Advanced SPSS Data Analysis Training Course

Participants can import, reshape and transform multi-source data for advanced analysis.

• Importing Excel, CSV, database and fixed-width files with validated formats
• Wide-to-long restructuring using VARSTOCASES for repeated-measures datasets
• Long-to-wide restructuring using CASESTOVARS for reporting and modelling
• Computing derived variables with conditional logic, date functions and recodes
• Merging files through keyed joins, append operations and duplicate checks
• Weight variables, split files and selection filters applied transparently

Participants can diagnose data-quality risks and apply justified treatment rules.

• Frequency tables and cross-tabulations identify invalid codes and impossible values
• Missing-data patterns assessed through percentages, indicators and diagnostic tables
• MCAR reasoning, MAR assumptions and consequences of listwise deletion
• Multiple imputation settings, pooled estimates and imputation diagnostics
• Univariate outliers examined with z-scores, boxplots and percentile rules
• Multivariate outliers screened using Mahalanobis distance and influence statistics

Participants can use exploratory methods to identify patterns before formal modelling.

• Distribution shape assessed with histograms, Q-Q plots and skewness statistics
• Explore procedure compares groups using descriptive summaries and confidence intervals
• Scatterplot matrices reveal nonlinear relationships, clusters and influential observations
• Chart Builder creates annotated bar charts, distributions and interaction plots
• Standardised scores support comparisons across variables with different scales
• Data transformations address skewness through logarithmic, square-root and rank methods

Participants can create reusable SPSS Syntax for auditable analytical workflows.

• Syntax Editor commands replace repeated menu-based analytical procedures
• Comment blocks document assumptions, inputs, transformations and decision rules
• DO IF, LOOP and VECTOR structures automate conditional data preparation
• DEFINE macros standardise repeated procedures across variables and datasets
• OMS commands capture selected output tables into structured files
• Temporary transformations prevent unintended changes to master analysis datasets

Participants can assess sample adequacy and interpret uncertainty in analytical results.

• Probability and non-probability sampling implications for statistical inference
• Sampling error expressed through standard errors, confidence intervals and margins
• Power analysis concepts: alpha, effect size, sample size and beta
• Type I and Type II errors linked to operational decisions
• Multiple testing risk managed through planned comparisons and correction methods
• Effect sizes reported using Cohen's d, eta squared and odds ratios

Participants can compare group outcomes while controlling covariates and error rates.

• One-way ANOVA assumptions assessed through residual plots and Levene's test
• Factorial ANOVA models main effects and interaction effects
• ANCOVA adjusts group comparisons for baseline or confounding covariates
• Estimated marginal means support adjusted comparisons and interaction interpretation
• Post-hoc procedures selected using variance equality and comparison objectives
• Partial eta squared and confidence intervals communicate practical significance

Participants can model correlated observations from longitudinal and repeated-measures studies.

• Repeated-measures ANOVA evaluates within-subject change across measurement occasions
• Mauchly's test and Greenhouse-Geisser correction address sphericity violations
• Linear mixed models accommodate incomplete repeated observations and clustering
• Fixed effects, random intercepts and random slopes specified deliberately
• Covariance structures compared using information criteria and residual diagnostics
• Predicted trajectories communicate intervention, cohort and time effects

Participants can build, diagnose and interpret multiple linear regression models.

• Model specification links dependent variables, predictors, controls and hypotheses
• Enter, hierarchical and stepwise selection methods compared critically
• Residual plots assess linearity, homoscedasticity, normality and independence
• Variance inflation factors identify multicollinearity among predictor variables
• Cook's distance and leverage values identify influential observations
• Standardised coefficients, confidence intervals and adjusted R-squared guide interpretation

Participants can model binary outcomes and communicate classification performance.

• Binary logistic regression models event probability from predictor variables
• Odds ratios convert regression coefficients into decision-relevant effect measures
• Categorical predictors coded with reference categories and indicator variables
• Hosmer-Lemeshow statistics, residuals and influence diagnostics assess model fit
• Classification tables, sensitivity and specificity evaluate prediction performance
• ROC curves and area under curve compare candidate models

Participants can analyse ordered and multi-category outcomes using appropriate regression models.

• Ordinal regression models ordered satisfaction, risk-rating and maturity outcomes
• Proportional odds assumption tested before interpreting ordinal model estimates
• Multinomial logistic regression compares nominal outcome categories against references
• Parameter estimates translated into relative risk ratios and predicted probabilities
• Goodness-of-fit tests and pseudo R-squared values assessed cautiously
• Interaction terms identify differing predictor effects between population groups

Participants can select generalised linear models for non-normal outcome distributions.

• Generalised linear model components: distribution, link function and linear predictor
• Poisson regression models event counts, incidents and transaction frequencies
• Negative binomial models address overdispersion in count outcome data
• Exposure offsets standardise event rates across unequal observation periods
• Gamma models handle positive, right-skewed cost and duration outcomes
• Deviance, Pearson residuals and information criteria compare fitted models

Participants can apply distribution-free tests and bootstrap inference where assumptions fail.

• Mann-Whitney and Wilcoxon tests compare independent and paired ranks
• Kruskal-Wallis tests examine group differences without normality assumptions
• Friedman tests assess repeated ranked observations across multiple conditions
• Spearman's rho measures monotonic association for ordinal or skewed variables
• Bootstrap confidence intervals estimate stability for statistics and model parameters
• Permutation-test logic distinguishes observed effects from random allocation

Participants can identify latent dimensions and reduce correlated survey variables.

• Correlation matrices, KMO and Bartlett's test assess factorability
• Principal components analysis reduces variables for descriptive data reduction
• Exploratory factor analysis estimates latent constructs behind survey items
• Eigenvalues, scree plots and parallel analysis guide factor retention
• Varimax and oblimin rotations clarify factor structure and correlations
• Factor scores support downstream regression, segmentation and reporting

Participants can evaluate the reliability and internal structure of multi-item scales.

• Cronbach's alpha interpreted alongside item-total correlations and scale purpose
• McDonald's omega considered when tau-equivalence assumptions are doubtful
• Reverse-coded items checked before calculating composite scale scores
• Item deletion decisions balance reliability, content coverage and comparability
• Inter-rater reliability assessed with Cohen's kappa and intraclass correlation
• Scale scoring rules document missing-item thresholds and response ranges

Participants can develop and assess statistically credible customer or population segments.

• Standardisation prevents high-variance variables dominating distance-based clustering
• Hierarchical clustering explores candidate segment structures using dendrograms
• K-means clustering refines segment membership around centroid solutions
• Distance measures selected according to variable type and analytical purpose
• Cluster profiles compare behaviours, needs, risks and demographic characteristics
• Segment stability assessed through split samples and alternative starting seeds

Participants can assess classification models and compare discriminant and logistic approaches.

• Discriminant analysis classifies groups using linear combinations of predictors
• Box's M and covariance assumptions checked before model interpretation
• Canonical discriminant functions identify variables separating defined groups
• Cross-validation estimates classification performance on held-out observations
• Confusion matrices distinguish false positives, false negatives and correct classifications
• Logistic and discriminant models compared for assumptions and operational use

Participants can prepare time-series data and evaluate forecasts for operational decisions.

• Time-series structure separates trend, seasonality, cycles and irregular variation
• Date definitions and periodicity settings prepare monthly, quarterly and weekly series
• Autocorrelation and partial autocorrelation plots identify temporal dependence
• Exponential smoothing models estimate level, trend and seasonal components
• ARIMA identification uses differencing, autoregressive and moving-average terms
• MAPE, RMSE and residual diagnostics evaluate forecast accuracy

Participants can convert SPSS output into concise, decision-ready analytical reporting.

• Results narratives distinguish statistical significance, practical significance and uncertainty
• Model tables present coefficients, intervals, fit measures and sample sizes
• Predicted probabilities and marginal means translate models for non-specialists
• Chart annotations identify comparisons, units, timeframes and material caveats
• Method notes document exclusions, transformations, assumptions and validation checks
• Recommendations connect analytical evidence to decisions, actions and monitoring measures

Participants can build, review and present a complete SPSS analysis addressing a defined organisational question.

• Problem statement defines decision context, stakeholders, outcomes and analytical scope
• Source data prepared through documented cleaning, transformation and quality checks
• Statistical method selected from research questions, measurement levels and assumptions
• SPSS Syntax records reproducible preparation, modelling and output generation steps
• Peer review tests interpretation, diagnostics, limitations and reporting clarity
• Management presentation communicates findings, recommendations, risks and next actions