# HSCI 307 Lecture Notes - Lecture 11: Alternative Hypothesis, Standard Error, Statistical Hypothesis Testing

59 views1 pages

Published on 24 Jan 2018

School

Department

Course

Professor

Types of data

Data: numbers that represent some feature of the thing you are measuring

Measurement: the process of assigning numbers to the thing

Categories NOIR

Nominal: mutually exclusive & exhaustive

Ordinal: nominal + in order

Interval: nominal + in order + equally spaced

Ratio: nominal + in order + equally spaced + a

meaningful zero

Discrete or Continuous

Discrete: who number,

no value between the

whole numbers

Continuous: any value

Pareto charts

•

Stem and leaf plots (SLP)

•

Box plots

•

Scatter plots

•

Graph data with some common tools that allow you to see data patterns

Each one

Describe data numerically

Correlation coefficient: straight line relationship between two variables

•

Descriptive statistics: happen to describe properties of the data and their distribution

One group of data

Minimum ~ maximum ~ range - a measure of variation

Median: two equally frequent parts

Decimal place (dp) 0.1

•

Mean:

Standard deviation: how much each observation varies from the mean

Three number report: sample size, median, IQR (interquartile range)

Lifetimes of people and electronic devices (computer)

○

Skewed left or has a tail that is longer on the left than it is on the right

•

Length of stay of patients in a hospital

○

Skewed right or has a longer tail on the right than on the left

•

Measure of symmetry - skewness (-1 ~ 1)

Five-number summary

Two groups of data

Correlation

The regression equation

Confidence intervals (CI)

Lower bound and upper bound - two-sided confidence interval

•

Parameter: a characteristic of a distribution of a variable in a

population that was used to generate the sample

•

Ideally - sample should be selected at random/ created as

part of a RCT - to ensure the groups represent the

population (otherwise bias)

○

data are independent sample from the population

○

Normally distributed

○

RIN for random, independent, normally distributed

•

CI is a statement about a characteristic in a population that was

used to generate the sample - Statistical inference, rather than a

descriptive statistic

•

95% - 95% confidence coefficient (19 times out of 20)

•

95% CI: 95% of the time the true value in the population will be

contained in the interval if the study were to be repeated many times

Assumptions

Standard error (SE)

Criterion value

Approximate CI

Difference between the means

Equal variances

Test hypothesis

Step 1: compute a test statistic

•

Step 2: compute a P-value from a suitable referent distribution

•

Step 3: compare the P-value to a criterion standard - α(0.05 - 95%)

•

P-value < α, hypothesis is rejected - statistically significant

○

P-value > α, hypothesis cannot be rejected

○

Null hypothesis, one expects to be rejected

○

Alternative hypothesis

○

Two-tailed test

○

Step 4: interpretation:

•

Hypothesis testing: another way to draw an inference about the population from which the sample of

data was collected

Quantitative methods: analysis

Monday, March 20, 2017

20:18

week 11 chpt 11 quantitative analysis Page 1

## Document Summary

Data: numbers that represent some feature of the thing you are measuring. Measurement: the process of assigning numbers to the thing. Graph data with some common tools that allow you to see data patterns. Interval: nominal + in order + equally spaced. Ratio: nominal + in order + equally spaced + a meaningful zero. Discrete: who number, no value between the whole numbers. 95% ci: 95% of the time the true value in the population will be contained in the interval if the study were to be repeated many times. Descriptive statistics: happen to describe properties of the data and their distribution. Correlation coefficient: straight line relationship between two variables. Minimum ~ maximum ~ range - a measure of variation. Standard deviation: how much each observation varies from the mean. Three number report: sample size, median, iqr (interquartile range) Measure of symmetry - skewness (-1 ~ 1)