When Pupil Books Meet Teacher Routines: Dataset for a Cluster Randomized Evaluation of Luganda Oral Reading Fluency in Ugandan Primary 4 Classrooms

Data collection included baseline and endline assessment waves as well as weekly implementation monitoring activities throughout the intervention period.The geographical scope covers Luganda-speaking educational contexts in central Uganda where Luganda is taught as a local-language literacy subject in Primary Four classrooms. Schools included both government and low-fee private schools operating under ordinary classroom conditions.The temporal resolution of the data varies across sheets:Learner reading assessments were collected at baseline and endline;Lesson implementation and monitoring data were collected weekly;Classroom observations were conducted repeatedly across the intervention period;School and teacher demographic data were collected once at baseline.The spatial resolution of the dataset is hierarchical and includes:School level;Classroom level;Teacher level; andIndividual learner level. Data Generation ProceduresThe dataset was generated through a pragmatic cluster-randomized educational trial.

Schools constituted the unit of randomization, while literacy assessments and instructional observations were conducted at classroom and learner levels.Thirty participating schools were screened for eligibility based on:Availability of Primary Four Luganda instruction;Presence of intact classrooms;Willingness to participate in classroom observation and literacy assessment procedures; andAbsence of existing weekly structured reading interventions.Eligible schools were randomly assigned to intervention or comparison conditions using computer-generated randomization procedures.Data generation involved multiple complementary procedures:Learner Reading AssessmentsLearners completed oral reading fluency tasks administered individually by trained literacy assessors.

During each assessment, learners read standardized Luganda passages aloud while assessors recorded:Reading accuracy;Correct words read;Omissions;Substitutions;Hesitations;Self-corrections; andReading time.Correct words per minute (CWPM) was computed from timed oral reading performance.Classroom ObservationsStructured classroom observations were conducted during Luganda literacy lessons using rubric-based observation instruments.

Observers documented:Reading modelling;Guided reading routines;Pair-reading activities;Corrective feedback practices;Comprehension support;Lesson pacing; andPupil-book usage.Monitoring and Fidelity ChecksWeekly monitoring visits documented implementation fidelity, protocol adherence, lesson delivery, and deviations from study procedures. Monitoring logs included narrative field notes and structured fidelity indicators.Teacher Surveys and Acceptability RatingsTeachers completed structured acceptability questionnaires assessing:Ease of implementation;Classroom feasibility;Usefulness of reading materials;Learner participation support; andWillingness to continue intervention practices.Administrative and Contextual DataSchool-level and classroom-level contextual variables were collected through administrative records and baseline surveys. Devices, Tools, and Instruments UsedData collection relied on standardized literacy-assessment tools, structured observation schedules, monitoring forms, and intervention implementation records.The following tools and instruments were used:Luganda oral reading passages;Stopwatch timing procedures for oral reading fluency measurement;Structured classroom-observation rubrics;Implementation-fidelity checklists;Teacher acceptability questionnaires;Monitoring visit forms;Lesson-dose tracking sheets; andData-quality verification logs.Observers and assessors underwent training and calibration prior to data collection.

Double-scoring procedures were implemented for selected reading assessments to evaluate inter-rater reliability.Data were entered and stored electronically using Microsoft Excel workbook structures with linked worksheets and coded identifiers. Data Processing and Cleaning ProceduresData processing involved several sequential stages designed to ensure accuracy, completeness, consistency, and confidentiality.Step 1: Field VerificationCompleted paper instruments were reviewed in the field immediately after data collection to identify:Missing values;Illegible entries;Timing inconsistencies; andIncomplete learner records.Step 2: Double Data Entry and ValidationSelected assessment records and fidelity scores were double-entered and cross-checked to reduce transcription errors.Step 3: De-identificationLearner names and direct personal identifiers were removed and replaced with coded identifiers linking:Schools;Teachers;Classrooms; andLearners.Step 4: Coding and StandardizationCategorical variables such as:School type;Intervention arm;Gender;Participation status; andObservation ratingswere standardized using consistent coding conventions.Step 5: Quality Assurance ChecksAutomated and manual checks were conducted to identify:Duplicate records;Impossible values;Missing timestamps;Out-of-range scores; andInternal inconsistencies between sheets.Step 6: Reliability VerificationA subset of oral reading fluency assessments underwent independent rescoring.

Absolute score differences and agreement percentages were calculated to estimate inter-rater reliability. Data Structure and File DescriptionThe dataset is contained in a single Microsoft Excel workbook (.xlsx) consisting of 18 worksheets.1. READMEContains a general overview of the project, workbook organization, and dataset purpose.Format: Text-based worksheetContent type: Metadata and project documentation 2.

Data_DictionaryDefines variable names, descriptions, coding conventions, and examples for all worksheets.Format: Tabular metadata sheetContent type: Variable definitions and coding reference 3. Study_QuestionsSummarizes research questions, analytic objectives, key findings, and interpretive cautions.Format: Text and summary table sheet 4. Analysis_DashboardContains summary indicators and visual analytical summaries of major intervention outcomes.Format: Summary table and dashboard worksheet 5.

Model_SummaryContains statistical outputs including:Means;Standard deviations;Confidence intervals;Effect sizes;p-values; andIntraclass correlation coefficients.Units:CWPM measured in correct words per minute;Effect sizes reported as Hedges’ g;Confidence intervals reported numerically. 6. SchoolsContains school-level information.Rows represent individual schools.Columns represent school characteristics.Variables include:School ID;District;Community type;School type;Enrollment;Eligibility status; andIntervention assignment. 7.

Teachers_ClassroomsContains teacher- and classroom-level variables.Rows represent teacher-classroom units.Columns represent instructional and demographic variables.Variables include:Teacher gender;Years of experience;Literacy training exposure;Class size;Monitoring visits; andAcceptability ratings.Units:Experience measured in years;Class size measured as number of learners. 8. Learners_MasterContains learner demographic and participation data.Rows represent individual learners.Columns represent learner characteristics and participation indicators.Variables include:Learner ID;Age;Gender;Home language;Consent status; andAssessment participation.Units:Age measured in years.The sheet contains more than 500 learner records. 9.

Learner_AssessmentsThis is the primary literacy assessment dataset.Rows represent learner assessment occasions.Columns represent assessment variables.Variables include:Assessment wave;Reading timing;Correct words;Errors;Hesitations;Omissions;Substitutions; andAssessor IDs.Units:Reading time measured in seconds;CWPM measured in correct words per minute;Error counts recorded as integers. 10. Lesson_DoseDocuments intervention exposure and instructional delivery.Rows represent instructional sessions.Columns represent implementation variables.Variables include:Planned lessons;Delivered lessons;Curriculum topics;Delivery dates; andTeacher notes. 11.

Observations_FidelityContains structured classroom-observation ratings.Rows represent observed lessons.Columns represent fidelity dimensions.Variables include rubric scores for:Reading modelling;Guided practice;Corrective feedback;Pair reading;Comprehension linkage; andPupil-book use.Most scores use ordinal rating scales. 12. Book_AvailabilityMeasures learner access to reading materials.Rows represent classroom observation points.Columns represent book-access indicators.Variables include:Learners with individual books;Shared books;Learners without texts; andPercentage access indicators.Units:Counts and percentages. 13.

Teacher_AcceptabilityContains teacher survey responses regarding intervention feasibility and usefulness.Variables include Likert-scale ratings and open-ended responses.Units:Ordinal ratings on 5-point scales. 14. Monitoring_VisitsDocuments implementation-monitoring activities.Variables include:Scheduled visits;Fidelity checks;Monitoring comments;Protocol deviations; andCorrective actions. 15. Reliability_DoubleScoringContains inter-rater reliability information.Rows represent double-scored learner assessments.Columns represent:Original scores;Secondary scores;Absolute differences; andAgreement percentages.Units:CWPM values;Percentage agreement. 16.

Protocol_DeviationsDocuments deviations from the research protocol.Variables include:Severity classification;Description of deviation; andCorrective action taken. 17. Data_Quality_ChecksContains records of data verification and cleaning procedures.Variables include:Missing-value checks;Duplicate detection;Logical consistency reviews; andCorrection notes. 18. Instrument_MappingMaps research instruments to corresponding worksheet variables.This sheet functions as an instrument-tracking reference. Row and Column MeaningsAcross the workbook:Rows generally represent observational units such as schools, teachers, classrooms, learners, lessons, or assessment occasions;Columns represent variables, indicators, scores, identifiers, dates, or categorical classifications.Each sheet uses coded identifiers to permit relational linkage between:Schools;Teachers;Classrooms; andLearners. Missing DataSome missing data are present in the dataset and primarily result from:Learner absenteeism;Incomplete reading attempts;School interruptions;Late lesson observations; andAdministrative scheduling constraints.Missing values are generally coded as:Blank cells;“NA”; orExplicit missing-value codes.Most missingness occurred in repeated learner assessments and implementation-monitoring variables rather than in baseline demographic information.The dataset includes quality-check documentation identifying:Missing records;Incomplete observations; andCorrective actions where feasible. Data Errors and ReliabilityPotential measurement error sources include:Learner hesitation timing inconsistencies;Observer scoring variation;Classroom noise during oral reading;Incomplete learner responses; andManual transcription errors.To reduce these risks:Assessors received standardized training;Calibration exercises were conducted;Double-scoring procedures were implemented; andData verification checks were performed.The Reliability_DoubleScoring sheet documents scorer agreement and absolute score differences.

Inter-rater reliability exceeded acceptable analytical thresholds for educational assessment research.Possible timing-related reading-assessment error margins may include minor variations of a few words per minute due to hesitation timing and scorer judgment. Ethical and Confidentiality ProceduresThe dataset is fully de-identified. No learner names or direct personal identifiers are included. Ethical procedures included:Teacher consent;Learner assent;Secure storage of assessment records; andRestricted access to identifiable source documents. Scientific Importance of the DatasetThis dataset is scientifically important because it integrates:Experimental educational intervention data;Literacy assessment outcomes;Classroom instructional-process measures;Implementation-fidelity documentation; andContextual school information from multilingual African classrooms.The dataset supports educational, linguistic, and implementation-science research focused on:Oral reading fluency;Structured pedagogy;Literacy instruction;Educational assessment;Reading development; andLow-resource educational systems.Its multilevel and longitudinal structure enables investigation of how classroom practices, instructional fidelity, and learner access to books influence literacy outcomes in multilingual schooling environments.