Mika
Lastusaari
*a,
Eero
Laakkonen
b and
Mari
Murtonen
b
aDepartment of Chemistry, University of Turku, Turku, Finland. E-mail: miklas@utu.fi
bDepartment of Teacher Education, University of Turku, Turku, Finland
First published on 31st May 2019
Changing majors or dropping out are of great concern to universities worldwide, but the role of learning approaches in terms of students’ persistence has not been previously studied. Changing majors, especially in chemistry, is a severe problem in Finland. Here, learning approach data were collected with the ChemApproach questionnaire from 733 bachelor-level students at four Finnish universities. Students intending to change majors showed stronger submissive surface approaches and weaker active deep approaches than those intending to persist in chemistry. The ChemApproach data were complemented with information on actual persistence and first-year grades from a smaller sample from one university (N = 177). A practical deep approach in chemistry studies combined with relatively high grades was shown to be connected to persistence, while a desire to change majors combined with high grades resulted in the actual changing of majors. A high submissive surface approach indicated students at risk of dropping out completely.
In Finland, universities receive financial aid from the government based on their performance, one criterion of which is their output of graduated students. The funding received from the government may then be divided within a university to its faculties, and within faculties to their departments, by using the performance criteria of each unit. This means that if a student starts his or her university studies, e.g., at a department of chemistry, but graduates from medicine, it is the medical department that receives performance points, not the chemistry department. Therefore, a higher dropout rate and greater number of students switching majors will have a clear effect on the money available for a given department in a given faculty at a given university. There are also international effects, since a high university dropout rate can reflect poorly on the perceived global competitiveness of a country as a whole (Ost, 2010). The obvious goal of universities and related institutions is thus to minimise dropout rates. Accordingly, it is essential to find ways of recognising students who are at risk of dropping out as well as determining which factors affect their persistence in studies. With this information, one can begin to discuss whether universities can influence students’ persistence – and if so, with what measures.
Until the end of the 1960s, the primary cause of student drop outs was considered to be shortcomings on the part of students themselves (Habley et al., 2012). In this regard, the first theory was proposed in 1970 by Spady (see e.g., Bean, 1982), who used the notion that shared group values and support from friends reduce suicide rates to model study persistence. Subsequently, several models have been developed to explain the factors that affect persistence or dropping out. The most influential of these models are considered to be those of Tinto (1975) and Bean (1980), both of which have been revised and augmented over subsequent years (e.g., Bean and Eaton, 2002; Tinto, 2010). Other models include e.g., those of Pascarella and Terenzini (1980): “student–faculty informal contact model”, Brower (1992): “life task model” and Altonji (1993): “sequential choice model” as well that of Cabrera et al. (1992a) that combines the models of Tinto and Bean. The earlier models have been reviewed in detail by e.g., Bean (1982), while both older and newer models were reviewed by, e.g., Kuh et al. (2006).
Tinto's (2010) model was influenced by Spady's (see e.g., Bean, 1982) and emphasises the fit between the student and his or her environment. First, the student is influenced by family, schooling and personal attributes, which together compose and direct his or her initial goals and institutional commitments. Second, during university studies, integration with the local academic and social systems refine the student's commitments further. Integration with the academic system is defined by grade performance and intellectual development, while integration with the social system comprises peer-group and faculty interactions. In Tinto's model, these refined commitments serve as the basis for choosing whether to persist or drop out. On the other hand, Bean's (Bean and Eaton, 2002) model stresses attitudes towards the intention to persist. In this model, academic, social–psychological and environmental factors, which are called socialisation and selection factors, affect grades, institutional fit and institutional commitments, thereby influencing one's decision to persist or drop out. However, in Bean's model, environmental factors (i.e., finances, opportunity to transfer and outside friends) may also directly affect one's decision to persist or drop out (Bean, 1985).
Although these models may appear conceptually straightforward, they are actually rather complex in practice. This is because they include a multitude of variables, each of which has the potential to correlate at different magnitudes and to vary in importance. Consequently, different sample groups have indicated different factors as the main contributors to persistence. These main contributors include the following: financing (Cabrera et al., 1992b; Bennett, 2003; Herzog, 2005; Lassibille and Gómez, 2008), first year grades and academic performance (Montmarquette et al., 2001; Wintre and Bowers, 2007; Araque et al., 2009; Ost, 2010; Valadas et al., 2017), major subject (Leppel, 2001; St. John et al., 2004), level of motivation (French et al., 2005) and commitment (Pascarella and Terenzini, 1983; Wintre and Bowers, 2007; Willcoxson et al., 2011), interaction (Pascarella and Terenzini, 1983; Lehmann, 2007; Wintre and Bowers, 2007), interest-major fit (Allen and Robbins, 2008), preparation for university studies (Allen and Robbins, 2008; Lassibille and Gómez, 2008), self-efficacy for educational requirements (Lent et al., 1984), major subject's first year curricula linkage (Lifton et al., 2007), student peers (Ost, 2010), parental support (Wintre and Bowers, 2007), age (Lassibille and Gómez, 2008; Araque et al., 2009), family characteristics (Lehmann, 2007; Lassibille and Gómez, 2008; Araque et al., 2009), quality of teaching (Bennett, 2003) and flexibility of studies (Di Pietro and Cutillo, 2008). As suggested from the small set of studies above, it is not enough to view persistence based only on one perspective, such as sociological, psychological, organisational, economic or cultural. Instead, an integrated viewpoint is needed (Habley et al., 2012). That said, regardless of the perspective employed, common factors are always at (inter)play between (1) who the student was before starting his or her studies and (2) how the studies influenced the student to (3) choose whether to persist or not persist. This interplay ultimately determines whether the student will persist. Here, we call these (1) pre-study factors, (2) during-study influences and (3) actualisation. A visual representation of this three-stage persistence framework in terms of the present work is provided in the Conclusions and implications.
The above-mentioned factors undoubtedly play a very important role in persistence. However, current research on student learning shows that the learning environment and the quality of learning it provokes also play a crucial role in the quality of learning outcomes and study success (Vermunt and Donche, 2017). In Finland, students comprise a homogeneous group in terms of background factors, such as economy, ethnicity and quality of prior education. This in turn presents a good opportunity to focus on the quality of learning.
The ways in which students approach learning can be roughly described by two distant ends, namely meaning- and reproduction-directed. According to a review by Vermunt and Donche (2017), meaning-directed learning is generally positively related to academic performance, while reproduction-directed learning is typically negatively related to academic performance. Research from the Student Approaches to Learning (SAL) tradition dates back to the same time period as research on persistence. During the 1970s, pioneering works were published by four groups led by Entwistle, Biggs, Marton and Pask (see e.g., Beattie et al., 1997). The learning approach theories are based in the dichotomy of deep and surface learning: A deep approach involves the intention to understand, relating ideas and using evidence to form an overall perception; whereas a surface approach involves memorising without understanding, acceptance of not thinking and fragmented knowledge (see e.g., Entwistle and McCune, 2004). A student using the deep approach will generally consider a subject to be learned as meaningful. Moreover, such a student will feel that learning presents a positive challenge and will typically derive excitement and contentment from it. On the other hand, a student using the surface approach will most often lack these positive feelings (Howie and Bagnall, 2013). However, it has been reported that the approach to learning used by a single student is not stable but varies according to time, situation, subject matter and personal circumstances (Vermunt and Vermetten, 2004; Lindblom-Ylänne et al., 2014). One can therefore consider the possibility that common factors influence both persistence and learning approaches. Keeping in mind that research has shown that the deep approach correlates with a good learning outcome and that students can be directed towards the increased use of the deep learning approach (Baeten et al., 2010; Dolmans et al., 2010), it is compelling to think that there could be a connection between learning approaches and persistence and that persistence could be fostered by influencing the learning approach. However, the question of the relationship between learning approaches and persistence has not yet been studied extensively. In fact, to the best of the authors’ knowledge, only one publication exists on this topic. In that work, it was reported that for law students, the features of the surface approach or those of an unorganised deep approach were the most important predictors of whether a master's degree was completed within seven years (Haarala-Muhonen et al., 2017). However, no indication was given as to whether the results were due to delays in studies or to dropping out. The lack of reports on the relationship between learning approaches and persistence may be due to the fact that both may be influenced by a great number of different factors (as discussed above), and thus there may be too many variables to be analysed. Fortunately, Finland has a rather distinct educational setup with a very homogeneous group of students (when concerning, e.g., economy, ethnicity and quality of prior education). Such a high degree of homogeneity reduces the number of potentially influencing factors, making the combined study of persistence and learning approaches more feasible.
As learning approaches have been found to be context-specific (e.g., Vermunt and Donche, 2017), we have developed a questionnaire directed especially at chemistry learning: the ChemApproach questionnaire (Lastusaari et al., 2016). In our earlier study (Lastusaari and Murtonen, 2013), we found that with chemistry students, we could separate different types of deep and surface approaches. What was especially interesting was that those students aiming to persist in chemistry rated high on the active deep approach scale, which emphasised the role practical laboratory exercises play in chemistry learning. This finding is congruent with the results for pharmacy students by Smith et al. (2007), who found that those students who used a strong application-directed learning approach also demonstrated significantly positive academic performance. Another important finding from our previous study that should be investigated further (Lastusaari and Murtonen, 2013) was that surface orientation was correlated with the willingness to change majors. To obtain a more coherent picture of chemistry students’ learning and persistence in studies, we demonstrated that the questionnaire yielded valid results (Lastusaari et al., 2016). In the present work, we inspect more closely the interplay between learning approaches, persistence and grades.
(1) Do students who want to persist in the chemistry major express different kinds of learning approaches than those students who want to change their major?
(2) What kinds of learning approaches are expressed by those students who are successful in changing their major?
(3) Are course grades connected to learning approaches and, further, to changing majors or dropping out?
(4) Can we predict who will change their major, and is it possible to increase the number of students who persist in the chemistry major?
A second dataset was subtracted from the whole dataset to include only bachelor-level chemistry majors from one university. This group (N = 177) comprised 51% females, and 41% of all students in this group wanted to change their major. This dataset was complemented with information on actual persistence (i.e., whether the students had persisted in chemistry, changed their major or dropped out) and first-year grades received through the student register.
All comparisons between the subgroups were analysed using the mean score variables, which were computed based on the factors of the CFA model. The internal consistency of these scales was evaluated using the Cronbach's alpha reliability coefficient. In the group comparisons, the between-within ANOVA was used as an omnibus test for the interaction and main effects. In case of a significant interaction effect, simple main effect analyses were conducted using one-way ANOVA with pairwise comparisons via Tukey's post hoc tests. In case of non-normality or small group sizes, the Kruskal–Wallis non-parametric ANOVA with related Bonferroni-adjusted pairwise comparisons were applied. Besides the significance tests, the effect size was evaluated with the eta squared (η2) or partial eta squared (ηp2) measurements, with the following cutoff values: 0.01 weak, 0.06 medium and 0.14 strong (Cohen, 1988).
Descriptive statistics, analyses of variance and non-parametric analyses were carried out using IBM SPSS Statistics 24.0 software. CFA was performed using the Mplus 7.4 statistical programme (Muthén and Muthén, 2015).
df | χ 2 | p | CFI | TLI | RMSEA | SRMR |
---|---|---|---|---|---|---|
Note: the cutoff values used for accepting a model were CFI and TLI above 0.90, RMSEA and SRMR below 0.08 (Hu and Bentler, 1999; Little, 2010; Wang and Wang, 2012).a The degrees of freedom are different compared to the model with the complete sample (N = 733), because one significant error covariance was added (between the items ActDeep1 ↔ ActDeep4). | ||||||
Complete sample (N = 733) | ||||||
110 | 282.60 | <0.001 | 0.93 | 0.91 | 0.05 | 0.06 |
Subsample1: chemistry majors (N = 343) | ||||||
109a | 201.29 | <0.001 | 0.92 | 0.90 | 0.05 | 0.06 |
Subsample2: one university (N = 177) | ||||||
109a | 143.98 | 0.014 | 0.94 | 0.92 | 0.04 | 0.075 |
Using the confirmed factor structure (see model fit information in Table 1), we computed the corresponding mean score variables and examined whether the desire to change majors would be reflected in these mean scores. The two-way interaction effect between the learning approaches and the subject of the changed major was significant (F(3, 978) = 5.89, p = 0.001, ηp2 = 0.018), and the effect size was small. Thus, simple main effects analyses were conducted for the differences between the ‘change major subject’ groups. The results (Table 2) show that the submissive surface (SubSurf) score was statistically significantly higher for students who intended to change majors. Furthermore, active deep (ActDeep) scores were statistically significantly higher for those students who did not want to change their major than for those who did want to change their major. The η2 values indicate that the effect size was medium (0.06 ≤ η2 < 0.14) for SubSurf and small (0.01 ≤ η2 < 0.06) for ActDeep. We also checked the data for students from majors other than chemistry to determine how the intention to change majors was connected, overall, to learning approaches, but we found no main effect of group (i.e., whether a student intends or does not intend to change majors) across all learning approach dimensions or factors (Appendix 4). This indicates that students who had chemistry as their minor employed the same learning approach features regardless of whether they wanted to change their major. This may be because the ChemApproach data only address the students’ learning approaches towards chemistry, not towards the subject of their major. One can assume that most of the students who had chemistry as a minor chose to take chemistry courses voluntarily. In other words, these students wanted to learn more about chemistry regardless of their commitment to their own majors. Therefore, it seems that the higher submissive surface and lower active deep scores were characteristic of chemistry majors who wanted to change their major. If we think about these results from the point of view of persistence theory (Tinto, 2006; Tinto, 2010), it can be stated that the students with high submissive surface and low active deep scores had low initial commitment toward chemistry. On the other hand, those students who indicated not wanting to change their major scored high in active deep (ActDeep). Of course, wanting to change majors is not the same as actually changing majors. There is always the possibility that one's plans will change or that these plans will not ultimately be realised. According to Tinto's (2010) model, one's goals and commitments will be refined during university study as a result of the degree of success achieved in integrating with the academic and social systems. It is only after this refinement has occurred that the decision about whether to persist is made. For this reason, we next discuss the factors that affect actual persistence.
Intends to change major (N = 117) | Does not intend to change major (N = 226) | F(1, 341) | p | η 2 | |||
---|---|---|---|---|---|---|---|
M | SD | M | SD | ||||
SubSurf | 2.61 | 0.87 | 2.41 | 0.81 | 4.84 | 0.029 | 0.014 |
TecSurf | 3.31 | 0.86 | 3.18 | 0.85 | 1.81 | 0.179 | 0.005 |
ActDeep | 2.71 | 0.87 | 2.97 | 0.86 | 7.57 | 0.006 | 0.021 |
PraDeep | 3.27 | 0.85 | 3.42 | 0.83 | 2.48 | 0.116 | 0.007 |
In the between-within ANOVA, the learning approach × persistence group interaction effect was significant (F-test with Greenhouse and Geisser correction: F(11.40, 458.99) = 3.52, p < 0.001, ηp2 = 0.081), and the effect size was medium. Thus, simple main effects analyses were conducted for the differences between the persistence groups. The results show that submissive surface (SubSurf) and practical deep (PraDeep) have statistically significant differences between the five groups of persistence (Table 3, see Appendix 5 for the post hoc tests). The corresponding η2 values indicate that the effect size was medium (0.06 < η2 < 0.14) for both SubSurf and PraDeep. Because of the small number of students especially in the group of drop outs, we also carried out a nonparametric Kruskal–Wallis H test (Appendix 6), which agreed with ANOVA. In the case of submissive surface, the statistical significance was obtained for the pairs (i) ‘dropped out’ – ‘intended and changed’ (p = 0.002), (ii) ‘dropped out’ – ‘did not intend but changed’ (p = 0.004) and (iii) ‘dropped out’ – ‘did not intend and did not change’ (p = 0.005). In each case, those who dropped out scored highest in submissive surface, suggesting that this approach could be used as an indicator of students at risk of dropping out. It is also worth noting that those who intended but did not change majors did not score statistically differently in submissive surface than those who dropped out. This may be because both of these approaches can be considered as failures in studies and may thus be due to a lack of motivation or commitment or to a bad interest-major fit. All of these are important factors in persistence (Pascarella and Terenzini, 1983; French et al., 2005; Wintre and Bowers, 2007; Allen and Robbins, 2008). Students with insufficient motivation or commitment or unsuitable interest-major fit would thus naturally not persist. However, in the present case, not having enough motivation or commitment to the desired major (instead of chemistry) actually increased persistence with chemistry, even though the interest-major fit would not be optimal. Tying in with the general concepts of persistence models (Kuh et al., 2006; Tinto, 2010; Habley et al., 2012) it is possible that those students who dropped out experienced a negative during-studies influence on their pre-study factors. On the other hand, those who intended to change majors yet persisted were influenced by chemistry studies positively enough to persist regardless of their initial goal and commitment to majoring elsewhere.
Intended and changed (N = 40) | Intended but did not change (N = 38) | Did not intend but changed (N = 16) | Did not intend and did not change (N = 63) | Dropped out (N = 8) | F(4, 160) | p | η 2 | Non-parametric Kruskal–Wallis | |||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
M | SD | M | SD | M | SD | M | SD | M | SD | H(4) | p | ||||
SubSurf | 2.17 | 0.63 | 2.53 | 0.72 | 2.11 | 0.66 | 2.28 | 0.72 | 3.19 | 0.59 | 4.96 | 0.001 | 0.110 | 19.17 | 0.001 |
TecSurf | 3.29 | 0.74 | 3.25 | 0.93 | 3.49 | 0.85 | 3.14 | 0.78 | 2.98 | 1.10 | 0.83 | 0.511 | 0.020 | 2.83 | 0.586 |
ActDeep | 2.77 | 0.71 | 2.78 | 0.80 | 3.25 | 0.95 | 2.98 | 0.79 | 2.41 | 0.63 | 2.25 | 0.066 | 0.054 | 8.85 | 0.003 |
PraDeep | 2.93 | 0.64 | 3.12 | 0.75 | 3.19 | 0.65 | 3.48 | 0.74 | 3.43 | 0.35 | 4.34 | 0.002 | 0.098 | 15.87 | 0.003 |
For practical deep (PraDeep), a statistically significant difference was obtained for the pair ‘intended and changed’ – ‘did not intend and did not change’ (p = 0.001). The result is such that the latter group scored higher than the former. Therefore, a low practical deep score seems to reveal those students who will succeed in changing their major. The group of dropouts also showed an interestingly high practical deep score. It can be speculated that because chemistry is very much an empirical science, the practical deep score can be considered a good indicator of a student's interest in chemistry. One must keep in mind, though, that a low practical deep score would not mean that the student could not have a practical deep approach towards some other discipline. Even though practical deep is not an indicator of commitment as such, it could thus serve as a feature for enhancing the refinement of commitment and goal setting towards persisting with chemistry. Therefore, combining this practical deep result with the submissive surface score of the dropouts, one is tempted to wonder whether the dropout rate could be lowered by influencing the submissive surface features of these students towards a more positive, less submissive approach, one which would enhance commitment and thus lead to persistence.
The results of the one-way ANOVA showed a significant main effect of the persistence group (F(4, 37.27) = 3.23; p = 0.023; η2 = 0.081), and the effect size was medium. The robust Brown–Forsythe ANOVA was used because of the unequal variances between the groups. The pairwise comparisons (Games–Howell post hoc tests) indicated a statistically significant difference in grades between ‘intended and changed’ and ‘intended but did not change’ (p = 0.001; see Appendix 7 for the complete post hoc comparison data). That is, of those students wishing to change their major, those who did change scored clearly higher than those who did not. It is important to note here that it was not the drop outs who had the lowest grades. Therefore, it was not the grades alone that affected persistence in the initially chosen major. This is a good example of academic integration being more than good academic performance. This is very well described with the concept of interest – major fit (Allen and Robbins, 2008). Again, because of the small number of students in the dropout group, we also carried out nonparametric testing. The non-parametric Kruskal–Wallis test gave a similar result (H(4) = 13.63, p = 0.009; please see Appendix 8 for the post hoc data).
Fig. 2 Mean values of the learning approach scale scores and first-year grades for students in the different persistence groups. |
Fig. 3 Mean values of the learning approach scale scores and first-year grades for students in the different persistence groups. |
To examine this more, we conducted a binary logistic regression analysis (Table 4) to determine how learning approaches and grades predict persistence. According to the results, the most powerful factor for explaining persistence in the chemistry major is the practical deep approach. In this analysis, we ignored the students’ self-reported intent to change majors. Thus, on the level of examining all those who persisted or changed majors, grades did not play a central role. However, when combined with students’ own will to change majors, as seen earlier in Fig. 2, grades did play a crucial role, separating those who intended to change majors and were successful in doing so from those who intended to change majors but were not successful in doing so.
Predictor | B | SE | Wald statistic | p | OR |
---|---|---|---|---|---|
Note: B = regression coefficient, SE = standard error, OR = odds ratio. | |||||
SubSurf | 0.60 | 0.29 | 4.21 | 0.040 | 1.82 |
TecSurf | −0.42 | 0.24 | 3.07 | 0.080 | 0.66 |
ActDeep | −0.14 | 0.26 | 0.28 | 0.599 | 0.87 |
PraDeep | 0.99 | 0.29 | 11.48 | 0.001 | 2.68 |
Grade | −0.42 | 0.20 | 4.46 | 0.035 | 0.66 |
According to the results, the practical deep approach to studying was the most powerful indicator for persistence in the chemistry major. Rating high on the submissive surface approach was connected to the willingness to change majors, and it also predicted dropping out totally from university. Grades did not play an important role in persistence when examining all students who persisted or changed majors; but when examining the more detailed groupings, grades did reveal that those who intended to change majors and succeeded in doing so had higher grades than those who intended to change majors but were not successful in doing so. It seems that a pre-study intention to not persist or to persist with chemistry has only a moderate effect on actual persistence, since only one-half of those who intended to change majors were successful in doing so. Also, there was a small proportion of those who did not plan to change majors but ultimately did so. If one outlines the results in the general three-stage framework common to the models of persistence discussed in the Introduction (summarised in Fig. 4), the during-studies influence, i.e., the learning process, is clearly most important. However, because the data were collected by teachers, it may be that some students felt pressure to give a socially desirable reply to the question of intention to change majors. This could be one explanation for why the pre-study factors had no effect on persistence.
Fig. 4 The findings of the present work inserted into the general three-stage framework common to models of persistence. |
The results reveal the great importance of integration to the academic system, i.e., how well the students adhere to the structural rules and requirements of the institution (Habley et al., 2012). This is reflected well in the submissive surface score. To decrease submissiveness (SubSurf score) and, at the same time, promote deep learning, one could engage in more interactive teaching methods, e.g., collaborative problem-solving practices and real working life cases, to better immerse these students in the studied themes. Donche's et al.'s (2013) study showed that teachers who were more learning-focused helped their students to score higher on processing strategies and to better regulate their learning. Similarly, changing teaching methods and the learning-teaching environment to support the development of deep learning approaches could both reduce the willingness to change majors and the likelihood of dropping out.
The persistence of chemistry majors’ interest in chemistry laboratory exercises in terms of the practical deep approach should definitely be fostered on the basis of these results. Application-directed learning has been shown to have a significant and positive relationship with academic performance in a study with pharmacy students (Smith et al., 2007). Thus, the number of practical exercises in education should not be reduced, as many contemporary educational reforms tend to do – on the contrary, they should be increased. Also, other ways to make study content more ‘concrete’ should be considered. Doing so might motivate some students to change their plans to change majors and instead persist in chemistry. This would likely also help surface-oriented students, especially those in danger of dropping out, to develop their learning approaches to be deeper-oriented. One way to develop teaching in this regard could be to incorporate smartphones or tablets. For example, Wijtmans et al. (2014) reported that connecting theory to practice, engaging students with chemistry research and maintaining student attention was improved by using multiple-choice questions, open-ended questions and molecular visualisation with mobile electronic devices as interactive methods during lectures.
It seems that the end result of getting rather high grades is driven by the fact that chemistry is one of the subjects that needs to be mastered in the admission examinations of the departments of medicine. Thus, the students with high grades are, on average, such who strongly strive towards the medical department. It would seem difficult to influence the intentions of these students, because in Finland, medical doctors not only have a higher societal status than chemists but also have considerably higher salaries. However, it may be possible to influence the practical deep score by linking first-year lecture teaching more tangibly with laboratory exercises. In addition, building better linkages between the subject matter taught and working life options could increase interest towards chemistry and thus increase persistence.
The present study shows that the ChemApproach questionnaire can be used to identify chemistry students at risk of not persisting. Thus, the questionnaire may serve as a tool for assessing the need for changes in teaching methods in order to minimise the risk of not persisting with chemistry as a major. Environmental factors have been proven to be influential in shaping students’ learning patterns (Vermunt and Donche, 2017); thus, the teaching–learning environment should be further developed in chemistry to support the practical deep learning approach and to reduce the surface approach. This study also showed that the ChemApproach questionnaire yields reliable data about students’ learning approaches, which were shown to be tightly connected to their persistence in studies. However, it must be noted that the present work was carried out using the Finnish version of the questionnaire and that no testing with the English translations presented in Appendix 1 has been done, yet.
Finally, we must emphasise that although the learning approaches discussed here are a product of a complex mixture of factors that may be difficult to change, evidence does exist that they can be influenced (see e.g., Parpala et al., 2010; Donche et al., 2013). With the ChemApproach questionnaire providing clues about which learning approach features require changes, we are one step closer to improving persistence.
Item | Statement |
---|---|
SubSurf1 | Many things that I learn remain isolated and do not link as a part of a bigger picture. |
SubSurf2 | When reading the course material, I often do not understand how a new topic relates with any old one. |
SubSurf3 | I have to memorize things without having the opportunity to understand them. |
SubSurf4 | During a chemistry lecture, I often do not understand what a new thing is connected with. |
TecSurf1 | I underline while reading for chemistry examination. |
TecSurf2 | I divide the course material to parts, which I learn for the chemistry examination. |
TecSurf3 | When reading for a chemistry examination, I try to make summaries of different unities with my own words |
TecSurf4 | I make my own notes when studying for an examination. |
TecSurf5 | I make mnemonics to learn things better. |
ActDeep1 | After a chemistry lecture, I often chew over the things taught. |
ActDeep2 | I usually search and read additional material concerning the course. |
ActDeep3 | I often chew over the thoughts awoken by scientific texts as well as connections between them. |
ActDeep4 | I look for justifications and evidence to make my own conclusions about things to be learned. |
PraDeep1 | I like to do practicals. |
PraDeep2 | I have often understood a chemical phenomenon only after doing practical work on it. |
PraDeep3 | When doing a practical, I usually try to understand what its different parts are based on. |
PraDeep4 | One can learn a chemical phenomenon only by doing practical work on it. |
University | Annual intake of chemistry majors, n | Years collected | Participating students per year, n | Participation, % |
---|---|---|---|---|
Turku | 63 | 3 | 59 | 94 |
Jyväskylä | 55 | 1 | 17 | 31 |
Eastern Finland | 28 | 1 | 14 | 50 |
Oulu | 53 | 1 | 29 | 55 |
Helsinki | 106 | 0 | 0 | 0 |
Åbo Akademi | 6 | 0 | 0 | 0 |
All participating | 199 | 119 | 60 | |
All together | 311 | 119 | 38 |
Items | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1. SubSurf1 | — | ||||||||||||||||
2. SubSurf2 | 0.51 | — | |||||||||||||||
3. SubSurf3 | 0.50 | 0.52 | — | ||||||||||||||
4. SubSurf4 | 0.42 | 0.59 | 0.44 | — | |||||||||||||
5. TecSurf1 | 0.03 | 0.11 | 0.05 | 0.06 | — | ||||||||||||
6. TecSurf2 | −0.10 | −0.01 | −0.04 | −0.03 | 0.14 | — | |||||||||||
7. TecSurf3 | −0.10 | 0.01 | −0.09 | 0.02 | 0.20 | 0.34 | — | ||||||||||
8. TecSurf4 | −0.01 | 0.08 | 0.03 | 0.05 | 0.41 | 0.26 | 0.38 | — | |||||||||
9. TecSurf5 | 0.04 | 0.01 | 0.01 | −0.04 | 0.21 | 0.19 | 0.20 | 0.33 | — | ||||||||
10. ActDeep1 | −0.14 | −0.14 | −0.10 | −0.12 | 0.02 | 0.15 | 0.22 | 0.15 | 0.09 | — | |||||||
11. ActDeep2 | −0.10 | −0.05 | −0.12 | 0.00 | 0.08 | 0.10 | 0.14 | 0.13 | 0.06 | 0.30 | — | ||||||
12. ActDeep3 | −0.21 | −0.13 | −0.14 | −0.11 | −0.01 | 0.11 | 0.20 | 0.08 | −0.05 | 0.42 | 0.37 | — | |||||
13. ActDeep4 | −0.15 | −0.11 | −0.13 | −0.14 | 0.07 | 0.12 | 0.21 | 0.07 | 0.09 | 0.29 | 0.35 | 0.49 | — | ||||
14. PraDeep1 | −0.22 | −0.16 | −0.20 | −0.18 | 0.04 | 0.10 | 0.09 | 0.04 | 0.08 | 0.17 | 0.09 | 0.17 | 0.12 | — | |||
15. PraDeep2 | 0.04 | 0.11 | 0.08 | 0.08 | 0.13 | 0.16 | 0.12 | 0.18 | 0.18 | 0.11 | 0.11 | 0.06 | 0.06 | 0.34 | — | ||
16. PraDeep3 | −0.24 | −0.21 | −0.24 | −0.24 | 0.01 | 0.23 | 0.24 | 0.13 | 0.14 | 0.31 | 0.27 | 0.33 | 0.30 | 0.42 | 0.32 | — | |
17. PraDeep4 | 0.05 | −0.07 | 0.07 | 0.10 | 0.10 | 0.12 | 0.09 | 0.07 | 0.07 | 0.08 | 0.11 | 0.09 | 0.08 | 0.23 | 0.51 | 0.22 | — |
M | 2.58 | 2.30 | 2.51 | 2.32 | 2.52 | 3.37 | 3.11 | 3.50 | 3.42 | 3.03 | 2.52 | 2.89 | 2.94 | 3.74 | 3.24 | 3.37 | 2.72 |
SD | 0.96 | 0.93 | 1.03 | 0.92 | 1.49 | 1.15 | 1.25 | 1.32 | 1.22 | 1.00 | 1.11 | 1.16 | 1.09 | 1.08 | 1.08 | 1.03 | 1.14 |
Skewness | 0.30 | 0.59 | 0.42 | 0.56 | 0.43 | −0.40 | −0.12 | −0.49 | −0.43 | −0.21 | 0.35 | 0.06 | −0.03 | −0.74 | −0.12 | −0.31 | 0.23 |
Kurtosis | −0.34 | 0.01 | −0.42 | −0.02 | -1.27 | −0.56 | −0.99 | −0.89 | −0.79 | −0.40 | −0.69 | −0.84 | −0.68 | 0.01 | −0.61 | −0.44 | −0.70 |
Mean score variables | 1 | 2 | 3 | 4 |
---|---|---|---|---|
1. SubSsurf | — | |||
2. TecSurf | 0.01 | — | ||
3. ActDeep | −0.21 | 0.21 | — | |
4. PraDeep | −0.12 | 0.25 | 0.29 | — |
M | 2.43 | 3.18 | 2.84 | 3.27 |
SD | 0.75 | 0.83 | 0.79 | 0.77 |
Skewness | 0.41 | −0.24 | −0.15 | −0.28 |
Kurtosis | −0.24 | −0.53 | −0.36 | −0.02 |
Intends to change major (N = 65) | Doesn’t intend to change major (N = 293) | F(1, 355) | p | η 2 | |||
---|---|---|---|---|---|---|---|
M | SD | M | SD | ||||
SubSurf | 2.53 | 0.82 | 2.47 | 0.74 | 0.329 | 0.567 | 0.001 |
TecSurf | 3.29 | 0.82 | 3.11 | 0.86 | 2.343 | 0.127 | 0.007 |
ActDeep | 2.65 | 0.83 | 2.77 | 0.79 | 1.189 | 0.276 | 0.003 |
PraDeep | 3.18 | 0.85 | 3.23 | 0.77 | 0.171 | 0.679 | 0.001 |
Dependent variable | (I) Persistence | (J) Persistence | Mean difference (I–J) | Std. error | Sig. | 95% confidence interval | |
---|---|---|---|---|---|---|---|
Lower bound | Upper bound | ||||||
a The mean difference is significant at the 0.05 level. | |||||||
(a) | |||||||
SubSurf | Intended and changed | Intended but didn't change | −0.36414 | 0.15584 | 0.139 | −0.7941 | 0.0658 |
Didn't intend but changed | 0.05938 | 0.20349 | 0.998 | −0.5021 | 0.6208 | ||
Didn't intend and didn't change | −0.11300 | 0.13908 | 0.926 | −0.4967 | 0.2707 | ||
Dropped out | −1.01875a | 0.26644 | 0.002 | −1.7539 | −0.2836 | ||
Intended but didn't change | Intended and changed | 0.36414 | 0.15584 | 0.139 | −0.0658 | 0.7941 | |
Didn't intend but changed | 0.42352 | 0.20502 | 0.240 | −0.1422 | 0.9892 | ||
Didn't intend and didn't change | 0.25115 | 0.14130 | 0.390 | −0.1387 | 0.6410 | ||
Dropped out | −0.65461 | 0.26760 | 0.109 | −1.3930 | 0.0837 | ||
Didn't intend but changed | Intended and changed | −0.05938 | 0.20349 | 0.998 | −0.6208 | 0.5021 | |
Intended but didn't change | −0.42352 | 0.20502 | 0.240 | −0.9892 | 0.1422 | ||
Didn't intend and didn't change | −0.17237 | 0.19259 | 0.898 | −0.7038 | 0.3590 | ||
Dropped out | −1.07813a | 0.29789 | 0.004 | −1.9000 | −0.2562 | ||
Didn't intend and didn't change | Intended and changed | 0.11300 | 0.13908 | 0.926 | −0.2707 | 0.4967 | |
Intended but didn't change | −0.25115 | 0.14130 | 0.390 | −0.6410 | 0.1387 | ||
Didn't intend but changed | 0.17237 | 0.19259 | 0.898 | −0.3590 | 0.7038 | ||
Dropped out | −0.90575a | 0.25820 | 0.005 | −1.6182 | −0.1933 | ||
Dropped out | Intended and changed | 1.01875a | 0.26644 | 0.002 | 0.2836 | 1.7539 | |
Intended but didn't change | 0.65461 | 0.26760 | 0.109 | −0.0837 | 1.3930 | ||
Didn't intend but changed | 1.07813a | 0.29789 | 0.004 | 0.2562 | 1.9000 | ||
Didn't intend and didn't change | 0.90575a | 0.25820 | 0.005 | 0.1933 | 1.6182 | ||
TecSurf | Intended and changed | Intended but didn't change | 0.03237 | 0.18771 | 1.000 | −0.4855 | 0.5503 |
Didn't intend but changed | −0.20250 | 0.24511 | 0.922 | −0.8788 | 0.4738 | ||
Didn't intend and didn't change | 0.14750 | 0.16702 | 0.903 | −0.3133 | 0.6083 | ||
Dropped out | 0.31000 | 0.32093 | 0.870 | −0.5754 | 1.1954 | ||
Intended but didn't change | Intended and changed | −0.03237 | 0.18771 | 1.000 | −0.5503 | 0.4855 | |
Didn't intend but changed | −0.23487 | 0.24695 | 0.876 | −0.9162 | 0.4464 | ||
Didn't intend and didn't change | 0.11513 | 0.16970 | 0.961 | −0.3531 | 0.5833 | ||
Dropped out | 0.27763 | 0.32233 | 0.911 | −0.6117 | 1.1669 | ||
Didn't intend but changed | Intended and changed | 0.20250 | 0.24511 | 0.922 | −0.4738 | 0.8788 | |
Intended but didn't change | 0.23487 | 0.24695 | 0.876 | −0.4464 | 0.9162 | ||
Didn't intend and didn't change | 0.35000 | 0.23161 | 0.557 | −0.2890 | 0.9890 | ||
Dropped out | 0.51250 | 0.35881 | 0.610 | −0.4774 | 1.5024 | ||
Didn't intend and didn't change | Intended and changed | −0.14750 | 0.16702 | 0.903 | −0.6083 | 0.3133 | |
Intended but didn't change | −0.11513 | 0.16970 | 0.961 | −0.5833 | 0.3531 | ||
Didn't intend but changed | −0.35000 | 0.23161 | 0.557 | −0.9890 | 0.2890 | ||
Dropped out | 0.16250 | 0.31074 | 0.985 | −0.6948 | 1.0198 | ||
Dropped out | Intended and changed | −0.31000 | 0.32093 | 0.870 | −1.1954 | 0.5754 | |
Intended but didn't change | −0.27763 | 0.32233 | 0.911 | −1.1669 | 0.6117 | ||
Didn't intend but changed | −0.51250 | 0.35881 | 0.610 | −1.5024 | 0.4774 | ||
Didn't intend and didn't change | −0.16250 | 0.31074 | 0.985 | −1.0198 | 0.6948 | ||
(b) | |||||||
ActDeep | Intended and changed | Intended but didn't change | −0.00828 | 0.17918 | 1.000 | −0.5027 | 0.4862 |
Didn't intend but changed | −0.48125 | 0.23238 | 0.238 | −1.1225 | 0.1600 | ||
Didn't intend and didn't change | −0.21538 | 0.15882 | 0.657 | −0.6536 | 0.2229 | ||
Dropped out | 0.36250 | 0.30425 | 0.756 | −0.4770 | 1.2020 | ||
Intended but didn't change | Intended and changed | 0.00828 | 0.17918 | 1.000 | −0.4862 | 0.5027 | |
Didn't intend but changed | −0.47297 | 0.23505 | 0.265 | −1.1216 | 0.1756 | ||
Didn't intend and didn't change | −0.20710 | 0.16271 | 0.708 | −0.6561 | 0.2419 | ||
Dropped out | 0.37078 | 0.30630 | 0.745 | −0.4744 | 1.2160 | ||
Didn't intend but changed | Intended and changed | 0.48125 | 0.23238 | 0.238 | −0.1600 | 1.1225 | |
Intended but didn't change | 0.47297 | 0.23505 | 0.265 | −0.1756 | 1.1216 | ||
Didn't intend and didn't change | 0.26587 | 0.21992 | 0.746 | −0.3410 | 0.8727 | ||
Dropped out | 0.84375 | 0.34016 | 0.100 | −0.0949 | 1.7824 | ||
Didn't intend and didn't change | Intended and changed | 0.21538 | 0.15882 | 0.657 | −0.2229 | 0.6536 | |
Intended but didn't change | 0.20710 | 0.16271 | 0.708 | −0.2419 | 0.6561 | ||
Didn't intend but changed | −0.26587 | 0.21992 | 0.746 | −0.8727 | 0.3410 | ||
Dropped out | 0.57788 | 0.29485 | 0.291 | −0.2357 | 1.3915 | ||
Dropped out | Intended and changed | −0.36250 | 0.30425 | 0.756 | −1.2020 | 0.4770 | |
Intended but didn't change | −0.37078 | 0.30630 | 0.745 | −1.2160 | 0.4744 | ||
Didn't intend but changed | −0.84375 | 0.34016 | 0.100 | −1.7824 | 0.0949 | ||
Didn't intend and didn't change | −0.57788 | 0.29485 | 0.291 | −1.3915 | 0.2357 | ||
PraDeep | Intended and changed | Intended but didn't change | −0.18717 | 0.15840 | 0.762 | −0.6242 | 0.2499 |
Didn't intend but changed | −0.25625 | 0.20683 | 0.729 | −0.8269 | 0.3144 | ||
Didn't intend and didn't change | −0.55313a | 0.14093 | 0.001 | −0.9420 | −0.1643 | ||
Dropped out | −0.49732 | 0.28648 | 0.415 | −1.2877 | 0.2931 | ||
Intended but didn't change | Intended and changed | 0.18717 | 0.15840 | 0.762 | −0.2499 | 0.6242 | |
Didn't intend but changed | −0.06908 | 0.20838 | 0.997 | −0.6440 | 0.5059 | ||
Didn't intend and didn't change | −0.36595 | 0.14320 | 0.084 | −0.7611 | 0.0291 | ||
Dropped out | −0.31015 | 0.28760 | 0.817 | −1.1037 | 0.4834 | ||
Didn't intend but changed | Intended and changed | 0.25625 | 0.20683 | 0.729 | −0.3144 | 0.8269 | |
Intended but didn't change | 0.06908 | 0.20838 | 0.997 | −0.5059 | 0.6440 | ||
Didn't intend and didn't change | −0.29688 | 0.19544 | 0.552 | −0.8361 | 0.2424 | ||
Dropped out | −0.24107 | 0.31686 | 0.941 | −1.1153 | 0.6332 | ||
Didn't intend and didn't change | Intended and changed | 0.55313a | 0.14093 | 0.001 | 0.1643 | 0.9420 | |
Intended but didn't change | 0.36595 | 0.14320 | 0.084 | −0.0291 | 0.7611 | ||
Didn't intend but changed | 0.29688 | 0.19544 | 0.552 | −0.2424 | 0.8361 | ||
Dropped out | 0.05580 | 0.27836 | 1.000 | −0.7122 | 0.8238 | ||
Dropped out | Intended and changed | 0.49732 | 0.28648 | 0.415 | −0.2931 | 1.2877 | |
Intended but didn't change | 0.31015 | 0.28760 | 0.817 | −0.4834 | 1.1037 | ||
Didn't intend but changed | 0.24107 | 0.31686 | 0.941 | −0.6332 | 1.1153 | ||
Didn't intend and didn't change | −0.05580 | 0.27836 | 1.000 | −0.8238 | 0.7122 |
Null hypothesis | Sig. | Decision |
---|---|---|
The distribution of SubSurf is the same across categories of Persistence | 0.001 | Reject the hypothesis |
The distribution of TecSurf is the same across categories of Persistence | 0.586 | Retain the null hypothesis |
The distribution of ActDeep is the same across categories of Persistence | 0.065 | Retain the null hypothesis |
The distribution of PraDeep is the same across categories of Persistence | 0.003 | Reject the hypothesis |
(I) Persistence | (J) Persistence | Mean difference (I–J) | Std. error | Sig. | 95% confidence interval | |
---|---|---|---|---|---|---|
Lower bound | Upper bound | |||||
a The mean difference is significant at the 0.05 level. | ||||||
Intended and changed | Intended but didn't change | 0.79a | 0.19 | 0.001 | 0.25 | 1.33 |
Didn't intend but changed | 0.23 | 0.24 | 0.877 | −0.48 | 0.95 | |
Didn't intend and didn't change | 0.33 | 0.19 | 0.423 | −0.20 | 0.86 | |
Dropped out | 0.60 | 0.47 | 0.717 | −1.02 | 2.22 | |
Intended but didn't change | Intended and changed | −0.79a | 0.19 | 0.001 | −1.33 | −0.25 |
Didn't intend but changed | −0.56 | 0.24 | 0.180 | −1.27 | 0.16 | |
Didn't intend and didn't change | −0.046 | 0.19 | 0.117 | −0.99 | 0.07 | |
Dropped out | −0.19 | 0.47 | 0.994 | −1.81 | 1.43 | |
Didn't intend but changed | Intended and changed | −0.23 | 0.24 | 0.877 | −0.95 | 0.48 |
Intended but didn't change | 0.56 | 0.24 | 0.180 | −0.16 | 1.27 | |
Didn't intend and didn't change | 0.10 | 0.24 | 0.994 | −0.61 | 0.80 | |
Dropped out | 0.37 | 0.50 | 0.941 | −1.27 | 2.01 | |
Didn't intend and didn't change | Intended and changed | −0.33 | 0.19 | 0.423 | −0.86 | 0.20 |
Intended but didn't change | 0.46 | 0.19 | 0.117 | −0.07 | 0.99 | |
Didn't intend but changed | −0.10 | 0.24 | 0.994 | −0.80 | 0.61 | |
Dropped out | 0.27 | 0.47 | 0.975 | −1.35 | 1.89 | |
Dropped out | Intended and changed | −0.60 | 0.47 | 0.717 | −2.22 | 1.02 |
Intended but didn't change | 0.19 | 0.47 | 0.994 | −1.43 | 1.81 | |
Didn't intend but changed | −0.37 | 0.50 | 0.941 | −2.01 | 1.27 | |
Didn't intend and didn't change | −0.27 | 0.47 | 0.975 | −1.89 | 1.35 |
Sample 1 – Sample 2 | Test statistic | Std. error | Std. test statistic | Sig. | Adj. sig. |
---|---|---|---|---|---|
Each row tests the null hypothesis that the Sample 1 and Sample 2 distributions are the same. Asymptotic significances (2-sided tests) are displayed. The significance level is 0.05. Significance values have been adjusted by the Bonferroni correction for multiple tests. | |||||
Intended but didn’t change – dropped out | −15.148 | 18.169 | −0.834 | 0.404 | 1.000 |
Intended but didn’t change – didn’t intend and didn’t change | −22.008 | 9.747 | −2.258 | 0.024 | 0.240 |
Intended but didn’t change – didn’t intend but changed | −27.514 | 14.662 | −1.877 | 0.061 | 0.606 |
Intended but didn’t change – intended and changed | 39.011 | 10.731 | 3.635 | 0.000 | 0.003 |
Dropped out – didn’t intend and didn’t change | 6.859 | 17.387 | 0.395 | 0.693 | 1.000 |
Dropped out – didn’t intend but changed | 12.366 | 20.549 | 0.602 | 0.547 | 1.000 |
Dropped out – intended and changed | 23.863 | 17.957 | 1.329 | 0.184 | 1.000 |
Didn’t intend and didn’t change – didn’t intend but changed | 5.507 | 13.680 | 0.403 | 0.687 | 1.000 |
Didn’t intend and didn’t change – intended and changed | 17.003 | 9.345 | 1.819 | 0.069 | 0.688 |
Didn’t intend but changed – intended and changed | 11.496 | 14.398 | 0.799 | 0.425 | 1.000 |
This journal is © The Royal Society of Chemistry 2019 |