Descriptive Statistics

..Download File..

Descriptive Statistics

Data is plural.
Datum is singular
All data is not equal.

Measurement systems:
1. Nominal: has the property of identity (0=female, 1=male)
2. Ordinal: has the properties of identity and order (ranked data – the units are not even in amount of property)
3. Interval: has the properties of identity, order, and additivity (the units are equal in amount of property)
4. Ratio: has the properties of identity, order, additivity, and absolute zero (like 0 degrees Kelvin)

Measures of Central Tendency
• Mean: average
• Median: middle most number
• Mode: the most common number

In a normal distribution, the mean, median and mode are the same (50th percentile). Most items in nature are equally distributed. Income is NOT normally distributed.

Measures of Variability
• Standard deviation (extent to which the scores cluster around the mean)
• Variance (SD squared)
• Range (highest minus lowest scores)

-2 standard dev

-1 standard dev

0

+1 standard dev

+2 standard dev

70

85

100 (WISC-IV)

115

130

MR (with other behaviours)

68 % of scores fall here

Gifted Education.

2nd percentile

16th percentile

50th percentile

84th percentile

98th percentile

96% of scores fall here

-2

-1

0 (Z-Score)

+1

+2

30

40

50 (T-Score)

60

70

300

400

500 (SAT)

600

700

Scores of 70 represent Mild Mental Retardation
Scores of 55 represent Moderate Mental Retardation (below first percentile)

99.74% of scores fall between -3 and +3 standard deviations.

Remember: you can’t score in the 100th percentile. (99 is the highest because that is the percent of cases that fall below.)

Mean and standard deviation of the SAT is 500 (on each test) and 100.

Stanine (Standard Nine) – originates from military use.

Example computation of a standard deviation from raw scores:




X

X-Xbar

(X –Xbar)squared

85

31 (85-54=31)

961 (31 squared)

80

26

676

70

16

256

60

6

36

55

1

1

50

-4

16

45

-9

81

40

-14

196

30

-24

576

25

-29

841

540 (mean = 54)

 

3640 (Sum above)


Variance = {sum(X-Xbar)}squared/n = 364
SD = square root of the above = 19.08

We now know the mean and standard deviation of our set of scores. (54 and 19.08)

Normal Curve Equivalent (NCE) – has a mean of 50 and a standard deviation of 21.06. (not all that useful)

Age Equivalent and Grade Equivalent scores assume that all school districts curriculums are equal, and that they move kids through the grade levels equally. In actuality, these concepts are not universal.

Reliability – consistency, stability over testing occasions and time spans.

Three types
1. Test-retest reliability
2. Internal consistency – a measure of reliability based on the average inter-correlations of the items on the test (do the odds correlate with the evens?).
3. Alternate form reliability – two versions of the test correlate.

Filed under: EDC 512-513 Cognitive Assessment and Practicum
Copyright: September, 2003 - David Profitt