2020
DOI: 10.21031/epod.602765
|View full text |Cite
|
Sign up to set email alerts
|

The Importance of Sample Weights and Plausible Values in Large-Scale Assessments

Abstract: Uluslararası Öğrenci Değerlendirme Programı (PISA), Yetişkin Becerileri Araştırması (PIAAC), Uluslararası Matematik ve Fen Bilimleri Eğilimi Araştırması (TIMSS) gibi geniş ölçekli testler, birincil amacı olan ölçme, değerlendirme ve gelişimi izlemenin yanı sıra, ülkelerin eğitim politikalarını belirleyici anahtar bir rol de üstlenmektedir. Makro boyutta politika değişikleriyle milyonlarca paydaşı etkileme potansiyeline sahip olan geniş ölçekli test verilerinin bilimsel olarak doğru istatistiksel yöntem ve tekn… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1

Citation Types

0
7
0

Year Published

2021
2021
2024
2024

Publication Types

Select...
8

Relationship

0
8

Authors

Journals

citations
Cited by 17 publications
(7 citation statements)
references
References 20 publications
0
7
0
Order By: Relevance
“…Teacher-level variables are shown in Table 2. Also, Arıkan et al (2020) suggested using sample weights and plausible values in data analysis in large-scale international assessments. Thus, a multilevel structure was taken into account using five plausible values and sample weights.…”
Section: Discussionmentioning
confidence: 99%
See 1 more Smart Citation
“…Teacher-level variables are shown in Table 2. Also, Arıkan et al (2020) suggested using sample weights and plausible values in data analysis in large-scale international assessments. Thus, a multilevel structure was taken into account using five plausible values and sample weights.…”
Section: Discussionmentioning
confidence: 99%
“…For example, teaching limited by students' needs was not correlated with mathematics achievement on TIMSS 2015 in Dinaric region countries such as Albania, Croatia, Kosova, and Serbia (Elezović et al, 2022); but was negatively correlated in Turkey (Sarı et al, 2017). In addition, the studies have shown that there were many differences between schools in Turkey, and therefore multilevel analyzes should be made according to schools (Akyüz-Aru, 2020;Arıkan et al, 2020;Sarı et al, 2017). However, Suna and Özer (2021) pointed out that the difference in achievement between schools decreased partially in TIMSS 2019 compared to other TIMSS assessments.…”
mentioning
confidence: 99%
“…The use of MANCOVA is important for two reasons. First, academic achievement indicators are expressed as possible values in item responses in theory-based large-scale studies such as the PISA and TIMSS, and researchers recommend considering all possible values for unbiased results (Arıkan et al, 2020;Rutkowski et al, 2010). MANCOVA enables simultaneous analysis of all possible values (5 possible values in the TIMSS and 10 possible values in the PISA, respectively) as dependent variables.…”
Section: Data Collection and Analysismentioning
confidence: 99%
“…For this reason, it is recommended to make a multivariate analysis that considers all these values to avoid bias (Wu, 2015). In other words, approaches such as choosing only one of the possible values or reducing it to a single value such as average are not recommended because they may cause information loss and biased results (Arıkan, Özer, Şeker, & Ertaş., 2020;Rutkowski, Gonzalez, Joncas and von Davier., 2010;Tat, Koyuncu, & Gelbal, 2019). In the study, HLM 8 software was used to calculate the intraclass correlation and to perform multilevel regression analysis.…”
Section: Data Collection and Analysismentioning
confidence: 99%
“…In this context, the number or percentage of schools represented in the population by the schools selected as sampling similar to students may also vary. In this manner, sampling weights should be used in order for the unbiased estimates to represent the population (Arıkan et al, 2020;Rutkowski, 2010;Tat, Koyuncu, & Gelbal, 2019). In the TIMSS, both students and schools are weighted and selected by a two-stage sampling methodology (Rutkowski et al, 2010;Mullis et al, 2020).…”
Section: Data Collection and Analysismentioning
confidence: 99%