A study on the evaluation of science projects of primary school students based on scientific criteria

The research universe consists of the science projects participating in the Science Projects Event for Primary School Students that was jointly conducted by MNE (The MNE) and TUBITAK (The Scientific and Technological Research Council of Turkey). The research sample is composed of 454 science projects applying to the above-mentioned event from Bursa region between 2007 and 2011.

The Evaluation Chart for Science Projects Event for Primary School Students was used for evaluating the project reports prepared by students. However, expert opinions were taken and some changes were made in the chart used for evaluation within the scope of the event (competition) while preparing the new chart. Used as an evaluation instrument, this chart consists of 8 main items and 23 sub-items. 8 main items are as follows: The Determination of Problem (DP), Originality and Creativity (OC), Scientific Method (SM), Consistency and Contribution (CC), Usefulness (U), Implementability (I), Literature Review (LR), and Result (R). The items in the chart were answered with the following responses: “Yes” (2), “Partly” (1), “No” (0) (Table 10).

The data obtained through the evaluation of 454 science projects were analyzed by means of SPSS 18.00. In addition, f and % values were calculated in regard to qualitative data.

The expert opinions and recommendations were taken concerning the Evaluation Chart for Science Projects Event for Primary School Students in order to ensure scope validity of the scale.

Kolmogorov-Smirnov test demonstrated that data did not display a normal distribution. In the event that two researchers come up with different scores concerning the evaluation of a formation, Kendall’s tau-c is used for determining the correspondence of evaluation scores. Kendall’s tau-c coefficient is a statistical element that tests the correspondence between asymmetrical peers in ordinal scale, near scale, or interval scale data (Özdamar, 2011). Thus, Kendall’s tau-c coefficient was used for interpreting the research data.

The meanings of Kendall’s tau-c coefficients are as follows:
>0.50               : High-level correlation,
0.36-0.49         : A Significant correlation,
0.20-0.35         : Intermediate level correlation,
0.10-0.19         : Low-level correlation,
< 0.10              : No correlation.

Fit values and percentages of the scores given by two researchers in re-evaluations were determined. Fit index (FI) was calculated as follows:

FI= ((Total number of correspondences) / (Total number of evaluations)) x 100

Total number of evaluations= the number of corresponding evaluations + the number of non-corresponding evaluations

Table 1. The percentages of correspondence between expert evaluators

Criteria	N	The number of correspondence	Correspondence %
a) The determination of problem	454	283	62.3
b) Originality and creativity	454	364	80.2
c) Scientific method	454	347	76.4
d) Consistency and contribution	454	389	85.7
e) Usefulness	454	403	88.8
f) Implementability	454	430	94.7
g) Literature review	454	388	85.5
h) Result	454	354	78.0

FI needs to be over 75% for inter-experts evaluation results that are to be considered reliable. A lower ratio means that observers think differently to a considerable extent. In the table 1 , correspondence values of two experts are given, and correspondence percentages and averages are separately calculated for each project. Accordingly, correspondence percentages vary between 62.3% and 94.7%. According to the table 1, correspondence ratio is over 75% in all criteria except for The Determination of Problem. In other words, there is a high-level correspondence between observers in 7 criteria.

Document review method was employed in the present study where documents were subjected to analysis. The projects were accessed via the “Bu Benim Eserim Matematik ve Fen Bilimleri Proje Yarışması (This Is My Work: Mathematics and Science Project Competition)” (http://tegm.meb.gov.tr/bubenimeserim/) page located on the homepage of the official website of the MNE (www.meb.gov.tr). The projects were separately evaluated by two different experts in accordance with the predetermined criteria. The obtained qualitative data were quantified through content analysis. The data obtained by expert evaluators were subjected to an analysis of normality. The related results are demonstrated in the Table 2.

Table 2. The analysis of normality of the data obtained by expert evaluators

Criteria	N		SD	Kolmogorov-Smirnov Z	p
a) The determination of problem	454	0.551	0.384	0.107	0.00*
b) Originality and creativity	454	0.412	0.515	0.292	0.00*
c) Scientific method	454	0.315	0.375	0.201	0.00*
d) Consistency and contribution	454	0.403	0.308	0.323	0.00*
e) Usefulness	454	0.189	0.306	0.349	0.00*
f) Implementability	454	0.189	0.384	0.478	0.00*
g) Literature review	454	0.266	0.289	0.216	0.00*
h) Result	454	1.117	0.404	0.227	0.00*
*p<0.05

According to the Table 2, the data obtained by expert evaluators do not display a normal distribution.

Table 3. The results of inter-evaluators kendall's tau-c analysis

Criteria	N	Kendall’s tau-c (τ_c)	P
a) The determination of problem	454	0.678**	0.00
b) Originality and creativity	454	0.612**	0.00
c) Scientific method	454	0.708**	0.00
d) Consistency and contribution	454	0.579**	0.00
e) Usefulness	454	0.517**	0.00
f) Implementability	454	0.380*	0.00
g) Literature review	454	0.670**	0.00
h) Result	454	0.577**	0.00
* 0.36-0.49 ** >0.50

According to the table 3, there is a high correspondence between the scores of expert evaluators pertaining to the criteria of The Determination of Problem (τ_c= 0.678). Originality and Creativity (τ_c= 0.612), Scientific Method (τ_c= 0.708), Consistency and Contribution (τ_c= 0.579), Usefulness (τ_c= 0.517), Literature Review ( τ_c= 0.670), and Result (τ_c= 0.577), and there is a significant correspondence between the scores of expert evaluators pertaining to the criterion of Implement ability (τ_c= 0.380).