📄 social_science

Teacher Qualification Deficits and Mathematics Achievement in Uzbekistan: Multilevel Evidence from PISA 2022

Using PISA 2022 data from Uzbekistan, this study employs multilevel modeling to demonstrate that while various student and school factors influence mathematics achievement, inadequately qualified teaching staff is the only consistent negative predictor, with its detrimental impact being 3.4 times more severe in rural schools than in urban ones.

Original authors: Indira

Published 2026-07-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Indira

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the education system of Uzbekistan as a massive, bustling school bus ride where 7,293 fifteen-year-old students are trying to solve a giant math puzzle. The bus just arrived at its first international stop, the PISA 2022 assessment, and the results were a bit of a shock. The average student scored 364 points, while the "average" student across the OECD (a club of mostly wealthier nations) scored 472. That's a gap of 108 points—roughly the difference between a student who can barely read a map and one who can navigate a city. Even worse, 81% of these students couldn't even reach the "minimum proficiency" line, meaning they were struggling with the basics.

The big question was: Why is the bus moving so slowly? Is it the students' families? The number of buses (schools)? Or the drivers (teachers)?

To find the answer, the researchers didn't just look at the students; they looked at the whole bus system using a special statistical tool called a "multilevel model." Think of this like a detective who doesn't just ask, "Who is tired?" but also asks, "Which bus is broken?"

The Big Discovery: It's the Drivers, Not the Passengers

The study found that about 18% of the difference in math scores wasn't about the students themselves, but about which school they attended. It's like realizing that if you get on a bus with a flat tire, you're going to arrive late, no matter how fit the passengers are.

When the researchers started testing different reasons for the delay, they ruled out a few common suspects. They looked at whether there were too many students per teacher (the bus was too crowded), whether there were simply not enough drivers (staff shortages), or whether the drivers had a piece of paper saying they were certified. Surprisingly, none of these were the main problem. The number of teachers and the official certification rates didn't consistently predict which schools were doing better or worse.

The one thing that did consistently predict a lower score was inadequately qualified teaching staff.

The researchers found that in schools where principals said, "Hey, our teachers aren't trained well enough to teach this curriculum," the math scores dropped by about 10 points. This wasn't a guess; it was a clear, measurable pattern that showed up again and again across ten different statistical checks. It's the difference between a driver who knows how to drive a bus and one who is just holding the steering wheel.

The Rural Penalty: Why the Back of the Bus Hurts More

Here is where the story gets even more dramatic. The researchers split the data into "city" schools and "rural" schools. They discovered a massive inequality in how much bad teaching hurts students.

In city schools, if a teacher wasn't well-qualified, the math scores dropped by about 6 points. But in rural schools? The drop was 20 points—more than three times worse!

Why? The paper suggests a "compensatory resources" theory. Imagine a student in the city whose teacher is struggling. That student can go home, hire a tutor, ask a smart neighbor, or use a parent's help. The city is full of "extra engines" that can help the bus move even if the main driver is weak.

But in rural Uzbekistan, that extra help often doesn't exist. The school is the only engine. If the teacher isn't qualified, there is no backup plan. The student is stuck with a flat tire and no spare. This means that fixing teacher quality in rural areas isn't just helpful; it's critical. The cost of doing nothing there is three times higher than in the city.

What About the Students' Feelings?

The study also looked at the students themselves. It turns out that how a student feels matters a lot. Students who felt confident in their math skills ("self-efficacy") and felt like they belonged at school did much better. Interestingly, these feelings mattered most for the students who were already struggling the most. It's like a life raft: it helps everyone, but it saves the people in the water the most.

The Bottom Line

The paper concludes that Uzbekistan has a clear path forward, but it needs to be precise. The government has already started a plan (Presidential Decree No. DP-79) to pay teachers 70% more if they get internationally recognized certificates. The data suggests this is the right move, but with a catch: it needs to reach the rural schools first.

The study doesn't claim this is a magic fix that will solve everything overnight. It's a snapshot in time, and we need to wait for the next PISA test in 2025 to see if the scores actually go up. But the evidence is strong: to fix the math scores, you don't need more buses or more certificates on the wall; you need to make sure the people driving the buses in the countryside are actually trained to drive. If you fix the drivers where the backup engines are missing, the whole bus ride gets a whole lot smoother.

Technical Summary: Teacher Qualification Deficits and Mathematics Achievement in Uzbekistan

Problem Statement
Uzbekistan's debut in the OECD Programme for International Student Assessment (PISA) in 2022 revealed a significant performance gap, with a mean mathematics score of 364 compared to the OECD average of 472. Approximately 81% of fifteen-year-old students failed to meet the minimum proficiency benchmark. Despite substantial recent policy investments, including a 70% salary supplement for certified teachers and a 46-trillion-soum education budget, there is a lack of empirical evidence regarding the specific school- and student-level determinants driving these outcomes. Previous literature has largely overlooked Central Asian transitional economies, and no prior study has applied analytically appropriate multilevel modeling to Uzbekistan's hierarchically structured PISA data to disentangle the effects of student background from institutional factors.

Methodology
The study utilizes the PISA 2022 Uzbekistan dataset, comprising 7,293 students nested within 202 schools. To address the clustered nature of the data and measurement uncertainty, the authors employ a four-model econometric sequence:

Unconditional Two-Level HLM: Used to decompose variance and calculate the Intraclass Correlation Coefficient (ICC) to determine the necessity of multilevel modeling.
OLS Baseline: A single-level regression with Fay's Balanced Repeated Replication (BRR) variance estimation to establish a benchmark for student-level predictors.
Full Two-Level HLM: The primary model incorporating student-level (Level 1) and school-level (Level 2) predictors, estimated via maximum likelihood.
Simultaneous Quantile Regression: Applied across five quantiles (10th, 25th, 50th, 75th, 90th) to examine how predictor effects vary across the performance distribution.

Data Handling and Estimation

Plausible Values: All models were estimated separately across ten plausible values (PV1MATH–PV10MATH) and combined using Rubin's (1987) rules for multiply imputed data to account for measurement uncertainty.
Variance Estimation: Standard errors were derived using Fay's BRR procedure with 80 replicate weights to reflect the complex stratified cluster sampling design.
Variable Operationalization:
- Dependent Variable: Mathematics performance.
- Level 1 Predictors: Socioeconomic status (ESCS), gender, mathematics self-efficacy, disciplinary climate, teacher support, and sense of belonging.
- Level 2 Predictors: Inadequately qualified teaching staff (UNQUALSTAFF, a binary indicator of principal-reported hindrance), staff shortage, certified teacher percentage, rural location, and student-teacher ratio.
Subsample Analysis: Separate HLMs were estimated for urban and rural subsamples to test the "compensatory resources hypothesis."

Key Results

Variance Decomposition: The unconditional model yielded an ICC of 0.181, indicating that 18.1% of the total variance in mathematics scores is attributable to differences between schools. This confirms that school-level factors are a meaningful source of inequality, justifying the use of multilevel modeling.
Student-Level Determinants: In the full HLM, mathematics self-efficacy, sense of belonging, gender (male advantage), teacher support, and socioeconomic status (ESCS) were all significant positive predictors. Notably, the coefficient for ESCS dropped by 54% when moving from the OLS baseline to the HLM, suggesting that much of the observed socioeconomic gradient is due to between-school sorting rather than within-school effects.
School-Level Determinants: Among all school-level variables, inadequately qualified teaching staff (UNQUALSTAFF) was the only factor showing a consistent and significant negative association with mathematics performance across all ten plausible values ( $\beta = -10.17, p < 0.05$ ). Variables such as staff shortage, teacher certification rates, and student-teacher ratios were not statistically significant.
Distributional Effects (Quantile Regression):
- Teacher support and sense of belonging had the strongest positive effects at the lower end of the performance distribution (Q10), benefiting students already below the proficiency threshold.
- Socioeconomic status (ESCS) exerted a progressively stronger influence toward the top of the distribution (Q90).
- The negative impact of inadequately qualified staff was consistent across the entire distribution.
Rural-Urban Disparities: The subsample analysis revealed a "rural amplification" effect. The penalty for inadequately qualified staff was 3.4 times larger in rural schools ( $\beta = -20.37$ ) compared to urban schools ( $\beta = -5.93$ ). In urban schools, the coefficient was often not statistically significant, whereas it was robust in rural contexts.

Significance and Claims
The paper claims to make three primary contributions to the literature on transitional education systems:

Methodological First: It provides the first multilevel analysis of mathematics performance in Uzbekistan using PISA 2022 data, filling a gap in Central Asian educational research.
Identification of the Binding Constraint: The study identifies that the adequacy of teacher qualifications (rather than the sheer number of teachers or domestic certification rates) is the single most consistent school-level predictor of performance. This challenges the assumption that resource quantity is the primary driver of outcomes in this context.
Geographic Targeting Evidence: By demonstrating that the cost of teacher qualification deficits is significantly higher in rural areas, the study supports the "compensatory resources hypothesis." It argues that rural students, lacking access to external learning supports (e.g., private tutoring), rely entirely on school instruction; thus, poor teacher quality in these areas imposes a disproportionately severe penalty.

Limitations and Future Directions
The authors acknowledge that the cross-sectional nature of the data limits causal inference; the findings represent conditional associations rather than causal estimates. Additionally, the measurement of rurality is binary, potentially masking heterogeneity within rural areas, and the reliance on principal-reported data introduces potential perception bias. The paper suggests that future research should utilize the upcoming PISA 2025 cycle for longitudinal difference-in-differences analysis and consider three-level models to account for regional variations.

Conclusion
The study concludes that for Uzbekistan to achieve its goal of ranking in the top 30 PISA countries by 2030, reform efforts must be geographically targeted. Specifically, the 70% salary supplement for internationally certified teachers (Presidential Decree No. DP-79) should be prioritized for rural schools, where the return on investment in teacher qualification is estimated to be 3.4 times higher than in urban centers.

Technical Summary: Teacher Qualification Deficits and Mathematics Achievement in Uzbekistan

More like this