Normal Curve and Statistical Inference

Page 1

+1 SD = 68% of the data

+2 SD = 95% of data

+3 SD = 99.7%

* It is possible to have your data at 4 or 5 SD, but it's a tiny amount

Common for many values and metrics for "normal" or healthy people

Standard Scores

allows us to compare things by getting a common denominator (z-score, t-score)

ROM following two different stretching protocols

Percentiles- a standard score that creates a "common denominator" for the data

decile - 1-10 scale

48th percentile = 5th decile

Quartile

48th percentile = 2nd quartile

T scores = 10z + 50

T scores vs Z scores

EX in SPSS for score of 10

Z score = -0.59

T score = 10 (-0.59) + 50= 44.1

percentile= 26th percentile

Probability

Likelihood an event will occur at random

Never an "automatic" guarantee

Tied to the SD and the percentages within

P of data falling within + 1SD = 68%

P of data being < + 1 SD = 84%

P of data being > + 2 SD = 2.28%

Odds

P of the outcome (for or against)

Odds = Probability of event occurring/ probability of event NOT occurring OR

P/1-P

Page 2

Statistical Inference

sampling error

type of standard score that helps us infer population mean (mu)

Level of confidence & confidence intervals

interval is the range calculated from the level

Null hypothesis testing

our alpha level set before testing is the type I error rate. How comfortable are we with accounting for type I error

Interpreting p-value

Probability that if the null were true, we would have a % chance of seeing the results that we got or more extreme