pith. sign in

arxiv: 2605.17646 · v1 · pith:7PNRIZQBnew · submitted 2026-05-17 · 📊 stat.ME

Starshaped Mean Residual Life Models for Non-Monotonic Survival Data: A Bayesian PMRL Regression Framework with Applications to Teacher Retention

Mohammad Sepehrifar This is my paper

Pith reviewed 2026-05-19 22:07 UTC · model grok-4.3

classification 📊 stat.ME
keywords starshaped mean residual lifenon-monotonic survivalproportional mean residual lifeBayesian survival modelsteacher retentionright censoringmean residual life models
0
0 comments X

The pith

A starshaped mean residual life model captures non-monotonic survival patterns by requiring only that the mean residual life divided by time is nondecreasing.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces a Starshaped Mean Residual Life (SMEL) framework to handle survival data where hazards are not monotonic, such as high early attrition followed by stabilization. This is achieved by assuming the ratio m(t)/t is nondecreasing, which models the move from vulnerability to equilibrium without needing full monotonicity like in Cox models. The approach is extended to a proportional mean residual life regression using Bayesian methods with Weibull distributions. Simulations on thousands of datasets demonstrate lower bias and better predictive accuracy than standard models, and the method is applied to rural STEM teacher retention data to confirm equilibrium and link persistence to achievement improvements.

Core claim

The central claim is that non-monotonic survival data can be effectively modeled using the starshaped mean residual life property, formalized by the nondecreasing condition on m(t)/t, and this property supports a proportional mean residual life regression model estimated via adaptive Bayesian methods, as validated by Monte Carlo simulations showing maintained low bias under censoring and improved scores over Cox, and confirmed in the teacher retention application with starshaped equilibrium and quantified gains from persistence.

What carries the argument

The starshaped mean residual life (SMEL) property requiring that the mean residual life m(t) divided by time t is nondecreasing, which allows modeling the transition from early vulnerability to mid-career equilibrium in survival processes.

If this is right

  • The SMEL-PMRL model maintains bias at most 0.02 under 40% right-censoring in 48,000 simulated datasets.
  • It reduces the integrated Brier score by 19% compared to Cox models.
  • The model achieves a 5.4% improvement in AIC.
  • In the teacher data, it identifies 38% early-career tenure decline in the first three years.
  • Joint modeling indicates that persistence beyond year 3 produces 31-point achievement gains over four years.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the starshaped condition proves common in workforce data, the framework could guide interventions targeting the first few years to boost long-term retention.
  • The Bayesian estimation allows easy extension to include time-varying covariates for more dynamic predictions.
  • Similar models might apply to patient survival in medicine where early risks differ from later stability.
  • Testing the model on other non-monotonic datasets like employee turnover in tech industries would further validate its generality.

Load-bearing premise

The modeling premise that the ratio of the mean residual life function to time is nondecreasing, which is needed for the starshaped equilibrium to hold and to formalize the shift from vulnerability to stability.

What would settle it

A dataset in which the estimated mean residual life ratio m(t)/t decreases over time intervals after the initial period would falsify the starshaped property for that application.

Figures

Figures reproduced from arXiv: 2605.17646 by Mohammad Sepehrifar.

Figure 1
Figure 1. Figure 1: Simulation results under balanced hazard regime (BHR) with [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Prior sensitivity and joint mean residual life recovery under balanced hazard [PITH_FULL_IMAGE:figures/full_fig_p015_2.png] view at source ↗
read the original abstract

We develop a Starshaped Mean Residual Life (SMEL) framework for survival data with non-monotonic hazard patterns, where early-stage attrition is followed by mid-career stabilization. Unlike Cox proportional hazards models or standard mean residual life models requiring monotonicity, SMEL accommodates complex temporal dynamics by requiring only that $m(t)/t$ be nondecreasing, formalizing the transition from vulnerability to equilibrium. We extend SMEL to regression settings via proportional mean residual life (PMRL) models, $m(t\mid Z)=m_0(t)\exp(Z^\top\gamma)$, with adaptive Bayesian estimation using three-parameter Weibull--resilience distributions and the No-U-Turn Sampler. Monte Carlo simulations across 48,000 datasets show SMEL-PMRL maintains bias $\leq 0.02$ under 40\% right-censoring, reduces integrated Brier score by 19\% over Cox models ($2.34$ vs.\ $2.88\times10^{-2}$), and achieves 5.4\% AIC improvement. Joint longitudinal-survival extensions via shared frailty enable simultaneous modeling of correlated time-to-event and continuous outcomes. Application to 169 rural STEM teachers (2018--2023, NSF Noyce) confirms starshaped equilibrium ($\Lambda=12.47$, $p=0.002$), with 38\% early-career tenure decline (years 1--3). The joint model ($\hat{\theta}=0.41$, 95\% CI: $[0.35,\,0.47]$) shows persistence beyond year~3 yields 31-point cumulative achievement gains (0.56~SD) over four years. SMEL-PMRL offers a flexible, theoretically grounded alternative to proportional hazards for workforce dynamics and high-attrition settings where equilibrium processes govern long-term stability.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript develops a Starshaped Mean Residual Life (SMEL) framework for survival data with non-monotonic hazard patterns, where early attrition transitions to mid-career stabilization. It extends this via proportional mean residual life (PMRL) regression m(t|Z)=m0(t)exp(Z^T γ), estimated adaptively with three-parameter Weibull-resilience distributions and NUTS sampling. Monte Carlo results across 48,000 datasets report bias ≤0.02 under 40% censoring, 19% lower integrated Brier score than Cox, and AIC gains; the application to 169 rural STEM teachers finds significant starshaped equilibrium (Λ=12.47, p=0.002) and benefits from joint frailty modeling of persistence and achievement.

Significance. If the starshaped condition is preserved under the proposed estimation, the framework supplies a theoretically grounded alternative to Cox or standard MRL models for workforce and high-attrition settings that exhibit vulnerability-to-equilibrium dynamics. The reported simulation improvements and the empirical finding of 31-point achievement gains from persistence beyond year 3 indicate potential practical value, though this depends on verification that the key modeling assumption holds in finite samples with censoring.

major comments (2)
  1. [§3] §3 (SMEL definition and PMRL extension): The central premise requires m0(t)/t to be nondecreasing for all t to formalize the vulnerability-to-equilibrium transition, yet the manuscript supplies neither an analytic proof that the three-parameter Weibull-resilience family enforces this nor any post-fit diagnostic (e.g., posterior checks on simulated or real m0(t)/t trajectories). Under 40% right-censoring and n=169, tail identification of m0(t) occurs primarily through the frailty term, so small perturbations can produce decreasing segments that invalidate the SMEL justification.
  2. [Simulation study] Simulation study (Monte Carlo section, 48,000 datasets): Bias ≤0.02 and Brier-score reduction are reported, but no verification is described that the fitted baselines in the simulated replicates satisfy the starshaped property. Without such checks, the performance gains cannot be attributed to a correctly specified SMEL-PMRL model rather than to the flexibility of the Weibull-resilience prior alone.
minor comments (2)
  1. [Abstract] Abstract: the integrated Brier scores (2.34 vs. 2.88×10^{-2}) should state the time horizon and scaling explicitly for comparability with other survival literature.
  2. [Application] Application section: the exact censoring mechanism, data exclusion rules, and definition of the 38% early-career decline should be stated more precisely to support replication.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments on our manuscript. We address each major comment below in a point-by-point manner, providing clarifications and committing to specific revisions that strengthen the theoretical and empirical support for the SMEL-PMRL framework.

read point-by-point responses

  1. Referee: [§3] The central premise requires m0(t)/t to be nondecreasing for all t to formalize the vulnerability-to-equilibrium transition, yet the manuscript supplies neither an analytic proof that the three-parameter Weibull-resilience family enforces this nor any post-fit diagnostic (e.g., posterior checks on simulated or real m0(t)/t trajectories). Under 40% right-censoring and n=169, tail identification of m0(t) occurs primarily through the frailty term, so small perturbations can produce decreasing segments that invalidate the SMEL justification.

    Authors: We agree that an explicit analytic verification and post-fit diagnostics would strengthen the justification for the SMEL assumption. The three-parameter Weibull-resilience family was selected for its ability to accommodate resilience parameters that promote nondecreasing m(t)/t behavior consistent with the starshaped condition, but the original submission did not include a formal proof or trajectory checks. In the revision we will add a short analytic demonstration that the family satisfies the nondecreasing m0(t)/t requirement under the parameterization used, together with posterior predictive checks (including plots of posterior mean trajectories and the fraction of draws satisfying the property) for both the simulation replicates and the teacher-retention data. These additions will directly address concerns about tail behavior under 40% censoring and the role of the frailty term. revision: yes

  2. Referee: [Simulation study] Bias ≤0.02 and Brier-score reduction are reported, but no verification is described that the fitted baselines in the simulated replicates satisfy the starshaped property. Without such checks, the performance gains cannot be attributed to a correctly specified SMEL-PMRL model rather than to the flexibility of the Weibull-resilience prior alone.

    Authors: We acknowledge that reporting verification of the starshaped property in the fitted baselines is necessary to attribute performance gains specifically to the SMEL-PMRL specification. Although the data-generating mechanisms in the Monte Carlo study were constructed to obey the starshaped condition, we did not present post-estimation diagnostics on the recovered m0(t)/t functions. In the revised manuscript we will include summary statistics across the 48,000 replicates (e.g., the proportion of replicates in which the estimated m0(t)/t is nondecreasing) and representative average trajectory plots. This will confirm that the reported bias and Brier-score improvements arise from correct specification rather than prior flexibility. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation or claims

full rationale

The paper defines the SMEL framework explicitly via the modeling assumption that m(t)/t is nondecreasing, then applies the standard PMRL regression form m(t|Z)=m0(t)exp(Z^T γ) with Bayesian estimation under a Weibull-resilience prior. Monte Carlo results (bias, Brier score, AIC) are generated from external simulated datasets rather than reducing fitted quantities to predictions by construction. The teacher-retention application reports posterior quantities (Λ, θ, achievement gains) from observed data. No load-bearing self-citation, uniqueness theorem, or ansatz smuggling appears in the provided text; the starshaped condition is stated as a premise, not derived from the estimation step itself. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

3 free parameters · 1 axioms · 0 invented entities

The framework rests on the starshaped modeling assumption and standard survival analysis properties; several parameters are estimated from data in both simulations and the teacher application.

free parameters (3)
  • regression coefficients gamma
    Fitted via Bayesian PMRL model to the teacher retention covariates.
  • Weibull-resilience distribution parameters
    Three-parameter Weibull-resilience distribution parameters estimated using NUTS sampler.
  • frailty parameter theta
    Estimated in the joint longitudinal-survival extension (reported as 0.41).
axioms (1)
  • domain assumption m(t)/t is nondecreasing
    Invoked as the defining requirement for the starshaped mean residual life property that accommodates non-monotonic hazards.

pith-pipeline@v0.9.0 · 5870 in / 1512 out tokens · 46580 ms · 2026-05-19T22:07:07.516305+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

65 extracted references · 65 canonical work pages

  1. [1]

    D., & Dowling, N

    Borman, G. D., & Dowling, N. M. (2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Review of Educational Research, 78(3), 367--409

    work page 2008

  2. [2]

    Carver-Thomas, D., & Darling-Hammond, L. (2017). Teacher turnover: Why it matters and what we can do about it. Learning Policy Institute

    work page 2017

  3. [3]

    Goldhaber, D., Gross, B., & Player, D. (2011). Teacher career paths, teacher quality, and persistence in the classroom: Are public schools keeping their best? Journal of Policy Analysis and Management, 30(1), 57--87

    work page 2011

  4. [4]

    Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis. American Educational Research Journal, 38(3), 499--534

    work page 2001

  5. [5]

    M., & May, H

    Ingersoll, R. M., & May, H. (2012). The magnitude, destinations, and determinants of mathematics and science teacher turnover. Educational Evaluation and Policy Analysis, 34(4), 435--464

    work page 2012

  6. [6]

    Monk, D. H. (2007). Recruiting and retaining high-quality teachers in rural areas. The Future of Children, 17(1), 155--174

    work page 2007

  7. [7]

    Podolsky, A., Kini, T., Bishop, J., & Darling-Hammond, L. (2019). Strategies for attracting and retaining educators: What does the evidence say? Learning Policy Institute

    work page 2019

  8. [8]

    G., Hanushek, E

    Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (2005). Teachers, schools, and academic achievement. Econometrica, 73(2), 417--458

    work page 2005

  9. [9]

    Rockoff, J. E. (2004). The impact of individual teachers on student achievement: Evidence from panel data. American Economic Review, 94(2), 247--252

    work page 2004

  10. [10]

    Ronfeldt, M., Loeb, S., & Wyckoff, J. (2013). How teacher turnover harms student achievement. American Educational Research Journal, 50(1), 4--36

    work page 2013

  11. [11]

    Sepehrifar, M. (2025). Using Starshaped Mean Equilibrium Class of Life Distributions. Manuscript

    work page 2025

  12. [12]

    L., Johnson, J., & Klein, B

    Showalter, D., Hartman, S. L., Johnson, J., & Klein, B. (2019). Why rural matters 2018--2019: The time is now. Rural School and Community Trust

    work page 2019

  13. [13]

    S., & Inventado, P

    Baker, R. S., & Inventado, P. S. (2014). Educational data mining and learning analytics. In K. Sawyer (Ed.), Cambridge Handbook of the Learning Sciences (2nd ed., pp. 253--274). Cambridge University Press

    work page 2014

  14. [14]

    Bienkowski, M., Feng, M., & Means, B. (2012). Enhancing teaching and learning through educational data mining and learning analytics. U.S. Department of Education, Office of Educational Technology

    work page 2012

  15. [15]

    E., & Proschan, F

    Barlow, R. E., & Proschan, F. (1981). Statistical Theory of Reliability and Life Testing: Probability Models. Silver Spring, MD: Holt, Rinehart and Winston

    work page 1981

  16. [16]

    Cao, J., & Zhou, M. (2016). Proportional mean residual life model for right-censored data. Biometrika, 103(1), 177--190

    work page 2016

  17. [17]

    Hsu, L., Tseng, G., & Li, H. (2018). Joint frailty models for correlated time-to-event data with applications in education research. Statistical Methods in Medical Research, 27(12), 3664--3680

    work page 2018

  18. [18]

    L., & Meier, P

    Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations. Journal of the American Statistical Association, 53(282), 457--481

    work page 1958

  19. [19]

    R., McLaughlin, E

    Koedinger, K. R., McLaughlin, E. A., & Heffernan, N. T. (2015). A data-driven approach to understanding learning in context. Journal of Educational Data Mining, 7(2), 1--27

    work page 2015

  20. [20]

    Rubin, D. B. (1987). Multiple Imputation for Nonresponse in Surveys. New York: John Wiley & Sons

    work page 1987

  21. [21]

    D., & Willett, J

    Singer, J. D., & Willett, J. B. (2003). Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford University Press

    work page 2003

  22. [22]

    Siemens, G., & Long, P. (2011). Penetrating the fog: Analytics in learning and education. EDUCAUSE Review, 46(5), 30--40

    work page 2011

  23. [23]

    M., Santibañez, L., & Daley, G

    Guarino, C. M., Santibañez, L., & Daley, G. (2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173--208

    work page 2006

  24. [24]

    Banerjee, M., Sen, B., & Woodroofe, M. (2007). Likelihood ratio tests for monotone functions. Journal of the Royal Statistical Society, Series B, 69(1), 115--135

    work page 2007

  25. [25]

    Player, D. (2008). Predicting teacher attrition and retention: Evidence from longitudinal data. Economics of Education Review, 27(4), 435--445

    work page 2008

  26. [26]

    Sutcher, L., Darling-Hammond, L., & Carver-Thomas, D. (2019). Understanding teacher shortages: An analysis of teacher supply and demand in the United States. Education Policy Analysis Archives, 27(35), 1--36

    work page 2019

  27. [27]

    Marder, M., Horn, C., Stephens, S., & Rhodes, A. (2022). Student learning and teacherretention for graduates of Texas Noyce programs. Education Policy Analysis Archives, 30(147).https://doi.org/10.14507/epaa.30.7254

    work page doi:10.14507/epaa.30.7254 2022

  28. [28]

    Siemens, G., & Long, P. (2011). Penetrating the fog: Analytics in learning and education. EDUCAUSE Review, 46(5), 30–40

    work page 2011

  29. [29]

    M., Yarahmadian, S., Sepehrifar, M., & Gelder, P

    Harandi, F. M., Yarahmadian, S., Sepehrifar, M., & Gelder, P. H. (2014). The

    work page 2014

  30. [30]

    Sepehrifar, M., & Yarahmadian, S. (2016). Decreasing renewal dichotomous

    work page 2016

  31. [31]

    Sepehrifar, M., & Yarahmadian, S. (2016). Testing monotonic equilibrium residual

    work page 2016

  32. [32]

    Ingersoll, R., & May, H. (2012). The magnitude, destinations, and determinants of mathematics and science teacher turnover. Educational Evaluation and Policy Analysis, 34(4), 435--464

    work page 2012

  33. [33]

    Sutcher, L., Darling-Hammond, L., & Carver-Thomas, D. (2019). Understanding teacher shortages: An analysis of teacher supply and demand in the United States. Education Policy Analysis Archives, 27(35), 1--40

    work page 2019

  34. [34]

    Nelson, W. (2009). Accelerated testing: Statistical models, test plans, and data analyses. Wiley

    work page 2009

  35. [35]

    Modarres, M. (2016). Reliability engineering and risk analysis: A practical guide. CRC Press

    work page 2016

  36. [36]

    E., & Baruwa, B

    Olalowo,I. E., & Baruwa, B. I., & Olalowo, H. O. & Aluko, F. O.. Reconceptualising early childhood education in Nigeria: a praxis for policy and research. Oxford Review of Education 0:0, pages 1-17

    work page

  37. [37]

    H., & Brown, C., & Kokotsaki, D

    Nguyen, D., & See, B. H., & Brown, C., & Kokotsaki, D. (2024). Leadership for teacher retention: exploring the evidence base on why and how to support teacher autonomy, development, and voice. Oxford Review of Education, 1–21

    work page 2024

  38. [38]

    D., & Dowling, N

    Borman, G. D., & Dowling, N. M. (2008). Teacher attrition and retention: A meta-analytic and narrative review of the research. Review of Educational Research, 78(3), 367–409

    work page 2008

  39. [39]

    Goldhaber, D., Gross, B., & Player, D. (2011). Teacher career paths, teacher quality, and persistence in the classroom: Are public schools keeping their best? Journal of Policy Analysis and Management, 30(1), 57–87

    work page 2011

  40. [40]

    Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis. American Educational Research Journal, 38(3), 499–534

    work page 2001

  41. [41]

    M., & May, H

    Ingersoll, R. M., & May, H. (2012). The magnitude, destinations, and determinants of mathematics and science teacher turnover. Educational Evaluation and Policy Analysis, 34(4), 435–464

    work page 2012

  42. [42]

    Monk, D. H. (2007). Recruiting and retaining high-quality teachers in rural areas. The Future of Children, 17(1), 155–174

    work page 2007

  43. [43]

    G., Hanushek, E

    Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (2005). Teachers, schools, and academic achievement. Econometrica, 73(2), 417–458

    work page 2005

  44. [44]

    Rockoff, J. E. (2004). The impact of individual teachers on student achievement: Evidence from panel data. American Economic Review, 94(2), 247–252

    work page 2004

  45. [45]

    Ronfeldt, M., Loeb, S., & Wyckoff, J. (2013). How teacher turnover harms student achievement. American Educational Research Journal, 50(1), 4–36

    work page 2013

  46. [46]

    Sepehrifar, M. (2024). Using Starshaped Mean Equilibrium Class of Life Distributions. Manuscript

    work page 2024

  47. [47]

    L., Johnson, J., & Klein, B

    Showalter, D., Hartman, S. L., Johnson, J., & Klein, B. (2019). Why rural matters 2018–2019: The time is now. Rural School and Community Trust

    work page 2019

  48. [48]

    S., & Inventado, P

    Baker, R. S., & Inventado, P. S. (2014). Educational data mining and learning analytics. In K. Sawyer (Ed.), Cambridge Handbook of the Learning Sciences (2nd ed., pp. 253–274). Cambridge University Press

    work page 2014

  49. [49]

    Cao, J., & Zhou, M. (2016). Proportional mean residual life model for right-censored data. Biometrika, 103(1), 177–190

    work page 2016

  50. [50]

    Hsu, L., Tseng, G., & Li, H. (2018). Joint frailty models for correlated time-to-event data with applications in education research. Statistical Methods in Medical Research, 27(12), 3664–3680

    work page 2018

  51. [51]

    L., & Meier, P

    Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incomplete observations. Journal of the American Statistical Association, 53(282), 457–481

    work page 1958

  52. [52]

    R., McLaughlin, E

    Koedinger, K. R., McLaughlin, E. A., & Heffernan, N. T. (2015). A data-driven approach to understanding learning in context. Journal of Educational Data Mining, 7(2), 1–27

    work page 2015

  53. [53]

    M., Santibañez, L., & Daley, G

    Guarino, C. M., Santibañez, L., & Daley, G. (2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173–208

    work page 2006

  54. [54]

    Banerjee, M., Sen, B., & Woodroofe, M. (2007). Likelihood ratio tests for monotone functions. Journal of the Royal Statistical Society, Series B, 69(1), 115–135

    work page 2007

  55. [55]

    Player, D. (2008). Predicting teacher attrition and retention: Evidence from longitudinal data. Economics of Education Review, 27(4), 435–445

    work page 2008

  56. [56]

    Sutcher, L., Darling-Hammond, L., & Carver-Thomas, D. (2019). Understanding teacher shortages: An analysis of teacher supply and demand in the United States. Education Policy Analysis Archives, 27(35), 1–36

    work page 2019

  57. [57]

    Sepehrifar, M. (2024). Using Starshaped Mean Equilibrium Class of Life Distributions. (Manuscript)

    work page 2024

  58. [58]

    S., & Inventado, P

    Baker, R. S., & Inventado, P. S. (2014). Educational data mining and learning analytics. In Learning analytics (pp. 61–75). Springer

    work page 2014

  59. [59]

    R., Kim, J., Jia, J

    Koedinger, K. R., Kim, J., Jia, J. Z., McLaughlin, E. A., & Bier, N. L. (2015). Learning is not a spectator sport: Doing is better than watching for learning from a MOOC. In Proceedings of the Second (2015) ACM Conference on Learning @ Scale (pp. 111–120)

    work page 2015

  60. [60]

    Player, D. (2008). Teacher attrition and retention in public schools. Washington, DC: U.S. Department of Education

    work page 2008

  61. [61]

    Guarino, C., Santibañez, L., & Daley, G. (2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173–208

    work page 2006

  62. [62]

    Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society, Series B, 34(2), 187–220

    work page 1972

  63. [63]

    B., Stern, H

    Gelman, A., Carlin, J. B., Stern, H. S., Dunson, D. B., Vehtari, A., & Rubin, D. B. (2013). Bayesian Data Analysis (3rd ed.). CRC Press

    work page 2013

  64. [64]

    W., & Bryk, A

    Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical Linear Models: Applications and Data Analysis Methods (2nd ed.). Sage Publications

    work page 2002

  65. [65]

    M., Santibañez, L., & Daley, G

    Guarino, C. M., Santibañez, L., & Daley, G. A. (2006). Teacher recruitment and retention: A review of the recent empirical literature. Review of Educational Research, 76(2), 173–208

    work page 2006