Sumários
28 Outubro 2026, 11:00
•
Yiu Lim LUI
We reviewed exercises 5.1, 5.2, and 5.4 together and examined the Portuguese CPI data.
27 Outubro 2026, 09:30
•
Yiu Lim LUI
Definition and Purpose: Index numbers are statistical measures that express the relative change of a variable or group of related variables over time or across locations relative to a base value (typically set to 100), making them essential for tracking inflation, economic policy, and comparing different data series.
Simple Index Numbers: Simple indices measure the relative change of a single variable against a base period, calculated using either a ratio form ($I_t = x_t / x_0$) or a percentage form ($I_t = (x_t / x_0) \times 100$).
Synthetic (Aggregate) Indices: These combine multiple variables into a single measure—either as composite indices when variables share the same units or complex indices when units differ—using methods like arithmetic, geometric, or harmonic means.
Weighted Indices (Laspeyres and Paasche): When items have unequal importance, weighted composite indices like Laspeyres (using base-period weights, which tend to overstate price increases) and Paasche (using current-period weights, which tend to understate them) are used, alongside combined measures like the Fisher Ideal index to reduce bias.
21 Outubro 2026, 11:00
•
Yiu Lim LUI
We used Excel to work through Exercises 3.1 and 4.1 together.
20 Outubro 2026, 09:30
•
Yiu Lim LUI
Contingency Tables and Independence: A contingency table summarizes the frequency distribution of two categorical variables. Two variables are independent if the proportion of each cell matches the product of its row and column marginal proportions.
Pearson's Chi-Square Statistic: This statistic measures the discrepancy between the observed frequencies and the expected frequencies assuming independence. A higher value indicates a larger departure from independence.General Association Measures for Multi-Category Tables: Measures such as the contingency square, the contingency coefficient, Tschuprow's coefficient, and Cramer's V are derived from the chi-square statistic to account for sample size and table dimensions.
Association Measures for Two-by-Two Tables: For binary variables, specialized measures like the Phi coefficient, Yule's Q, and Yule's Y are used. Unlike general measures, these binary coefficients indicate both the strength and the positive or negative direction of the association.
Sensitivity to Marginal Distributions: While the Phi coefficient changes significantly when sample sizes or marginal distributions are heavily skewed, Yule's Q remains completely constant and stable despite severe sample size imbalances.
14 Outubro 2026, 11:00
•
Yiu Lim LUI
We continued the discussion of linear regression. We finished Exercise 4.1 and 4.2.