or less than two minutes per day. Such small doses of science homework can hardly be expected to yield any significant effect, because of low statistical power. As expected, there was no significant indirect effect of homework on science achievement. The means of homework for the Nordic countries, and how these change over time, is provided in Fig. 5.3 in Chap. 5.

Mathematics. Table 7.3 provides results for relations between changes in teacher practices and changes in mathematics achievement. As for science, in the first two rows we also explain in detail the first two aspects of teacher practice, clarity of instruction and cognitive activation, and thereafter we only comment on noteworthy findings for mathematics. The first aspect of teacher practice, clarity of instruction (reported by students) has a positive and significant relation to mathematics achievement in all countries. Students’ perception of clarity of instruction decreased from 2011 to 2019 in all countries, except for Finland. For Norway, clarity of instruction could have changed over time or over grades, or both. Changes in clarity of instruction were significantly related to changes in achievement in all countries. In Sweden, Norway, and Denmark the indirect effects were negative, while in Finland it was positive. This means that Norway and Sweden’s inclining achievements may have increased even further, and Denmark’s achievement could have declined less, if clarity of instruction had increased rather than decreased for these three countries. In Finland, on the other hand, clarity of instruction increased, and this may have prevented a further decrease of their achievements.

Cognitive activation in mathematics was not significantly related to mathematics achievements in any of the countries. This is probably due to the low validity of the construct that only includes two items. However, it increased significantly over time in Sweden. In Norway it decreased, either over time or over grades, or both. There were no significant indirect effects, meaning that changes in cognitive activation are not related to changes in achievement.

Table 7.3 further provides results on limitations to teaching, how homework is used in class, and the time and frequency of homework. As for science, there are two noteworthy findings. First, limitations to teaching worsened over time in all countries (in Norway it decreased over time and/or from grade four to five). This means that the teachers’ reported that the following aspects limited instruction more in 2019 than they did in 2011: more students lacking sleep, more students being hungry in school, less students with prior knowledge, and more students disturbing and being less interested in the instruction. This negative change is related to changes in achievement over time (and/or change in grades in Norway). In Sweden, limitations to teaching decreased changes in achievement by six points, in Norway, by three points, in Finland by nine points, and in Denmark by six points. This means that, for instance, Finland could have turned their decreased achievement, to increased achievement. Again, changes in limitation to teaching in Norway, could be related to changes in achievement over time and/or grade.

The amount of homework assigned in mathematics was somewhat larger than in science, and particularly so for Norway and Denmark (24 and 14 min, respectively). The weekly mean was 13 and nine minutes for Finland and Sweden, respectively. No significant indirect effect of homework was found for mathematics, but for Sweden an effect of 1.4 (t = 1.92) was found. However, until this borderline significance has been replicated it should be regarded as a chance effect, given the large number of statistical tests conducted.

7.6 Discussion

7.6.1 Interpretation of Findings and Discussion in Light of Previous Research

Relations to achievement. In both mathematics and science, there is a strong pattern of positive and significant relations between teacher practices and student achievement, except for homework where there were no/few significant findings. The positive, significant relations between clarity of instruction and student achievement in both subject domains are in line with previous research (Fauth et al., 2014; Nilsen et al., 2018). The relation between cognitive activation and achievement was only positive and significant in science but was insignificant in all countries in mathematics. In science, cognitive activation was measured in terms of inquiry, and the positive results have been identified in previous research (Teig et al., 2018). The measure of inquiry has remained the same across all cycles and includes five items. The measure of cognitive activation in mathematics, however, has changed substantially over time; only two items were equal over time (i.e., how often the teacher asks the students to: “listen to me explain how to solve problems,” and “work problems (individually or with peers) with my guidance”). The validity of this measure is hence very low, and we believe this may explain the insignificant findings.

Besides the aspects of teaching quality, there was also a strong pattern of significant and positive relations across all countries and in both subject domains between limitations to teaching and student achievement. This means that students with teachers who perceived less limitations to teaching, perform better. This finding is supported by previous research (Kaarstein & Nilsen, 2021; Vik et al., 2022a).

Changes in teacher practices over time. In regard to changes in clarity of instruction, there is a clear pattern across both subject domains: students perceive their teachers to provide less clear instruction in 2019 than in 2011. The exception is Finland: in mathematics, Finnish students perceive their teachers to provide clearer instruction in 2019 than in 2011, while there are no significant changes in science.

However, there is no clear pattern of findings for cognitive activation, except that in both subjects there are no significant changes in Denmark or Finland. In science, teachers reported implementing inquiry practices more often in 2019 than in 2011 (and/or in grade five than in grade four) in Norway, and less often in Sweden. In mathematics, teachers reported implementing cognitive activation less often in 2019 than in 2011 in Norway (and/or in grade five than in grade four) and more often in Sweden. These opposite results and lack of patterns across mathematics and science are not surprising, due to the aforementioned low validity of the measure of cognitive activation in mathematics.

Findings on limitations to teaching show a clear pattern; a negative change in all countries. This means that teachers reported more limitations to teaching in 2019 than in 2011 (and/or more limitations to teaching in grade five than grade 4 in Norway). The samples in TIMSS are representative, so this could imply that the composition and habits of the populations of students have changed over time. To a large extent, this is backed by previous research; students’ habits in terms of gaming and social media have changed over time, and this could cause lack of sleep (Vik et al., 2022b). Moreover, there was a larger number of minority students who did not speak the language of the test as well as majority students did in 2019 as compared to in 2011 (Mullis et al., 2020). However, more research is needed to investigate how the student population has changed over time, and the consequences of this change.

Changes in teacher practices related to changes in student achievement. Clarity of instruction decreased over the time period for all countries except for Finland. However, if the change is not related to achievement, and if there is no significant indirect effect, this change cannot explain changes in achievement. For science, the indirect effect was only significant for Norway, and the effect was small and negative (one point). Norway’s increased achievement of 22 points (due to changes in time and/or grade), would have been 23 points had it not been for the decrease in clarity of instruction. For mathematics, on the other hand, the changes in clarity of instruction were related to changes in achievement in all countries. This could indicate that clarity of instruction is more important for student achievement in mathematics than in science. This is indeed confirmed by our findings through the stronger relations between clarity of instruction and achievement in mathematics than in science. While clarity of instruction is also important in science (Nilsen et al., 2018), it could be that other aspects of teaching quality are also important in science, such as inquiry-based teaching. Our findings confirm that the aspect of cognitive activation (inquiry), is positively associated with student science achievement, as has been found in previous research (e.g. Teig et al., 2018). This aspect of teaching quality in science only changed over time in Sweden and over time and/or grade in Norway. In Sweden, the increase in achievement would have been one point, had it not been for the decrease in inquiry. In Norway, the increase in inquiry explained two of the 22 points of their increased achievement over time and/or grade.

The analyses of effects of time spent on homework in science and mathematics did not yield any significant effects. However, this lack of findings must be interpreted cautiously, given that only limited amounts of homework were assigned to students, and particularly so for science. The fact that only a few students were actually assigned homework caused the statistical power to be limited, in spite of the relatively large samples of students involved in the study.

The most striking results of our analyses are those of limitations to teaching. As pointed out earlier, less limitations to teaching were positively associated with achievement, but in all countries, the teachers perceived more limitations to teaching in 2019 than in 2011. If it had not been for this negative development of limitations to teaching, Finland and Denmark could have turned their negative developments of achievements to the positive, and in Norway and Sweden, their increased achievements would have been increased even more.

In conclusion, our findings indicate that while teaching quality is important for student achievement, there have been no major changes in teaching quality, and these small changes only explain minor parts of changes in achievement. However, limitations to teaching have changed in all the Nordic countries, and this change is strongly related to changes in achievements. In Norway, more in depth studies of this have been published, where findings point to a negative development over time; students got less sleep and were more often tired in school in 2019 than in 2015 (Kaarstein & Nilsen, 2021; Vik et al., 2022a, 2022b). Moreover, this explained changes in achievement in mathematics and science (ibid). Further, the decrease in how often students spoke Norwegian at home explained the decreased achievements of eighth grade science students from 2015 to 2019 (Lehre & Nilsen, 2021). Students’ conceptual understanding in science decreased during this time period (Lehre et al., 2021). The decreased science achievement in eighth grade was further related to more students being uninterested and interrupting the instruction, and more students lacking prior knowledge (Kaarstein & Nilsen, 2021), as well as a less safe school environment (Nilsen et al., 2022). The changes in the students’ habits and composition were confirmed by previous research that found: a negative development of school environment (Wendelborg et al., 2020); increased challenges related to sleep and nutrition (Vik et al., 2022a, 2022b); and an increase in the number of minority students (Sentralbyrå, 2023). The findings from Norway from 2015 to 2019, support the findings in the present study from 2011 to 2019, and may point to effects of negative developments of limitations to teaching being associated with changes in achievements over time rather than over grade. Moreover, two of the three timepoints (i.e., 2015 and 2019) included students from the same grade.

However, further in-depth studies, preferably with longitudinal designs, are needed to confirm these findings, especially in Denmark, Finland, and Sweden.

7.6.2 Limitations, Reliability, and Validity

The method used in the present study is more robust than analyses of one cycle of TIMSS, and more robust than comparing results from separate analyses of each cycle. Still, this trend study is based on cross-sectional data, and hence no causal inferences can be made. The validity is low for some constructs, due to changes of items in the construct over time. This is especially the case for cognitive activation in mathematics.

Since Norway changed their target grade in 2015, it is hard to interpret whether changes in teacher practices are related to changes in achievement over time or over grades, or both. This weakens the inferences related to Norway. However, as pointed out earlier, previous analyses for Norway from 2015 to 2019 regarding limitations to teaching, support our findings on this in the present study.

7.6.3 Concluding Remarks, Contributions, and Implications

The main aim of the present study was to examine the relationship between changes in teacher practices and changes in achievement over time. The findings indicate that teaching quality matters for students, that clarity of instruction decreased somewhat in all countries except in Finland where it increased, and that these changes are related to changes in achievement. However, changes in teaching quality only explain a minor part of the changes in achievement. The increased limitations on teaching quality do, however, explain a large part of the changes in achievement in all countries and in both subject domains. This finding is a serious one and points to a need for further research into why student composition and students’ habits in terms of sleep and hunger at school have changed in the Nordic countries and what consequences this may have.

This study contributes to the field of research on teacher practices and teaching quality by confirming previous research from other countries (e.g. Baumert et al., 2010; Charalambous & Praetorius, 2020; Darling-Hammond, 2000) in the context of the Nordic countries. It further contributes to educational policy, stakeholders and practice, and points to the need for helping students in terms of their interest in mathematics and science, avoiding hungry students at school, and helping students who lack the necessary language skills. Parents may help their children get sufficient sleep, and researchers may want to further examine the impact of avoiding social media and gaming on sleep deprivation (Vik et al., 2022b).