Teaching about nature of science through short lab activities in Hong Kong classroom

Students' views in the pre-test and post-test are compared using Wilcoxon rank test since they were repeated measures of the same students. The levels of scientific theory as truth as shown in items 1-3: absolute truth, tentative, and never absolute truth, are ranked 1,2,3 respectively. The levels of understanding of underdetermination as shown in items f, d, e: contextual, weak and strong, are also ranked 1,2, 3 respectively, while 0 is given to those not choosing any underdetermination option. The higher score thus represents a more constructivist, anti-realist view that regards scientific theory not as truth and underdetermined by its evidence.

For the lab class, there found significant growth in views toward scientific theory being never true, z=-2.53, p=0.011, and toward strong underdetermination, z=-3.79, p=0.000 (Table 3). The effect sizes r (z/√N) are 0.47 and 0.70 respectively, a medium to high size effect. As for the control class, there was also significant change in views toward strong underdetermination, z=-2.48, p=0.013, with an effect size of 0.43. The results indicate that both the manipulated lab inquiry and the case of dinosaur extinction can effectively foster students’ understanding about underdetermination. To find out if students' views on scientific theory as truth are related to their views on underdetermination, Kendall rank correlation was calculated in posttest. Only in the dinosaur class the correlation is close to be significant, Kendall's tau= 0.30, p=0.065.

The students held various reasons for why scientific theory does or does not represent absolute truth (Table 4). In the lab class, there were substantial changes in views from pre-test to post-test: none believing in absolute proof and much more believing in inadequate evidence and underdetermination as the causes of scientific theory being tentative. Error, the main reason that students chose in the pretest, became less important in the posttest after the students were aware of the uncertainty between data and conclusion. Much more students chose inadequate evidence because underdetermination partly arises from the lack of evidence to discriminate between alternative hypotheses. In the dinosaur class, however, the views of pre-test and post-test showed no great differences because the students were already holding more constructivist views in the pre-test. It seems that students’ NOS views are context-dependent, with more constructivist views regarding authentic science than school science.

Table 3. Comparisons of students' views between pre-test and post-test using Wilcoxon Sign rank test

	Change from pre-test to post-test
Class	Scientific theory as truth	Underdetermination
Lab class (n=29)	+0.27* (1.83)	+1.1* (1.52)
Dinosaur class (n=34)	0 (2.10)	+0.47* (1.44)

*The difference between pre-test and post-test is significant, p<0.05
Figures in blankets are pre-test scores.

Table 4. Reasons of students in support of their views on scientific theory as truth

Class	Test	Absolute proof	Error	Inadequate evidence	Underdetermination
Lab class (n=29)	Pre-test	17.2%	75.9%	44.8%	72.4%
	Post-test	0%	65.5%	69%	96.6%
Dinosaur class (n=34)	Pre-test	5.9%	70.6%	64.7%	94.1%
	Post-test	2.9%	73.5%	67.6%	94.1%

Some of the students’ responses in class could show how the lab experience changed their NOS views:

S1: In the past I used to think science experiment gives certain answers, but now I know it is not!
S2: I never think of a different conclusion can be made from the same experiment!
S3: This experiment confused me – is science reliable or not?
S4: Can I ask if all true scientific experiments are like that? This really changed my view about science.