This is the complete study package for Statistical Analysis using SPSS (STS-207). Its nine headings are nine bodies of statistics, and eight of them are taught in full elsewhere on this site — with the formulae derived and the arithmetic worked by hand. None of that is repeated. What this course contributes, and what these pages therefore contain, is the SPSS procedure: which dialog, which syntax, which options change the answer, what the output tables are called, and which footnote has to be acted on.
| Prescribed area | The theory, on this site |
|---|---|
| 1. Data entry, import and export | specific to SPSS — written out on the practical page |
| 2. Data visualization | STS-108, section 4 — which diagram suits which variable |
| 3. Descriptive statistics | Descriptive Statistics |
| 4. Parametric tests | Inferential Statistics and Estimation Theory (STS-201) |
| 5. Non-parametric tests | STS-108, section 8 |
| 6. Design and analysis of experiments | Design and Analysis of Experiments and STS-203 |
| 7. Regression analysis | Statistical Methods, Unit 4 and Linear Algebra and Linear Models, Unit 4 — but model selection, logistic regression and probit analysis are not on the site, so they are worked in full here |
| 8. Multivariate data analysis | Multivariate Analysis (STS-202), Unit 3 and Unit 4 — discriminant, principal components, factor, scaling and cluster analysis, every eigenvalue computed |
| 9. Statistical quality control | Statistical Quality Control, Units 2 and 3 |
Able to carry out the statistical analysis and write a statistical report using SPSS for any data set.
The three windows and the Measure column that decides which procedures a variable may
enter; syntax rather than menus, and why; the charts, including the two SPSS has no button
for; the three descriptive procedures and which gives the mode; Levene before the \(t\);
every design through one UNIANOVA command, where leaving an interaction out
pools it into the error; all eight subsets of a three-predictor regression scored by
adjusted \(R^{2}\), Mallows \(C_p\), AIC and BIC, with forward, backward and stepwise run
step by step; logistic against probit on the same outcome; the five multivariate methods and
the three warnings SPSS will let you walk past; and the control charts.
The prescribed objective and the nine-item list of practicals, as printed.
The practical page uses the same twenty records as STS-108, Data Handling using R, with one column added. That is deliberate: the output of the two packages can be compared line for line on identical data, and a figure that differs is either a convention worth knowing about or a mistake worth finding. Where the two differ by convention — SPSS's sample-adjusted skewness against the moment coefficient, for instance — the practical page says so and gives both.
STS-105 Statistical Methods using Python completes the set from the other end, building the same methods from arithmetic with no package at all. Between the three papers a student sees every method three times: derived, scripted, and driven through a menu.