PSYC209 Notes Types of Tests PDF

Preview

Summary

Lectures notes on types of tests (Z, T, ANOVA, etc), Professor Leigh...

Description

Normal Distribuation ● Bell-shaped, unimodal and symmetric. ● Many variables are nearly normal, but none are exactly normal. Standardizing with Z-Scores ● (X-mean) / standard deviation ● Pam’s Score is (1800-1500) / 300 = 1 standard deviation above the mean. ● Jim’s Score is (24-21) / 5 = 0.6 standard deviations above the mean. ● Observations above or below 2 standard deviations from the mean are considered unusual. Percentile: ● The percentage of observations that fall below a given data point. ● Usually in a curve, it’s all the data to the left of an observation. ● In R- pnorm(mean =, sd=) ● 35.8-36/0.11 = -1.8 standard deviations ● Finding the exact probability = pnorm (z score, mean, sd) ● qnorm(percentile) = cutoff in terms of a zscore. ● 68, 95, 99.7 rule for SD of the mean. Parameter Estimation ● Point Estimates & Margin of Error Margin of Error ● Margin of sampling error is plus or minus a percent at a certain confidence level. ○ Ex. 41%+-2.9 at the 95% confidence level means that we’re 95% confident that 38.1 % - 43.9 % of respondents feel a certain way. Sampling Distributions ● Sampling distributions are never observed. ● Understanding the sampling distribution can help us understand the point estimates we observe. Central Limit Theorem ● The more samples we take, the more of a normal distribution it is. ● BUT ○ Random assignment must be used. ○ There should be at least 10 expected successes and 10 expected failures. When P is Unknown ● When we don’t know the value of p, we use p-hat. ● P-Value in R - pnorm(value, mean, sd) or pnorm(z-score). ● Subtracting the P-Value from 1 gives you an idea of the probability. When NP is Small ● Multiply NP and see if it fits success/failure condition. ● Success/Failure condition is 10. Confidence Intervals ● A range of values for a population parameter is called a confidence interval. ● We have a better chance of capturing the range of values with a confidence interval than a point estimate. ● SE=sqrt(p(1-p)/n)) Two groups, one variable numeric, one variable categorical. ● ANOVA Test -> More than two groups, one variable numeric, one variable categorical. ● Regression -> Two groups, both variables numeric. Correlation Coefficient: ● Next to the variable, under intercept in the table (b). ● Predicts the change in one unit of that variable, the other variable goes up as much as the coefficient when the first variable goes up one unit. ○ Ex. if the variables are resting heart rate and cholesterol and the c.c. is 0.36, cholesterol goes up 0.36 for every 1 unit increase in heart rate. ● This is called the point shift (a one-unit shift in variable A is associated with a 0.36 pointshift in variable B). Summary Table ● Predicted first, then predictor. ● Y ~ X. Both Numerical = Regression. One Numerical, One Categorical, More than Two Levels = ANOVA. One Numerical, One Categorical, Two Levels = T-Test....