Quantitative Qualitative Data Analysis with R Python Training Course

Participants can apply ethical, privacy and quality controls throughout an analysis project.

• Informed consent requirements for surveys, interviews and recorded discussions
• Personally identifiable information classified before collection, storage and sharing
• Data minimisation applied to forms, extracts and analytical datasets
• Quality dimensions assessed through accuracy, completeness, timeliness and consistency
• Version control conventions for datasets, scripts, codebooks and reports
• Disclosure risk checked before publishing small groups and sensitive findings

Participants can use R scripts to import, inspect and transform tabular datasets.

• RStudio projects organise scripts, data files, outputs and documentation
• Tidyverse packages installed and loaded for repeatable data preparation
• CSV and Excel files imported with explicit variable type checks
• Data frames inspected using structure, summaries, dimensions and sample records
• dplyr verbs filter, select, mutate, arrange and summarise operational data
• R scripts annotated with assumptions, sources and transformation decisions

Participants can use Python notebooks to inspect, transform and analyse structured data.

• Jupyter notebooks combine executable Python code, commentary and results
• pandas DataFrames created from CSV, Excel and database extracts
• Missing values identified using isna, counts and field-level profiling
• Boolean filters isolate records meeting defined business or programme criteria
• Groupby operations calculate counts, rates, averages and aggregated measures
• Reusable functions package recurring cleaning rules and analytical calculations

Participants can design ODK forms that capture valid field data for later analysis.

• XLSForm structure uses survey, choices and settings worksheets correctly
• Question types selected for text, numbers, dates, locations and media
• Relevance expressions implement skip logic based on prior responses
• Constraints prevent invalid ranges, impossible dates and incomplete identifiers
• Repeat groups capture household members, transactions or repeated observations
• ODK Central workflows manage form publication, submissions and data exports

Participants can produce a documented, analysis-ready dataset from multiple sources.

• Data dictionaries define variable names, labels, types and permitted values
• Duplicate records detected through identifiers, dates and matching fields
• Missing-data patterns profiled before deletion, imputation or exclusion decisions
• Dates, currencies and categories standardised using documented transformation rules
• Join keys tested before merging survey, operational and reference datasets
• Cleaning logs record source issues, corrections and unresolved data limitations

Participants can summarise distributions, patterns and exceptions using appropriate descriptive methods.

• Frequency tables reveal category distributions, missingness and unusual response patterns
• Central tendency compared using mean, median and mode appropriately
• Dispersion measured through range, variance, standard deviation and interquartile range
• Cross-tabulations examine relationships between categorical variables and population groups
• Histograms, boxplots and scatterplots expose skewness, outliers and clustering
• Segmentation compares results by geography, customer type, service channel or cohort

Participants can select, run and interpret common statistical tests for business and programme questions.

• Sampling frames distinguish target populations, accessible populations and selected samples
• Confidence intervals express plausible ranges around estimated population values
• Null and alternative hypotheses framed before selecting statistical tests
• t-tests compare means between independent groups or matched observations
• Chi-square tests assess association between categorical variables and outcomes
• p-values, effect sizes and practical significance interpreted together

Participants can build and interpret regression models while checking core assumptions.

• Linear regression estimates relationships between continuous outcomes and predictor variables
• Logistic regression models binary outcomes such as default, uptake or completion
• Predictor selection considers theory, data availability and multicollinearity diagnostics
• Residual plots assess linearity, normality, independence and constant variance
• Model coefficients translated into operational implications and decision thresholds
• Train-test splits assess predictive performance without overstating model certainty

Participants can prepare qualitative material for credible, traceable analysis.

• Interview guides use open questions, probes and neutral wording
• Purposive sampling identifies participants with relevant experience and perspectives
• Transcription conventions preserve speaker turns, pauses and significant contextual notes
• Anonymisation removes names, locations and other direct identifying details
• Case classifications capture participant attributes, sites and collection metadata
• Reflexive memos document researcher assumptions, decisions and emerging observations

Participants can structure an NVivo project for organised source review and retrieval.

• NVivo projects organise transcripts, documents, media, cases and classification sheets
• Source folders separate interviews, focus groups, policy documents and observations
• Imported transcripts checked against originals for speaker labels and formatting
• Cases created for participants, sites, teams or customer groups
• Classification sheets import attributes for comparison across defined participant groups
• Memos and annotations retain context beside source material and coded passages

Participants can create and apply a transparent coding framework to qualitative evidence.

• Deductive codes derived from research questions, theory and programme logic
• Inductive coding captures unanticipated issues emerging from source material
• Codebooks define code names, inclusion criteria, exclusions and examples
• Hierarchical nodes organise parent themes, subthemes and descriptive codes
• Double-coding procedures compare interpretation across analysts and reconcile differences
• Coding density reviewed to identify thin evidence and over-broad categories

Participants can develop themes and integrate them with quantitative findings.

• Braun and Clarke thematic analysis moves from codes to candidate themes
• Theme definitions distinguish patterns, mechanisms, exceptions and illustrative quotations
• Matrix coding queries compare themes across groups, locations and classifications
• Joint displays align statistical results, qualitative explanations and evidence gaps
• Convergence coding identifies agreement, complementarity, dissonance and unanswered questions
• Mixed-methods conclusions state how each evidence source informed interpretations

Participants can create a Power BI model that supports reliable reporting and analysis.

• Power Query profiles, filters and transforms source data before loading
• Star schemas separate fact tables from dimensions and reference attributes
• Relationships configured using appropriate cardinality, direction and active status
• Date tables support monthly, quarterly and year-to-date time intelligence
• DAX measures calculate totals, rates, variances and rolling period results
• Model validation reconciles dashboard totals against approved source reports

Participants can design Power BI dashboards that support specific management decisions.

• Dashboard purpose defined through users, decisions, measures and action triggers
• Visual selection matches comparisons, trends, distributions, maps and composition questions
• KPI cards display current value, target, variance and reporting period
• Slicers and drill-through pages support focused exploration without visual clutter
• Annotations explain material changes, data limitations and recommended actions
• Accessibility checks cover colour contrast, labels, reading order and alternative text

Participants can validate analytical outputs and package work for review or reuse.

• Reproducible workflows separate raw data, cleaned data, scripts and outputs
• R Markdown and Jupyter exports combine methods, code, tables and narrative
• Validation checks compare totals, ranges, duplicates and category values
• Sensitivity analysis tests whether conclusions change under reasonable assumptions
• Peer review checklists assess logic, methods, calculations, coding and interpretation
• Issue logs assign corrections, decisions, owners and completion status

Participants can present findings, limitations and recommendations for a decision-making audience.

• Executive summaries state decision context, principal findings and recommended actions
• Methods sections document sources, sampling, transformations, coding and statistical tests
• Findings sequence evidence from headline results to supporting detail
• Charts and quotations selected to substantiate rather than decorate conclusions
• Limitations distinguish data quality concerns from analytical uncertainty and scope
• Recommendations specify owners, actions, timing, measures and evidence requirements

Participants can build, review and present a complete mixed-methods analysis from raw data to decision briefing.

• Project brief defines a decision question, users and required outputs
• ODK, operational, survey and interview sources assessed for fitness
• R or Python scripts create a documented analysis-ready dataset
• NVivo coding produces themes supported by traceable source evidence
• Statistical findings and qualitative themes integrated through a joint display
• Power BI dashboard presents measures, exceptions, implications and recommended actions
• Peer review tests analytical logic, reproducibility, evidence quality and presentation