

What is SD?
SD = Standard Deviation

What is S_{EM}?
SEM = Standard Error of Measurement

Name the measures of Central Tendency.
Mean, Mode, Median

Name the measures of Variation.
Standard Deviation, Variance, Range

What is r?
r = correlation

How do I summarize data for a single variable?
Charts, Graphs, Central Tendency, Variability

What is a Histogram? What do the Vertical bars and Baseline mean?
A Histogram (History of Data) is a bar graph. Vertical lines depict the data frequency. Baselines depict observed scores on the dependent variable

What is a frequency polygon? What to the vertical lines and baseline depict?
A frequency polygon is a line graph. The vertical line depict score intervals or individual scores. The baseline (horizontal axis) depicts frequencies or percentages.

How does a positive scew differ from the normal curve?
There is a lot of negative or low scores with few people having very high scores. So, a curve with a positive scew has a longer tail on the right side (higher score side).

How does a negative scew differ from a normal curve?
Within the data there are a lot of high scores and fewer lower scores. So, the curve has a tail on the left side (lower score side).

What is Kurtosis?
Kurtosis describes the peakedness of a curve or the rate at which a curce rises.

Name both types of Kurtosis.
Platykurtic and Leptokurtic Curves

What is Platykurtic Curve?
Platykurtic curves have distributions that are flat and rise slowly. They are flat.

What is a Leptokurtic Curve?
A leptokurtic curve is a fastrising curve. This depicts tests that do not spread out or dicriminiate among the scores.

What is tendency?
Tendency = Data Trends

Name the two types of trends.
Central Tencency and Measures of Dispersion

What are the measures of central tendency?
Meadian, Mean, Mode

What are the measures of dispersion?
Range, Variance, and Standard Deviation

What is mode?
The score that most people got.

What is bimodal?
When a set of data has more than one mode it is termed, bimodal.

What is the median and how do you calculate it?
The median is the middle score. If there is an odd number of data, the median is the middle number. If there is an even number of data, you take the sum of the middle two numbers and divide that by two.

How do you calculate mean?
Add up all of the scores and divide by the number of scores.

Where do the mean median and mode fall in a Symmetrical Distribution?
All measures fall in the very middle of the curve.

Where do the mean, median, and mode fall in a positively skewed distribution?
The mode is at the peak of the curve. The the median and the mean follow to the right of the mode.
From left to right: mode median mean

Where do the mean, median, and mode fall in a negatively skewed distribution?
The mode is at the peak of the distribution and the median and mean follow to the left.
From left to right: mean median mode

What is a measure of variablility?
A measure of variability measures how spred out the scores are. It suggests dispersion. Measures of variability cannot be negative.

What do homogeneous scores mean?
There is little variability in the data.

What does heterogenous scores mean?
There is high variability in the data.

Name the measures of variabilty.
Range, Interquartile Range, Variance, and Standard Deviation

What is range?
The difference between the lowest and the highest data points.

What is interquartile range?
The spread among the middle 50% of the scores.

What is variance? How do you measure it?
Variance is how much the scores deviate from the mean. To find variance, you must first find deviation. Deviation is found by distracting the mean of a set of scores from each raw score. Variation is then the average of the squared deviation scores.
Variance is the sum of (the individual scores minus the mean) squared and divided by the number of data points.

What is standard deviation? How do you calculate it?
Standard deviation is another measure of dispersion, it is useful for describing scores (how much they deviate from the mean), puts scores back in original metric. SD = square root of variance

If scores approximate a normal distribution then how many scores in included in + or  1 Standard Deviation?
68%

If scores approximate a normal distribution then how many scores are there in + or  2 Standard Deviations?
About 95%

What is a normal curve? What can it do descriptively and inferentially?
A normal curve is a unimodal, symmetrical graphic depiction of scores.
Descriptively is can locate position of scores and important for normreferenced interpretation.
Inferentially it has reliability to derive confidence intervals.

How do you cummarize data on more than one variable?
With Bivariate relations, correlations

What are correlations?
Correlations measure the magnitude and direction of the relation between two variables. Pearson r

Name the scales of measurment.
Nominal, Ordinal, Interval, Ratio

What is the nominal scale of measurement?
Classify, no logical order

What is the ordinal scale of measurement?
Classify and rank or order

What is the interval scale of measurement?
Classify, orders, adds zero (arbitrary), and equal units

What is the ratio scale of measurement?
Classify, order, equal units, true zero point, has a true ratio

What are raw scores?
number correct, completion time, etc.
They lack comparability.

Name types of derived scores.
Age/Grade equivalents, Percentile Ranks, Standard Scores

What are some problems with age and grade scores?
They have differeny meanings (raw score of 5 can mean several different things), lots of misenterpretation, ordinal sclae (can't get standard error of measurement), imply constant rate of growth, imply a flase standard, different meanings across tests.

What are percentiles?
Not the percentage correct. As high or higher than xx% of children on this test (not as well or better),

Why are precentiles not useful in comparing scores?
Percentiles overestimeate differences in the center of a normal curve and underestimate differences on the tails of normal curve.

What are standard scores?
They make use of means and standard deviations. They are derived from zscores. Best scores for comparing.

What are zscores? How do you calculate zscores?
zscores are the number of standard deviations from the mean. Zscores are easy to convert.
Z scores = (rawmean) / Standard Deviation.

What is a normative sample?
Normative samples are a representative sample that match the population, is it comparable.

Name two types of measurement error.
Systematic and Random

What is systematic error?
Like a scale that always measures a person 10 pounds lighter

What is random error?
A person reading the weight of scale wrong.

What is a true score?
Hypothetical entities that would result from error free measurement.

What is reliability? How do you measure reliabilty?
Ratio of true score variance to total test score variance
Consistency, precision, trustworthiness
Reliability = true score variance/total variation in a set of scores
The more error, the less likely a peron's score is the result of true variation and less reliable
Extent to which differences in true scores are reflected in differences in observed scores

Name methods of estimating reliability.
Testretest, interrater, alternate form, internal consistency

What does testretest measure?
Consistency across repeated testing, stability coefficient.

What is interrater reliability?
Consistency across raters

What is alternate form reliability?
Consistency on parallel forms

What is internal consistency?
Consistency in measurement

Name two ways of calculating internal consistency.
Splithalf, coefficient alpha

What affects reliability?
Length (more items increase reliability estimates), Interval between testing (how stable is the trait), Constriction vs Extension of range, Anything that adds error reduces reliability

What are the desirable standards?
.60 for group uses (some say it should even be higher)
.90 for individual uses

What is the Standard Error of Measurment? How is SEM calculated?
Used to determine upper and lower limit of the range within the person's "true" score is likely to fall.
SEM represents the standard deviation of the hypothetical distribution of scores for person if they took the test an infinite number of times.
High reliability = small standard error of measurement or less error
SEM = Standard Deviation times the square root of 1 minus reliability
Reliability equals true variance divided by total variance

What percentage of likely scores is included in + or  1 SEM?
68%

What percentage of likely scores is included in + or  2 SEM?
95%

What are the most convenient ranges of confidence intervals?
68% 90% 95%

What is validity?
Does the test measure what it is supposed to measure?
Appropriateness, meaningfulness, usefulness of the inferences made from test scores or other measurements
A property of inferences, not the test.

Name types of validity.
Content validity, criterionrelated validity, construc validity

What is content validity?
Test items are representative of the range of possible items that should be covered

What is criterionrelated validity?
It is predictive. How well does the test predict future performance?

What is construct validity?
It is the essence of vailidity, factor analysis,
Concurrent: Should find a relationship if there is supposed to be one
Divergent: Lack of relation between constructs if there should be one

What affects validity?
Reliability (accuracy requires consistency) Systematic Bias (affects inferences)


V_{t}
True = reliability
Common + specific

Why do test scores differ?
Age of norms, floor and ceiling effects, regression to mean (correlations between tests), reliability and age, content difference, construct differences, practice effects, stability of trait

What is regression to the mean?
1correlations
Regression to mean effects
Information needed: score on one measure, population mean on both measure, correlation between the two measures

What are floor effects?
the lowest possible score on a test, indadequate floor can grossly inflate persons scores, a raw score of one should be at least 2 standard deviations below the mean

What it item gradient?
Steepness of how much a standard score changes with raw score change
Poor=large standard score change with small raw score change
Adequate gradient is sensitive to small differences

