Setting up your dataset
In Variable View, name each variable, set the type, label it clearly and define the measurement level (nominal, ordinal or scale). Add value labels for coded categories, such as 1 = Male, 2 = Female. In Data View, each row is a respondent and each column a variable.
Use short names without spaces and a consistent coding system. Keep a codebook so you can explain every variable later.
Cleaning and coding
Check for data entry errors by running frequencies and looking for values outside the expected range. Handle missing data in a way you can justify, and note how many responses you removed and why.
Recode reverse-worded items and compute composite scores (for example the mean of items in a scale) with Transform > Compute Variable. Check for outliers using boxplots or standardised scores, and decide whether to keep, adjust or remove them with a clear reason.
Descriptive statistics
Use Analyze > Descriptive Statistics > Frequencies for categorical variables such as gender or age group, and Descriptives for scale variables to get mean and standard deviation. Report sample characteristics first, usually in a table.
For Likert scale variables, report the mean and standard deviation, and be careful about how you interpret ordinal data. Include the sample size for each analysis, since missing data may change it.
Reliability and normality
Cronbach's alpha (Analyze > Scale > Reliability Analysis) measures the internal consistency of a multi-item scale. A value of about 0.70 or higher is commonly accepted, though acceptable levels vary by field.
To check normality, review skewness and kurtosis values, histograms and Q-Q plots, and consider the Shapiro-Wilk test, which can be sensitive in large samples. Use the result to decide between parametric and non-parametric tests, and explain your reasoning.
Correlation and regression
Pearson correlation measures the strength and direction of a linear relationship between two scale variables (use Spearman for ordinal or non-normal data). Report r, the p-value and N.
Multiple regression tests how well several predictors explain an outcome. Check assumptions such as linearity, independence, normality of residuals, constant variance and multicollinearity (VIF). Report R-squared, the F-test and each predictor's unstandardised and standardised coefficients with significance.
Group comparisons (t-test, ANOVA)
An independent-samples t-test compares the means of two groups, such as male and female students. One-way ANOVA compares three or more groups, followed by post-hoc tests (for example Tukey) if the overall result is significant. For related measures use paired t-tests or repeated-measures ANOVA.
Check assumptions such as normality and equality of variances (Levene's test). Where assumptions fail, consider non-parametric alternatives such as Mann-Whitney U or Kruskal-Wallis.
Reporting SPSS output
Present results in tables that follow your referencing style (for example APA), and describe the key findings in the text instead of pasting raw SPSS output. Report test statistics, degrees of freedom, p-values and effect sizes, for example: t(98) = 2.45, p = .016.
Say whether each hypothesis is supported, then interpret the result in context. Do not report p-values as proof: explain practical meaning too. If you are unsure which test fits your data, ask your supervisor or an SPSS tutor before running multiple tests.