This document provides guidance on several statistical topics:
1. It outlines the steps for hypothesis testing: stating hypotheses, standardizing observations, determining a decision rule, and comparing results to reach a conclusion.
2. It lists the key assumptions in linear regression models including that errors have zero mean and constant variance, follow a normal distribution, and are uncorrelated.
3. It clarifies that t-statistics and p-values in Excel assume a two-tailed test with a null hypothesis that the population parameter equals zero.
This document provides guidance on several statistical topics:
1. It outlines the steps for hypothesis testing: stating hypotheses, standardizing observations, determining a decision rule, and comparing results to reach a conclusion.
2. It lists the key assumptions in linear regression models including that errors have zero mean and constant variance, follow a normal distribution, and are uncorrelated.
3. It clarifies that t-statistics and p-values in Excel assume a two-tailed test with a null hypothesis that the population parameter equals zero.
This document provides guidance on several statistical topics:
1. It outlines the steps for hypothesis testing: stating hypotheses, standardizing observations, determining a decision rule, and comparing results to reach a conclusion.
2. It lists the key assumptions in linear regression models including that errors have zero mean and constant variance, follow a normal distribution, and are uncorrelated.
3. It clarifies that t-statistics and p-values in Excel assume a two-tailed test with a null hypothesis that the population parameter equals zero.
DNAQ too many times in past papers - read the question very carefully. You have reading time for this exam and there are no MCQ to worry about!
For CI, find half-width first. Then set up CI around the observation.
Define type I and type II error in terms of probability e.g. P(Type II error) = P(xbar>340|mu=342)
Hypothesis testing steps: 1. Hypotheses 2. Standardise observation to test statistic 3. Decision rule 4. Compare & conclude
Assumptions about disturbances: E(epsilon_i | x_i) = 0 Var(epsilon_i) = (sigma_epsilon)^2 epsilon_i~N Cor(epsilon_i, epsilon_j) = 0 for given i,j Other assumptions: Variance in X Simple Random Sampling Linear Model Plus, no perfect multicollinearity
"the t-stat and P-value in the EXCEL output are derived from two-tail tests with H0: associated population parameter = 0"
oh, and when analysing coefficients - it is the ceteris paribus marginal effect! REMEMBER THE CETERIS PARIBUS.
T-distribution tl;dr two different issues: 1. is sample mean distribution actually normal-shaped? 2. what is the underlying population variance?
(1) becomes a problem if underlying distribution is not normal. We fix this with CLT. (2) becomes a problem if we only know sample variance. We fix this by using t-distribution.
Our fixes work independently, but CANNOT be used at the same time. But CLT always takes precedence over t-distribution because it says BOOM! YOUR DISTRIBUTION IS NORMAL!