Flashcards in Round 2 Lecture 1 Deck (22)

Loading flashcards...

1

##
How to Determine Sample Size

###
1) Heuristics ( 100/group, Minimum of 20 to 50 per subgroup, Disproportionate sampling)

2) Budget constraints( Sample size = total budget/cost per subject)

3) Previous Research

4) Formal Statistics

2

## precision

### the size of the estimating interval

3

## confidence

### the degree of certainty that the true value falls within the estimating interval

4

## formal statistics

### trade off btw precision and confidence

5

## Confidence interval for a population mean is:

### X + Z* (s / sq. rt n)

6

## T =

###
margin of error you are willing to tolerate, i.e.,

zone around sample mean where population mean lies, or half CI width

7

## T = (equation)

### z * (s / sq. rt. n)

8

## Three pieces of information are needed for estimating for a mean sample size

###
1. Margin of error T (zone around sample mean where population mean lies)

2. Your confidence level

For 90% confidence, z = 1.645 For 95% confidence, z = 1.96 For 99% confidence, z= 2.575

3. Estimate of the population standard deviation (SD) s

9

## Estimating the population SD

###
Review prior research Conduct a pilot study

Use conservation guesses

10

## conservation guess for 5-item rating/Likert scale

### use s=1.4

11

## conservation guess for 7-item rating/Likert scale

### use s=2

12

## conservation guess for 9, 10, 11-item rating scale

###
use s=3

13

## conservation guess for responses measured by ratio scales

### use s=range/4

14

##
Three pieces of information are needed for estimating a proportion sample size

###
1. Margin of error T

2. Confidence level

For 90% confidence, z = 1.645 For 95% confidence, z = 1.96 For 99% confidence, z= 2.575

3. Estimate of the population proportion p

15

##
Estimating the Population Proportion p

###
Review prior research

Find secondary data (e.g., US census) Conduct a pilot study

Make an informed guess

Use the most conservative guess: p=0.5

16

## Sample Size Required for a Survey with Multiple Questions

###
-calculate ideal sample size for each question

-go with highest among all questions and the most critical questions

17

## total sampling elements (TSE) =

### n / { (1- BCI) (1-U) (1-R) (1-I) }

18

## BCI =

### proportion that are expected to have bad contact

19

## U =

### proportion that are expected to be unreachable

20

## R =

### proportion that are expected to refuse to participate

21

## I =

### proportion that are expected to be ineligible

22