The Long-Term Effects of the COVID-19 Pandemic on Children’s Writing: a Follow-up Replication Study

This study investigates whether the initial learning losses in young children's writing associated with emergency remote instruction during COVID-19 persisted one year later. The context is Norway, where schools closed to in-person instruction for a relatively short period in Spring 2020. Prior work by the authors found first-grade students tested immediately after remote instruction had lower writing quality, handwriting fluency, and attitude toward writing compared to a pre-pandemic cohort. The purpose here is to determine if those losses persisted into second grade. The research question: Did COVID-19-related instructional disruptions negatively impact second-graders’ writing quality, handwriting fluency, and attitude toward writing one year after the pandemic began? The study is important because writing is foundational for communication and learning, and early deficits may compound over time. Guided by the Writer(s)-within-Community (WWC) model, the study focuses on motivation (attitude), transcription (handwriting fluency), and overall text quality within a specific national context (Norway).

Two systematic reviews quantify pandemic-related learning losses, primarily in reading and mathematics: Hammerstein et al. (2021) reported median drops of −0.10 SD (math) and −0.09 SD (reading) immediately after Spring 2020 closures; König and Frey (2022) found an overall −0.18 SD across academic outcomes, with trends suggesting larger losses for younger students and earlier in the pandemic. Evidence on writing is scarce; beyond the authors’ prior Norwegian first-grade study, Haelermans et al. (2021) found a −0.06 SD drop in spelling among Dutch primary students. Theoretical framing via the Writer(s)-within-Community (WWC) model (Graham, 2018a, 2018b) emphasizes that writing development is shaped by community context and learners’ cognitive and motivational resources, justifying focus on a single-country context and on writing quality, handwriting fluency, and attitude. Prior research also indicates gender and language background relate to writing outcomes, and school-level characteristics can influence performance.

Design: Longitudinal replication comparing second-grade cohorts from the same Norwegian schools before and during COVID-19 using multilevel modeling controlling for school- and student-level covariates. Setting: Norway implemented emergency remote instruction March 12–April 27, 2020. Subsequent 2020/2021 in-person schooling included mitigation measures; some schools experienced short closures. Teacher absence and mitigation levels varied across schools. Participants: N=2309 second-graders from 185 classrooms in 59 schools across four municipalities. Before COVID-19 cohort (2019 testing, pre-pandemic): n=1668 (53.2% girls). During COVID-19 cohort (2021 testing, one year into pandemic): n=641 (51.9% girls), all of whom had been first-grade participants in the prior COVID-19 study. Language groups: L1 Norwegian, L2 (Norwegian as second language), and bilingual. Available school and student data suggest representativeness relative to national indicators. Sampling: Students originated from a large RCT writing intervention project (2019–2021). To avoid intervention confounds, only control-condition schools contributed to the During COVID-19 cohort; Before COVID-19 cohort data were collected pre-randomization. Consent rates were 91.2% (Before) and 84.8% (During); inclusion required complete data on all writing measures. Measures: - Handwriting fluency: 90-second copy task (from Monroe & Sherman, 1996), letters per minute score; coding reliability κ=0.812, ICC=0.99. - Writing quality: Discursive letter writing task (45 minutes) scored on eight 1–5 scales (audience awareness, organization, content relevance, vocabulary, sentence construction, spelling, legibility, punctuation). Many-facet Rasch measurement (FACETS) produced a fair average adjusted for rater severity and scale difficulty; reliability of separation: 0.94 (Before), 0.95 (During); standardized residuals within recommended bounds. - Attitude toward writing: Four-item student survey (3-point star ratings) averaged to a single score; EFA supported a single factor (53.8% variance), alpha=0.71. Covariates: School-level (standardized): national test results (Grade 5), school size, proportion of certified teachers, students per special education teacher, school hours per student. Student-level: gender, language background (L1, bilingual, L2), and cohort membership (Before vs During COVID-19). Procedures: Teachers administered tasks during regular class time following standardized manuals and videos; counterbalanced order of writing and copying tasks. Data collection windows were May/June 2019 (Before) and May/June 2021 (During). Writing texts were anonymized and double-rated by trained raters; cross-occasion linking using 50 common texts ensured scoring comparability. Analytical strategy: Three-level multilevel linear regressions (students nested in classrooms nested in schools) using lme4 in R. Separate models for writing quality, handwriting fluency, and attitude. Null models estimated ICCs. Full models included all covariates; students with any missing outcome data were excluded to maintain comparability across outcomes. Effect size f2 reported for cohort contribution to R2.

- Cohort effects (Before vs During COVID-19) were not statistically significant for any outcome: - Writing quality: Estimate = 0.074, p = 0.089 - Handwriting fluency: Estimate = −1.707 letters/min, p = 0.055 (trend only) - Attitude toward writing: Estimate = −0.044, p = 0.155 The added R2 due to cohort was minimal: 0.003 (quality; f2≈0.003), 0.006 (fluency; f2≈0.006), 0.001 (attitude; f2≈0.001), all well below a small effect. - ICCs indicated clustering, strongest for text quality: ICC(class)=0.255, ICC(school)=0.136; handwriting fluency ICCs were lower (class=0.183, school=0.090); attitude ICCs were small (class=0.066, school=0.010). - Student-level predictors: - Gender: Girls scored higher on all outcomes (quality: +0.312, p<0.001; fluency: +4.731 letters/min, p<0.001; attitude: +0.233, p<0.001). - Language: Relative to L1, bilingual and L2 had lower writing quality (bilingual: −0.080, p=0.009; L2: −0.083, p=0.050). L2 had lower handwriting fluency (−2.039, p=0.042); bilingual fluency difference not significant (−0.700, p=0.337). No significant language effects on attitude. - School-level predictors: - National test results positively predicted writing quality (β=0.084, p=0.013) but not fluency or attitude. - School size positively predicted handwriting fluency (β=1.426, p=0.018) but not other outcomes. - Students per special education teacher negatively predicted writing quality (β=−0.081, p=0.008). Overall, initial first-grade writing losses reported previously were no longer evident by end of second grade.

The study addressed whether learning losses in writing persisted one year into the pandemic. Results indicate that by the end of second grade, the cohort that experienced remote instruction in Spring 2020 performed comparably to pre-pandemic peers on writing quality, handwriting fluency, and attitudes, suggesting that early losses dissipated with a return to in-person schooling. This aligns with trends in broader meta-analyses suggesting smaller losses later in the pandemic and highlights contextual factors that may have facilitated recovery in Norway: short duration of national remote instruction, strong and autonomous educational system, and adaptive instructional responses as schools and teachers adjusted to pandemic realities. Gender differences favoring girls persisted across outcomes, and school-level factors (national test performance, school size, and special education staffing ratios) were associated with specific writing outcomes, underscoring the role of educational context. The absence of persistent negative effects in this setting suggests that sustained in-person instruction and supportive systems can mitigate initial disruptions, though results may not generalize to countries with longer or repeated closures or different resources. The unexpected association of larger school size with higher handwriting fluency warrants further investigation into resources and instructional emphasis in larger schools.

This longitudinal replication finds no sustained negative impact of COVID-19-related disruptions on second-graders’ writing quality, handwriting fluency, or attitudes in Norway, indicating recovery from initial first-grade losses. Contributions include evidence that early writing deficits may be transient under conditions of brief closures and robust schooling contexts, and identification of student- and school-level correlates of writing outcomes. Future research should: (1) examine long-term writing trajectories across different countries and grades with varying closure lengths; (2) investigate instructional adaptations and supports that facilitated recovery; (3) explore mechanisms behind school size effects on handwriting fluency; and (4) include broader writing constructs and standardized cross-national assessments where feasible.

- Non-randomized exposure: Students could not be randomly assigned to COVID vs non-COVID conditions; cohort comparisons may include unmeasured differences despite controls. - Sampling differences: Before cohort (n=1668) was larger than During cohort (n=641) and the During cohort was limited to RCT control schools, potentially affecting generalizability. - Grade and constructs: Focus limited to second grade and three constructs (quality, handwriting fluency, attitude), not the full range of writing skills. - Missing contextual data: No detailed data on writing instruction during remote instruction or the following year, nor parental involvement, limiting explanation of recovery mechanisms. - Country-specific context: Norway’s short closures and educational context may limit generalizability to settings with longer or repeated closures or different resources.