The students made Powerpoint presentations about the projects developed during the course. The projects were carried out using data from the "Tobacco Database", owned by the Primary Health Care Centre ACES Arco Ribeirinho, Setúbal, district of Lisbon. The main research question was: what are the most important variables (at both individual and community levels) to predict cigarette use? Students also wrote the respective essays.

Correlation between two populations: Pearson correlation coefficient; Spearman correlation coefficient. Explaining the framework underlying each of these measures.
These methodologies were applied to the databases "Control levels after heart attack" and "Evaluating risk for cardiovascular diseases in the Portuguese population". The analyses were made by using the spreadsheet Microsoft Excel and the IBM SPSS Software.

One-Way Analysis of Variance to compare the means of three or more populations. Evaluating the assumptions underlying this methodology. Multiple comparisons: orthogonal contrasts; post-hoc tests (as, for instance, LSD test, Tukey's LSD test, Duncan test, Bonferroni test, etc.).

An alternative approach to compare the means of three or more populations when the homogeneity assumption is not verified: the Brown-Forsythe test; and the Welch test.

 
All these methodologies were applied to the databases "Control levels after heart attack" and "Evaluating risk for cardiovascular diseases in the Portuguese population". The analyses were made by using the spreadsheet Microsoft Excel and the IBM SPSS Software.

By the end of the lesson, the logistic regression model (studied in the course "Value in Health Care") was applied to the variables of the database "Evaluating risk for cardiovascular diseases in the Portuguese population" in order to characterise the individuals who are at risk of a heart attack. The analysis was conducted using the IBM SPSS Software.

Summarising the main points of the last lesson. Explaining in detail the concept of the p-value.

The Binomial test: an exact alternative test to the large-sample hypothesis test to a single population proportion.

Hypothesis testing to compare the parameters of two independent populations: the difference between two population means (for the cases of homogeneous, and non-homogeneous populations); the ratio between two population variances (which can also evaluate the homogeneity of the populations); the difference between two population proportions. Application of these methodologies to the databases: "cholesterol levels after heart attack" and "Evaluating risk for cardiovascular diseases in the Portuguese population".

Paired sample t-test. Pared large-samples test (z-test). Using the SPSS to run the data analysis. An example: cholesterol levels after a heart attack. Comparing the control group to the group with the disease.

All the analyses were developed using the spreadsheet Microsoft Excel and the IBM SPSS Software. Once again, the transformation of the p-values from the bilateral tests (the outputs from the SPSS) into the unilateral versions was carried out whenever appropriate.

A review of the most important probability distributions (highlighting their shapes), with special emphasis to the Gaussian distribution. The most important sampling distributions. The Central Limit Theorem.

Confidence intervals: a population mean; a population variance (and also standard deviation); a population proportion. Application to the database "Control levels after heart attack" and "Evaluating risk for cardiovascular diseases in the Portuguese population" (addressed in Lesson 6).

Hypothesis testing:

- An introduction;

- Basic definitions: the meaning of a hypothesis in the statistical context; formulating a hypothesis test: the null hypothesis versus the alternative hypothesis, unilateral and bilateral tests; test statistic; rejection region; p-value;

- The analogy between the philosophy underlying hypothesis testing and the framework of the criminal trial;

- Different types of errors: type I error; type II error. The power of a test.

An outline on how to conduct a hypothesis test (one population), with emphasis to the large-sample test and t-tests.

Hypothesis test on a single mean, a single variance, and a single proportion. Conducting hypothesis tests to the relevant variables of the databases cited above, namely: single population mean; single population proportion; single population proportion. Analysing and discussing the main conclusions in the context of the health quality assessment. The analyses were developed with the spreadsheet Microsoft Excel and IBM SPSS Software. Emphasis was given to the transformation of the outputs obtained from SPSS (p-values associated to bilateral tests) into the unilateral versions.