
Measurement:
The act of collection of information on which a decision is made

Evaluation
The use of measurement in making decisions

Law
Concise statement of fact that has been proven time and time again, generally accepted as true and universal

Theory
an explanation of a set of related observations that is based upon proof that has been verified

Hypothesis
attempt to explain some basic observations before precise data has been rigorously collected and analyzed

Quantitative
 deals with numbers
 can be measured
ex. length, speed

Qualitative
 descriptions
 can be observed
ex. yellow, soft

Statistics
a collection of methods for planning experiments, obtaining data, then organizing, summarizing, presenting, analyzing, interpreting, and drawing conclusions based on the data

Population
complete collection of all elements to be studied (scores, people)

Census
the collection of data from members of a population

Sample
a sub collection of elements drawn from a population

Statistic
a numerical measurement describing some characteristic of a sample

Parameter
a numerical measurement describing some characteristic of a population

Descriptive Statistic
summarize or describe characteristics of a known set of data

Inferential Statistics
use sample data to make inferences (or conclusions and predictions) about a sample
correlation or experimental designs

Important characteristics of data
Center: value that shows the middle of data set is
Variation: a measure of the amount that values vary among themselves
Distribution: nature or shape of distribution of data (bell shaped, uniform, or skewed)
Outliers: sample values that are far from the majority of other values
Time: changing characteristics of the data over time

Measure of Central Tendency
value at the center or middle of a data set
 median  use when there are extreme values
 mode  when data is categorical
 mean  every other time

Variability
how different scores are from the mean (spread, dispersion)
 range
 standard deviation
 variance

Range
 Max  Min
 used to get a general estimate of different scores are from either other

Exclusive range
Highest score  lowest

Inclusive range
Highest  lowest + 1

Standard deviation
measure of variation of values about the mean
s can increase dramatically with inclusion of outliers
units are the same as data
larger sd, greater the variance

Variance
the same thing as standard deviation except squared

Descriptive
 X is Y
 how things are
 most common type of study
 observe and measure specific characteristics without attempting to modify the subjects that are being studied

Correlational
 x is related to y
 how things are in relation to other things
 used most commonly in health science studies
 observations not manipulated but related to each other

Experimental
x causes y
 how things are and how they got that way
 hard to do well; apply treatments and observe effects
 used sometimes in evaluation but usually to explain descriptive evaluations

Methods of sampling
 Random  equal chance of being selected
 Systematic  every nth element in a population (ex. every third person)
 Convenience  data easy to get
 Stratified  but into subgroups then choose randomly from the group
 Cluster  divide population into clusters then choose random clusters and use all the population within the cluster

Experimental Designs
Cross sectional  all data observed, measured and collected at ONE point in time
Retrospective  data collected from the past
Prospective  data collected in the future from groups (cohorts) sharing common factors

Confounding
Occurs in an experiment when the experimenter is not able to distinguish between the effects of different factors
 Plan an experiment to avoid confounding
 Can avoid it by:
 Binding  participants dont know whether they are receiving treatment or placebo
 Matching  participants with similar characteristics
 Randomized Controlled Trial  randomly assign to each experimental group

Frequency Distribution
 lists data values (individually or groups of intervals)
 interval is called class or bin; helpful for large data sets

Skewness
Distribution extends to one side more then the other
 Skewed to the left (negatively)
 Skewed to the right (positively)

Histogram
 a type of graph that portrays the nature of a data distribution
 Normal distribution has a bell shape

Kurtosis
Has to do with how flat or peaked a distribution appears
 Platykurtic  more flat
 Leptokurtic  more peaked

Charts
 Column  to compare, bars horizontal
 Bar  same except vertical
 Line  to show trend
 Pie  to show proportions

Correlation
 relationship between two variables
 can be generated for predicting the value of one variable given the value of the other variable
 good for data that comes in pairs

Experimental research
 aims to find casual mechanisms and determine predictability
 always at least one independent variable and one dependent variable
 relationships can be bivariate or multivariate

Correlation vs. Experimental
 Correlation:
 investigates linear relationship between two variables
 continuous variables
 data can be graphically presented
 neither is truly the ind. or dep. variable
 called a bivariate relationship
 no causation

Correlation coefficient (r)
 a numerical measure of the strength of the relationship between two variables representing quantitative data
 r is in between 1 and 1
 value of r does not change even if units change
 measures strength of a linear relationship only

Homoscedasticity (homogeneity or variance)
 variance or errors are randomly and evenly distributed
 variance or errors on one variable are not correlated with variance or errors on another variable

Requirements for r
 Sample of pair x,y is a random sample of independent quantitative data
 approximate straight line pattern
 outliers need to be removed if their known to be errors

Common errors involving correlation
 Causation: wrong to conclude that correlation implies casualty
 Averages: averages suppress individual variation and may inflate the correlation coefficient
 Linearity: there may be some relationship between x and y even when there is no linear correlation

Measurement
consists of rules for assigning numbers to (objects) in such a way as to represent quantities of attributes
most measurement is indirect

Variables of interest
 what do you want to know and how an you know it
 empirical or operational definitions: what can be measured that bests reflects what we want to measuere

Classical test theory: O = T + E
 Observed score: actual score n a test
 True score: theoretical reflection of the actual amount of a trait or characteristic an individual possesses
 Error score: part of the score that is random

True Score
 The actual amount of the attribute you want to measure (ex. true dietary intake)
 Assumption: the construct is real and exists much like blood level or atomic weight if only we could measure it accurately

Errors
 Error  did not intend to measure that messed up the score
 Systematic error  repeatedly occurs and affects scores predictably
 Non systematic error  unpredictable and varies

Levels of Measurement
 Nominal  characteristic, names, least precise measure, mutually exclusive (cant be both)
 Ordinal  order, ranking
 Interval  where a test or assessment tool is based on something we can talk about how much higher performance is compared to a lower one
 Ratio  characterized by the presence of an absolute zero; absence of any of the trait that is being measured

Reliability
the degree to which scores are: free from errors of measurement; consistent, or stable across a variety of conditions
 types of reliability:
 Test retest reliability
 Interrater reliability
 Internal consistency reliability
 Parallel forms reliability

Testretest reliability
used when you want to examine whether a test is reliable over time (do it again in time by the same people) then find the correlation efficient when comparing scores aka correlation on a test given at two diff times
ex. same test is taken in july and january by the same people
 longer times require greater stability
 affected by change, carry over effects

Interrater Reliability
 measure that tells you how much two raters agree on their judgments of some outcome
 correlation of scores measured by two different observers or raters
number of agreements/ number of possible agreements

Internal consistency reliability
used when you want to know whether the items are consistent with one another in that they represent one dimension, construct, or area of interest
 ex. different test forms
 a function of the relationship between items on a scale and number of items

Parallel forms of reliability
 when you wan to examine the equivalence or similarity between two different forms of the same test (correlation of scores between two different versions of the test)
 ex. studying two different things same method
then find correlation coefficient

