Stats

Question 0

Ordinal

Answer 0

Numbers indicate rank order of observations (greater then-less relationships)
Measured by median, mode

Question 1

Nominal

Answer 1

Numerals represent category labels only (sex, nationality)

Measured by mode

Question 2

Interval

Answer 2

Equal intervals between numbers but not related to true zero; do not represent absolute quantity (calendar years, degrees centigrade)
Measured by median, mean

Question 3

Ratio

Answer 3

Numbers represent unit with equal intervals, measured from true zero
Median, mean, mode

Question 4

What type of grading is available to measure reliability?

Answer 4

Cohens kappa (.40-.75 excellent); ICC (.59 - .75 good) which is based on an ANOVA
- used to be measured by Pearson's or Spearman however these measured covariant, NOT agreement

Question 5

Define construct validity

Answer 5

The extent to which a tool measures what it claims to measure
Ie: does weight measure overall health?

Question 6

Define concurrent validity

Answer 6

Relationship between test 1, test 2; typically gold standard

Question 7

Define internal validity

Answer 7

More control of a study, the better

Question 8

Define external validity

Answer 8

Generalization from the sample studied to the general population

Question 9

Define covariant

Answer 9

Extraneous phenomenon that affects dependent variable that the researcher cannot control
It is not of interest to the researcher

Question 10

Type I error

Answer 10

False positive rate (alpha)
Set before the study by researchers
Error concluding that your sample stats show there is a significant change (reject null) when there really is not

Question 11

Type II error

Answer 11

False negative rate (beta)
Error of not rejecting a “false” null hypothesis
Concluding there is NOT a difference between treatment and control when there really is

Question 12

What is power and what is our goal?

Answer 12

The ability of our statistical tests to find stat difference when they exist
1-beta (type II error) = 80%

Question 13

What can impact power?

Answer 13

Sample size, variance (between and within groups), effect size

Question 14

Define effect size

Answer 14

Measure of the degree to which the null hypothesis is false
An expression of the magnitude of the difference between two treatments
.2-.4 small; .5-.7 moderate, .8 to >2.0 large

Question 15

Define +LR scale and it’s shift in probability

Answer 15

1-2 alter P to a small degree
2-5 generate small but sometimes important shift in P
5-10 generate moderate shifts in P
> 10 generate large and often conclusive shifts in P

Question 16

Define LR- ranges and interpretation

Answer 16

.5-1 alter P to small degree
.2-.5 generate small but sometimes important shifts in P
.1-.2 generate moderate shifts in P
<.10 generate large and often conclusive shifts in P