L. H.
Montes
a,
R. A.
Ferreira
*a and
C.
Rodríguez
ab
aFacultad de Educación, Universidad Católica de la Santísima Concepción, Concepción, Chile. E-mail: roberto.ferreira.c@gmail.com
bFacultad de Psicología, Universidad de la Laguna, La Laguna, Spain
First published on 15th February 2018
Research into attitudes towards chemistry in Latin America and indeed towards science in general is very limited. The present study aimed to adapt and validate a shortened version of Bauer's Attitude toward the Subject of Chemistry Inventory version 2 (ASCIv2) for use in a Latin American context. It also explored attitudes towards chemistry of Chilean secondary school students, and assessed the effect of school type, year group, gender, and chemistry achievement on both cognitive and affective dimensions. The participants were 523 secondary school students from public, private subsidised, and private schools in Chile. Exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were first carried out to validate ASCIv2. The results of CFA showed that ASCIv2 retained the two-factor structure and showed optimal model fit, but three items had to be removed from the original instrument. The research also showed that attitudes towards science were neither positive nor negative, a reality similar to that of other countries. The results of multivariate and univariate analyses of variance showed significant effects of year group and chemistry achievement on attitudes towards chemistry. No effects of school type, gender or interactions between factors were found. Follow-up analyses revealed that as students advance through school their attitudes decline, but that the higher their chemistry marks, the more positive their attitudes become. These findings are partially in line with previous data from other countries and are a starting point for more research into attitudes towards chemistry in Latin America.
School teachers often state that attitudes play an important role in achieving goals, and that failure in a subject can provoke negative feelings such as apathy and disinterest (Macías, 2010; Kanafiah and Jumadi, 2013). They are not alone in recognising this threat; research has also identified attitudes as a factor that can strongly influence school performance and the choice of university degree (Brown et al., 2014). In fact, there is evidence that students with intentions to pursue a scientific career have a positive attitude towards disciplines such as chemistry (Lyons and Quinn 2010). At the same time, attitudes towards chemistry have been found to predict academic performance in the subject (Brandriet et al., 2011).
While school achievement and attitudes towards science seem to affect each other positively, other factors can have a negative effect on attitudes. For instance, attitudes towards science tend to become less favourable with age, which seems to be related to the content of the science curriculum. While the primary school curriculum focuses on materials such as natural phenomena or the human body, the secondary school content is more abstract and includes concepts that are not observable to the naked eye, and therefore require the use of higher thinking skills (George, 2006; Said et al., 2016). These findings, however, need to be interpreted with caution, as age or year group may interact with other factors. This was the case in a study conducted by Cheung (2009), who investigated the interaction effect between year group and gender in explaining attitudes towards chemistry. He found that male students enjoyed chemistry theory lessons more than their female counterparts during the first two years of secondary school. However, as they progressed into the final two years, their liking for chemistry laboratory work declined, matching that of their female counterparts. In contrast, female students’ enjoyment of chemistry grew steadily across year groups, before sinking again around the final year. Another important factor that can influence beliefs and feelings towards the chemical sciences is the sociocultural context in which students are immersed, for instance, the neighbourhood in which they live, including friends and family relations (Anwar and Bhutta, 2014).
Hence, attitudes seem to be modulated by a number of factors that constitute a more complex construct than expected, and this reveals the need to address them across different educational levels and cultural contexts. In Latin America, and particularly in Chile, there are virtually no studies that have addressed attitudes towards chemistry – or even science in general – and the relation of these attitudes to the above factors. Thus, the main aims of the present work are to validate a scale (ASCIv2) and to assess a number of factors that may affect attitudes towards chemistry of Chilean secondary school students, including gender, school type, year group, and chemistry achievement.
Over the years, a number of instruments to measure attitudes towards chemistry have been developed and validated, including the Cognitive Expectations for Learning Chemistry Survey (CHEMX) (Grove and Bretz, 2007), Colorado Learning Attitudes about Science Survey (CLASS) (Adams et al., 2008), Chemistry Self-Concept Inventory (CSI) (Bauer, 2005), and Attitude towards the Subject of Chemistry Inventory (ASCI) (Bauer, 2008). More recently, Xu and Lewis (2011) devised a new shorter version of ASCI (ASCIv2), which measures intellectual accessibility and emotional satisfaction. This instrument satisfied internal consistency and test–retest reliability, and was further supported by confirmatory factor analysis (CFA). As the authors claim, the two-factor scale represents measures of cognitive and affective components of attitude. The same instrument has been used with other samples at different universities within the United States and abroad, with the aim of obtaining further reliability and validity. For instance, Brandriet et al. (2011) used ASCIv2 to diagnose changes in attitude in first-year chemistry students at two universities in the United States. They found that ASCIv2 showed good internal consistency reliability as measured by Cronbach's alpha, and relatively good validity obtained using CFA. Xu et al. (2012) and Vishnumolakala et al. (2017) used ASCIv2 to measure students’ attitudes towards chemistry, as well as other constructs, among first year undergraduate students in Australia. In these two studies, ASCIv2 showed relatively good or good internal consistency and validity. However, when used outside English-speaking countries, ASCIv2 has not shown the same consistency and validity. For instance, in a study conducted at three universities in Saudi Arabia (Xu et al., 2015), ASCIv2 showed relatively poor internal consistency and validity for the two-factor and 8-item scale. Only after dropping item 6 (chemistry is challenging–unchallenging) did fit indices improve to an acceptable level. Another study that used ASCIv2 was that of Kahveci (2015) on high school students in Turkey. She used exploratory factor analysis (EFA) to adapt and validate the scale, and reported strong internal consistency with Cronbach's alpha values of 0.86 and 0.79 for “intellectual accessibility” and “emotional satisfaction”, respectively. In addition, use of the two-factor structure explained 56% of the variance, which suggests the applicability of using this scale in other populations of secondary school students. One difference to the original structure of ASCIv2 was that one item (chaotic–organized) loaded strongly on “intellectual accessibility” instead of “emotional satisfaction”, as in the original instrument. Since Kahveci only reported EFA indices, it is difficult to evaluate whether the structure obtained under this analysis can be confirmed using CFA, which would be critical for validation. The above results suggest that ASCIv2 is appropriate for use in different contextual settings, but caution must be taken regarding the way in which individuals respond to each of the items, and whether items will load on the same factors. It should therefore be noted that ASCIv2 seems adequate for measuring cognitive and affective attitudes towards chemistry in Anglo-Saxon contexts, but when used elsewhere, some adaptation to the structure may be necessary.
There is only one study that has looked into the cognitive and affective factors of chemistry using ASCIv2. Kahveci (2015) found that secondary school students in Turkey present neither positive nor negative attitudes towards chemistry, and reported an effect of marks obtained in the subject on the cognitive component. In other words, students with higher performance in chemistry showed more positive attitudes towards the subject than students who had low performance. Other variables such as year group, school type, and gender did not produce significant effects on the cognitive dimension. Regarding the affective component, the analyses indicated that “emotional satisfaction” was more positive for students with higher performance in chemistry.
While numerous studies on attitudes towards chemistry have been conducted in developed countries, information on this topic from less developed countries is scarce. For instance, in Latin America there are virtually no studies investigating attitudes towards chemistry, particularly among secondary school students. In Chile, research on attitudes is still incipient, with only one recent study having looked at attitudes towards science in general among secondary school students (Navarro and Förster, 2012). This research used the Test of Science Related Attitudes (Fraser, 1978) to assess attitudes and how they are modulated by scientific literacy and socioeconomic status (SES). The results showed that SES, scientific literacy and attitudes towards science are positively correlated; that is, the higher the SES and the level of scientific literacy, the better the attitudes towards science. However, this research did not address attitudes towards chemistry specifically, or their relation to academic performance; this was only investigated with reference to biology. In addition, the study used a scale of attitudes relating to interests, social implications of science and beliefs about science, factors that are linked only to the affective component. Due to the multidimensional nature of attitude, it is also necessary to examine its cognitive component, “intellectual accessibility”. Likewise, the study in question only examined a sample of students from year 11, so it is important to understand how age or year group affect attitudes towards science, particularly chemistry. As mentioned previously, SES was found to affect the emotional component of attitudes towards science; however, there is no clear reason to think that all sciences are regarded as equal. In fact, it has been found that science is not observed as homogeneous, so inquiring about attitudes towards science in general can partially mask attitudes towards a particular discipline (Spall et al., 2004). This last aspect is important because chemistry in particular is the branch that generates the highest percentage of drop-out and failure in school both nationally and internationally (Lyons, 2006; Porro, 2007; Farías et al., 2016; Flores Ávila, 2016). Going back to SES, Chile has a highly segregated school system that includes public, private subsidised, and fully private schools, which correspond roughly to low, middle, and upper socioeconomic levels. Hence, by studying school type it is possible to gain a deeper understanding of the effect of SES (Bellei, 2000; García-Huidobro and Bellei, 2003).
Intellectual accessibility | Emotional satisfaction |
---|---|
Hard–easy | Uncomfortable–comfortable |
Complicated–simple | Frustrating–satisfactory |
Confusing–clear | Unpleasant–pleasant |
Challenging–unchallenging | Chaotic–organized |
School type | Year group | Total | |||||
---|---|---|---|---|---|---|---|
Year 11 | Year 12 | Year 13 | |||||
Male | Female | Male | Female | Male | Female | ||
Public | 40 | 47 | 33 | 41 | 33 | 40 | 234 |
Private subsidised | 27 | 21 | 39 | 24 | 39 | 28 | 178 |
Private | 20 | 22 | 18 | 18 | 12 | 21 | 111 |
Total | 87 | 90 | 90 | 83 | 84 | 89 | 523 |
Descriptive statistics showed that asymmetry and kurtosis were greater than −1.000 and less than 1.000, which revealed good normality of the item scores. In order to explore the internal structure of ASCIv2, we first used exploratory factor analysis (EFA) with oblimin rotation, given that the cognitive and affective dimensions of ASCIv2 are correlated. We examined the suitability of the data for EFA by looking at correlations between the different items and ensuring that variables were sufficiently inter-correlated. Then we used Bartlett's test to confirm that our population correlation matrix was different from an identity matrix; that is, that our items were inter-correlated or not independent from one another. We also used the Kaiser–Meyer–Olkin (KMO) index, a measure of sampling adequacy, to further assess the suitability of each of the items to be included in EFA. We also checked the determinant of the correlation matrix in order to assess multicollinearity.
We used three criteria in order to decide the number of factors to extract. The first criterion consisted of retaining items that had eigenvalues equal to or greater than 1. Although it is a popular criterion among researchers, it often overestimates the number of factors to be retained, so the scree plot was also used for its accuracy in identifying the factors with eigenvalues in the sharp descent prior to levelling off for samples above 200 participants (Stevens, 2012). Apart from these two criteria reported as preferred by researchers, we also used parallel analysis, as it is the most accurate of all (Henson and Roberts, 2006).
Evidence for the internal consistency of the entire scale and all subscales was assessed using Cronbach's alpha (α) and McDonald's omega (ω_h). Taken on the recommendations made by Taber (2017), we clarify that when high values of alpha are obtained, we assume that this reflects good internal consistency of each subscale, and not simply redundancy given that ASCIv2 has a small number of items. Additionally, high overall values of alpha are regarded as reflecting mainly shared variance within subscales. Additionally, due to the limitations of alpha in that it relies on assumptions that are rarely met, and violation of these assumptions implies inflating or attenuating internal consistency, we also report omega, which performs better than alpha in estimating internal consistency, especially when assumptions of alpha are not met (see Dunn et al., 2014).
Apart from EFA, we additionally conducted confirmatory factor analysis (CFA) on the original structure of ASCIv2 and the new structure obtained using EFA on our data. CFA is more appropriate than EFA in later stages of construct validation, when there is prior evidence supporting specific predictions regarding a latent structure (Brown and Moore, 2012), as is the case in the present study. We tried 5 different CFA models and assessed their fitness to the data by looking at the chi-square (χ2), Comparative Fit Index (Bentler, 1990), Tucker–Lewis index (Tucker and Lewis, 1973), Standardized Root Mean Square Residual (SRMR), and RMSEA (Steiger, 1980). Simulation studies have found that these indices show overall satisfactory performance (Hu and Bentler, 1998). Guided by suggestions provided by Hu and Bentler (1999) and Steiger (2007), acceptable model fit was defined by the following criteria: RMSEA (≤0.07, 90% CI ≤ 0.06), SRMR (≤0.08), CFI (≥0.95), and TLI (≥0.95).
After selecting the model that best fit the data, the estimated factor scores were extracted and their values were added to the main dataset in R to be used for further analysis. A multivariate analysis of variance (MANOVA) was used to assess the effect of year group, school type, gender, and chemistry achievement on both attitude subscales measured by the adapted ASCIv2. Then we also performed separate analyses of variance (ANOVA) and Bonferroni corrected t-tests as follow-up analyses. All statistical analyses were carried out in R (R core Team, 2016). It is worth noting that chemistry achievement corresponded to the mark the students received the term prior to the administration of ASCIv2, and were obtained from official school documents. For ease of analysis, the chemistry mark was divided into quartiles and used as a factor in the inferential analyses.
Item | Emotional satisfaction | Intellectual accessibility | |||
---|---|---|---|---|---|
Adjectives | 8 items | 5 items | 8 items | 5 items | |
Note: items 1, 4, 5, and 7 were recoded for analysis. | |||||
4 | Comfortable–uncomfortable | 0.82 | 0.85 | 0.26 | −0.05 |
5 | Satisfying–frustrating | 0.81 | 0.81 | −0.01 | 0.00 |
7 | Pleasant–unpleasant | 0.73 | 0.66 | 0.04 | 0.12 |
1 | Hard–easy | 0.50 | — | 0.26 | — |
8 | Chaotic–organized | 0.37 | — | 0.04 | — |
2 | Complicated–simple | 0.02 | 0.07 | 0.84 | 0.63 |
3 | Confusing–clear | 0.15 | −0.02 | 0.61 | 0.88 |
6 | Challenging–unchallenging | −0.25 | — | 0.50 | — |
The internal consistency analyses of ASCIv2 with all items and ASCIV2 without 3 items revealed that the shortened version of the instrument had better overall internal consistency and for each of the scales, as shown by overall Cronbach's alpha and McDonald's omega. See Table 4.
Items | Cronbach's alpha | McDonald's omega | |
---|---|---|---|
ASCIv2 | 1, 2, 3, 4, 5, 6, 7, 8 | 0.78 | 0.85 |
2, 3, 4, 5, 7 | 0.78 | 0.85 | |
Intellectual accessibility | 2, 3, 6 | 0.65 | 0.57 |
2, 3 | 0.73 | 0.74 | |
Emotional satisfaction | 1, 4, 5, 7, 8 | 0.80 | 0.85 |
4, 5, 7 | 0.83 | 0.83 |
After running the EFA, we ran two CFA models, one for the original ASCIv2 structure and one for the structure that best fit our data (ASCIv2 Chile). CFA for original ASCIv2 showed poor fit, as shown by the indexes listed in Table 5. The model with the new structure showed slightly better indices than the original one. However, TLI, RMSE and SRMR were still not within the acceptable range. Hence, we tested a new CFA model without item 6, which had a relatively high negative competing loading. The model improved its fit significantly, with CFI and SRMR reaching very good fit; TLI was acceptable, but RMSE still did not meet our acceptance criterion. Finally, we decided to remove item 1 with a relatively high competing loading, and item 8 with a low loading. Hence, we tested a new model with only 5 items, but keeping the two-factor structure of ASCIv2. This model showed optimal fit indices, so we decided to retain these scores for the analyses that follow (see Table 5).
Model description | χ 2 | df | p-Value | CFI | TLI | RMSEA | SRMR |
---|---|---|---|---|---|---|---|
Original ASCIv2 | 210.79 | 19 | <0.001 | 0.85 | 0.78 | 0.11 | 0.14 |
ASCIv2 Chile (8 items) | 128.92 | 19 | <0.001 | 0.92 | 0.88 | 0.11 | 0.10 |
ASCIv2 Chile (7 items) | 52.61 | 8 | <0.001 | 0.96 | 0.93 | 0.10 | 0.05 |
ASCIv2 Chile (5 items) | 14.20 | 4 | <0.01 | 0.99 | 0.97 | 0.07 | 0.03 |
Variable | Wilks’ lambda | df | F | p-Value |
---|---|---|---|---|
Year group | 0.042 | 4 | 4.906 | <0.001 |
Chemistry achievement | 0.121 | 9 | 9.669 | <0.001 |
School type | 0.021 | 4 | 2.347 | 0.053 |
Gender | 0.002 | 2 | 0.556 | 0.573 |
Univariate analyses of variance (ANOVA) were also conducted as follow-up tests to the MANOVA. There was a significant effect of year group on “intellectual accessibility”, F(2, 523) = 9.024, p < 0.001, η2 = 0.034. In order to explore the differences between year groups for “intellectual accessibility”, Bonferroni-corrected t-tests were performed, thus avoiding Type I error. The result showed significant differences between year 11 and year 13 (p < 0.001) and between year 12 and year 13 (p < 0.05). Cognitive attitudes of the students in year 13 were significantly lower (M = 3.46) than those of their counterparts in year 11 (M = 3.80) and year 12 (M = 3.93). The comparison between year 11 and year 12 was not significant (p = 0.5) (see Table 7). The effect of year group was not significant for “emotional satisfaction” (F(2, 523) = 1.715, p = 0.18).
Attitude (subscales) | Variable | Level comparison | df | t | p-Value | r |
---|---|---|---|---|---|---|
Intellectual accessibility | Year group | Year 11 and year 12 | 347 | 0.630 | 0.529 | — |
Year 11 and year 13 | 346 | 3.929 | <0.001 | 0.21 | ||
Year 12 and year 13 | 344 | 3.419 | <0.05 | 0.18 | ||
Chemistry achievement | Quartiles 1 and 2 | 250 | −1.546 | 0.123 | — | |
Quartiles 1 and 3 | 238 | −2.822 | <0.05 | 0.17 | ||
Quartiles 1 and 4 | 250 | −4.653 | <0.001 | 0.28 | ||
Quartiles 2 and 4 | 259 | −3.464 | <0.05 | 0.21 | ||
Quartiles 2 and 3 | 257 | −1.362 | 0.174 | — | ||
Quartiles 3 and 4 | 255 | −2.293 | 0.136 | — | ||
Emotional satisfaction | Chemistry achievement | Quartiles 1 and 2 | 259 | −3.246 | <0.05 | 0.20 |
Quartiles 1 and 3 | 259 | −4.936 | <0.001 | 0.29 | ||
Quartiles 1 and 4 | 259 | −6.797 | <0.001 | 0.39 | ||
Quartiles 2 and 4 | 259 | −3.702 | <0.001 | 0.22 | ||
Quartiles 2 and 3 | 260 | −1.741 | 0.082 | — | ||
Quartiles 3 and 4 | 260 | −1.986 | 0.288 | — |
A significant effect of chemistry achievement was observed on “intellectual accessibility”, F(3, 523) = 8.912, p < 0.001, η2 = 0.049. In order to understand how cognitive attitudes varied between the different categories of chemistry achievement, Bonferroni-corrected t-tests were performed. The result showed significant differences between quartiles 1 and 3 (p < 0.05), between quartiles 1 and 4 (p < 0.001) and between quartiles 2 and 4 (p < 0.05). The comparison between quartile 2 and quartile 3 did not produce significant differences (p = 0.17). There were also no differences between quartile 3 and quartile 4 (p = 0.136), and between quartiles 1 and 2 (p = 0.123).
There was also a significant effect of chemistry achievement on “emotional satisfaction”, F(2, 523) = 13.324, p < 0.001, η2 = 0.091. Bonferroni-corrected t-tests revealed differences between quartile 1 and quartile 2 (p < 0.05), quartile 1 and quartile 3 (p < 0.001), quartile 1 and quartile 4 (p < 0.001), and between quartile 2 and quartile 4 (p < 0.001). Between quartile 2 and quartile 3, there were no significant differences (p = 0.082), neither were there between quartile 3 and quartile 4 (p = 0.288) (see Table 7).
We first used EFA in order to check whether the original structure of ASCIv2 was retained. The original structure was indeed retained, but internal consistency for one subscale was rather poor, and CFA results showed goodness of fit below acceptance levels. After inspecting each of the items, three were dropped from the original instrument because of high competing loading (item 1), loading below 0.40 (item 8), and relatively high competing negative loading (item 6). The final instrument had two factors and 5 items, with goodness of fit indices being optimal. These results show that ASCIv2 can be used in a different cultural setting, but caution needs to be exercised regarding some items. It is worth noting that the only time ASCIv2 had been used to test secondary school students outside the United States, the scale displayed good internal consistency, but one item (“chaotic–organised”) loaded on “intellectual accessibility” instead of “emotional satisfaction”, as in the original structure (Xu and Lewis, 2011). Furthermore, Kahveci (2015) reported good validity, but only used EFA during the validation process, which raises doubts as to whether ASCIv2 was really validated. A critical step during construct validation is the use of CFA, which has the advantage of accounting for measurement error, thus providing a stronger analytical framework than other methods such as EFA (Brown and Moore, 2012). In the present study, we used EFA at an early stage during validation, and results of EFA were very similar to those obtained by Kahveci, with one item also loading on a different factor. However, when CFA was used, this initial structure obtained goodness of fit indices below our acceptance threshold, unlike our final two-factor and 5-item structure, which had good internal consistency overall and across subscales, as well as excellent validity indexes. Given these results, we believe that ASCIv2 in its original structure produces excellent results in culturally similar contexts (e.g., English speaking countries) (e.g., Brandriet et al., 2011; Xu and Lewis, 2011; Xu et al., 2012). However, when used to measure different populations, as in the case of Turkey (Kahveci, 2015), Saudi Arabia (Xu et al., 2015) and now Chile, the results are not stable for a number of reasons. For instance, Kahveci (2015) warned that translating “chaotic” to Turkish was difficult; a similar situation was experienced in Chile with the same word “chaotic”, and the bipolar adjectives “challenging” vs. “unchallenging”. According to our language experts, the latter have perfectly clear equivalent terms in Spanish, but they are not used as often as in English, which might distort participants’ responses. It is worth mentioning that in Saudi Arabia item 6 (“challenging–unchallenging”) was deleted, as it was demonstrated that the students did not understand it properly and consequently responded to it differently compared with students from Australia and the USA (Brandriet et al., 2011; Xu and Lewis, 2011; Xu et al., 2012).
Another important aim of the present study was to assess attitudes towards chemistry of Chilean secondary school students. Since the structure of ASCIv2 with 2 latent variables and 5 items obtained the best psychometric properties, we used this structure and the factor scores obtained with it for conducting inferential analyses. Descriptive statistics showed that Chilean secondary school students have slightly higher affective than cognitive attitude towards chemistry. The average scores for the cognitive and affective dimensions in the present study were 3.80 and 4.18, respectively, which are somewhat higher than those found by Xu and Lewis (2.91 and 3.63, respectively) with first-year chemistry university students. However, they are very similar to those of the study by Kahveci (2015) in Turkey, where the mean score was 3.60 for the cognitive dimension and 3.93 for the affective dimension. They also closely matched those reported by Vishnumolakala et al. (2017), with 3.75 and 4.10 for cognition and affect. Similar patterns were observed across these studies; the affective dimension received a much higher rating across different countries. Overall, these results also suggest that attitudes towards chemistry among Chilean secondary school students are neither positive nor negative, in line with previous studies where the same questionnaire has been used (e.g., Xu and Lewis, 2011; Kahveci, 2015) and even other studies using different attitude scales (e.g., Salta and Tzougraki, 2004). The results presented here are from a convenience sample from a single region in Chile. However, it covers all three types of schools present in the country (public, private subsidised, and private) representing low, middle and upper socioeconomic groups.
We conducted multivariate analysis of variance (MANOVA) and univariate analysis of variance (ANOVA) to assess the effect of school type, year group, gender, and chemistry achievement on attitudes towards chemistry for both subscales of ASCIv2 combined (MANOVA), and for each of the subscales separately (ANOVA). We also explored possible interactions between the factors. The MANOVA showed that only year group and chemistry achievement influenced attitudes towards chemistry; school type and gender did not, nor were there any interactions between the factors. Chemistry achievement also affected each subscale (“intellectual accessibility” and “emotional satisfaction”) separately, whereas year group influenced only “intellectual accessibility”.
In the present study, we found that school type did not affect attitudes towards chemistry. This means that regardless of the school students attend – public, private subsidised or private – they value chemistry equally. Since school type in Chile can be used as an indicator of socioeconomic status (SES) (Bellei, 2000; García-Huidobro and Bellei, 2003), we can extend our findings and argue that SES does not modulate attitudes towards chemistry. To our understanding, this is the first time that the attitudes towards chemistry have been assessed in Chile, so it is impossible to directly compare these results with previous data. Navarro and Förster (2012) investigated the impact of SES on attitudes towards science in Chile, but they only considered biology. Their results show that SES does affect attitudes towards science, with higher scores associated with high SES.
Regarding the gender effect on attitudes towards chemistry, no significant differences were observed between girls and boys, nor did gender interact with other factors. This finding contrasts heavily with most international studies that show divergences in students’ attitudes towards science, with male students holding more favourable attitudes than female students (Cheung, 2009; Lyons and Quinn 2010; Bybee and McCrae, 2011). For instance, Can (2012) reported gender differences between years 10 and 11 on the dimensions of enjoyment and importance of chemistry as a school subject, with boys displaying more favourable attitudes than girls. In view of the above studies, the absence of a gender effect in the present research is quite surprising, although previous studies conducted in Chile have found that girls score higher than boys in adoption of scientific attitudes, as evidenced by Navarro and Förster (2012). This, according to the authors, implies that girls are more willing to accept new ideas and more open to modifying their beliefs about science. The findings by Navarro and Förster – along with those of the present work – are encouraging, as they show that gender stereotypes in relation to the discipline in Chile are not necessarily present. Our results coincide with those of Said et al. (2016) in Qatar, where no differences between boys’ and girls’ attitudes towards science were found.
Year group had an overall effect on attitude towards chemistry, but separate analysis on each subscale showed that the effect was significant only for “intellectual accessibility”, with more positive attitude towards chemistry in the lower years (years 11 and 12) than for year 13. Our findings are in line with previous studies from other parts of the world that show a decreasing trend in students’ attitudes towards chemistry as the students moved to higher year groups (George, 2006; Said et al., 2016). One of the most plausible explanations for this is that the curriculum contents for years 12 and 13 are more complex and require high levels of abstraction, as well as the use of more specific scientific language. Data were collected at the beginning of the academic year, which possibly explains why students in year 12 did not differ from those in year 11, as they had not yet been influenced by year 12 content that includes topics such as oxidation–reduction reactions, chemical equilibrium, and thermodynamics. As for the students in year 13, they are exposed to topics such as nuclear chemistry and polymers. Unlike higher year groups, the curriculum content of year 11 and earlier is related to chemical solutions and organic chemistry such as alkanes, alkenes, alkynes and functional groups. They tend to be easier to associate with everyday situations, allowing the acquisition of meanings in a much simpler way.
Chemistry achievement had a beneficial effect on attitudes across both subscales, combined and separately. The higher the marks in chemistry from the previous term, the higher both the cognitive and affective dimensions of attitudes towards chemistry. This relation was also tested earlier among college students (Brandriet et al., 2011; Xu and Lewis, 2011) and secondary school students using the same scale as in the current work (Kahveci, 2015). These new data, in a different contextual setting, confirm the strong relationship between these two variables.
The aim of the present study was to validate and adapt ASCIv2 to the context of a Latin American country. We devised a shorter version of ASCIv2 that retained the two-factor structure, but with only 5 items. Even though three items were eliminated, CFA showed that this shorter version of the instrument performed much better than the original 8-item instrument, which suggests that some items do not perform well across different cultures, in particular within a Latin American population. We recommend the use of ASCIv2 in different settings, but that caution be exercised with certain items that do not seem to generalise well. After seeking validation of ASCIv2, we explored the attitudes of Chilean secondary school students towards chemistry. We found that attitudes towards chemistry in general were neither positive nor negative, but were somewhat similar to those in other countries. However, affective attitudes were slightly higher than cognitive ones, so it seems that secondary school students have a relatively positive view of chemistry, but consider it to be intellectually difficult or challenging. Attitudes tend to become less favourable as students advance through the year groups, possibly as a result of the introduction of more abstract content which is often far removed from day-to-day activities. This is worrying given the strong association between attitudes towards chemistry and achievement in chemistry, as it means that students’ attitudes might decrease when they are most needed. We propose that chemistry educators develop lessons with varied examples that are applicable to daily life situations, beyond the isolated learning of concepts and principles. The above can be strengthened within an integrated curriculum that includes Biology and Physics since most phenomena do not occur in isolation. Another useful idea would be to conduct demonstration experiments in such a way that students can appreciate the phenomena and changes of the subject directly, thus provoking interest and possibly positive attitudes towards the discipline. Finally, we found no gender effects in our sample, which is encouraging since most studies find that boys have a more positive attitude towards science than girls. Given the importance of attitudes, we would suggest considering them for assessment at a similar level to academic performance in order to obtain much richer information about students.
This journal is © The Royal Society of Chemistry 2018 |