Ways to Gain Information
Empiricism, Faith/Intuition, Tenacity, Authority, Observe, Tradition, Rational/Logical
Empiricism
Research Based
Faith/Intuition
Gut, No Source, Can not separate accurate v. inaccurate
Tenacity
Superstitions, "Lucky Charm
Authority
Experts in the field, Bias, Wrong, Subjective
Tradition
Info accepted as true/ Always have; "Can't teach dogs new tricks
Rational/Logical
Assumptions/Wrong
Observe
See it, but cannot observe all situations
Inductive
Start specific move toward generalizabilty
Deductive
General topic narrowed down to specific topics
Research Hypothesis
Have a topic not yet confirmed; educated guess for the outcome; Design an experiment to test the assumption (Must be Testable)
Null Hypothesis
Ho (No Change)
Alternative Hypothesis
Ha (Modify IV (Intervention))
Independent Variable
IV; The variable being tested/observed
Dependent Variable
DV; Outcome
Operational Definition
Specific & Observable Definitions of Variables
Positive Correlation
1 thing goes up; other thing goes up; (Same Direction); EX: Pressure on Gas, Speed Increases
Negative Correlation
One thing goes up, other thing goes down (Opposite directions); Pressure on breaks; Car slows down
Correlation
Cause & Effect Relationship
Types of Correlation
Temporal Order, Correlation/Co-variation, No Alternative Explanations
Temporal Order
1st thing comes before something less
Correlation/Co-variation
1 thing to move in relation to another
No Alternative Explanations
(! Ha); ex: Ice cream sales, heat & crime)
Validity
Internal & External
Internal Validity
Internal Relationship
External Validity
Generalizability
Prevent Internal Validity Threats
Random Assignment
Prevent External Validity Threats
Replication
Construct
Umbrella that holds everything about/goes into the experiment
Types of Constructs
Face Validity, Concurrent Validity, Convergent Validity
Face Validity
If experiment looks like it's measures what it is supposed to measure (right); checked with experts in field
Concurrent Validity
Comparing results to other similar research; Measuring something in generally the same way
Convergent Validity
Same stimulus/Using same set of data; Come to same conclusion
Reliability Measures
Regression to the Mean, Interrater, Intra-rater
Regression to the Mean
Math test today vs. Math test Tomorrow (Test; Retest)
Interrater Reliability
Different people rating the same thing
Intra-rater Reliability
Same person rating the same thing
Range
Upper limit of highest number to Lower limit of lowest number; Highest number to Lowest number + Class Interval; Interval
Scientific Method
(1) Observe,
(2) Hypothesize,
(3) Use hypothesis to Generate Testable Prediction,
(4) Evaluate Prediction by making Systematic, Planned Observations,
(5) Use Observations to support, refute, or refine hypothesis.
Bias
Error, Deviate, Residuals (Average = Expected Value)
How to reduce Bias
Bigger sample; Way present Information; Standardize
Demand Characteristics
How people behave in Study; Help or harm you/the experiment
Hawthorne Effect
Change behavior when know you're being watched
Good Participant
Know what you want to heard & give it to you
Bad Participant
Make up data; Know what you want; gives you bad data
Institutional Review Board (IRB)
Make sure experiments do not harm any human beings (Informed Consent, Confidentiality, Deception)
Probability Sampling
Random, Stratified Random, Cluster, Proportionate
Simple Random
Equal chance to be selected; Choice of 1 doesn't affect others
Stratified Random
Define groups (left out); Try to get equal numbers for the groups; can oversample groups; EX: Age groups (18-25, 25-30)
Cluster
Large sample; Not representative; Random unbias size unknown; main equal chance to be selected; EX: neighborhood
Proportionate
Divide population calculate probability of all groups; Reflect correct proportions of groups (Underrepresented)
Non-Probability Sampling
Convenient, Quota, Purpose, Snowball
Convenient
Voluntary (selected, available, have to (for a grade); Bias
Quota
Identify subgroups, but take all available people; Fill the spots/#'s; 50 men & women (or whoever is available)
Purpose
Select people - certain characteristics; Extreme groups; (EX: Top 5% vs Bottom 5%)
Snowball
Not random; Only way to get unlikely participants; Hard to generalize; Population that would not volunteer [EX: Addicts, Prostitutes]
Fraud
Fudging your information; Filling in any answer; Explicit effort to falsify/ manipulate data
True Experiments
...
Non-Experiment
No control group; pre-experiment
Correlational Study
2 independent studies to see if they correlate; Ex: anxiety & depression (Not Causation)
Quasi-Experiment
Missing Randomization; Analyzing; No IV/DV
Greek
Population (Parameters); Constant; (u, o)
Latin
Statistic; Variable; (x, s)
Population
N; All possible options for people (Whole population)
Sample
n; People you collect for your research
Occam's Razor
Simplest Solution tends to be the best one"; Ex of Parsimony
Scales of Measurement
Nominal, Ordinal, Interval, Ratio
Nominal
Categories with no relation to others; (EX: Hair color; Sports jerseys)
Ordinal
Categories grouped; Order Matters; Groups do not have to be equal; (Silver, Bronze)
Interval
Categories are increments are the same; Zero: no meaning; absence; (Temperature)
Ratio
Categories with intervals and an absolute zero; Absent zero; (Distance Traveled)
Kurtosis
Measures thickness of the tails*; not center mass; (Variance)
Leptokurtic
Tales very narrow; Scores closer to mean
Mesokurtic
(Normal) Baseline to compare other examples
Platykurtic
Incredible difference; (EX: Height of circus performer)
Positive Skew
Tail in + direction away from 0 (Zero)
Negative Skew
Tail in - Direction toward 0 (Zero)
Normal Standard Deviation
Mean, Median. Mode = 0 (Perfectly Standardized); Measure in Z Scores; Percentile - Closer to 0; Closer to percentile points
The Standard Deviation
Principles of Science
Empirical, Public, Objective
Empirical
Structured; Observations; Support/ Reject Hypothesis
Public
Available for others to Replicate* it
Objective
Not Skewed/Bias
Pseudoscience
Hypothesis not refutable/ testable
Types of Psychological Research
Qualitative & Quantitative
Qualitative
Numbers, Scores, Statistical Analysis; Interpret Results
Quantitative
Observation, Understanding something (ex. Behavior)
How to Find Research Topics
(1) Interests
(2) Observations/Practical Issues
(3) Theories
(4) Previous Research
2 Types of Variables:
(1) Well-Defined
(2) Constructs
Well-Defined
Easily measurable (Height, Weight)
Constructs
Intangible/Abstract; Hypothetical entities that predicts behavior
Validity.
Measure what it is supposed to measure
Reliability
Able to be replicated; Generalized; Consistency
No Correlation
Constructs
Scores from measurements behave same as construct measured
Types of Constructs
Convergent & Divergent
Convergent
Measure correlates with similar constructs; Different scales/tests that measure the same answer
Divergent
Does not correlate with different constructs; EX: Happiness test measures with Depression
Factor Analysis
Structure of Scale; Support that sub-domains exist; Verbal Comprehensions (Dimension Scales; Sub-scores)
Content
Items on measure/Test; Items represent Sample/ Content/Construct (Educational)
Criterion
Measure persons standing; (Psych Boards; Intervention)
Types of Criterion Validity
Predictive; Concurrent; Incremental
Predictive
The test & future criterion; GPA+SAT=College Exam
Concurrent
Same point in time
Incremental
Increased value when adding multiple factors
True Score Theory
Every score has true level of knowledge & some error; (X = T + E)
Observed Score = True Score + Error
Test & Retest
Success of Measurements; Problem: Learned from first test (Parallel tests, Change wording)
Interrater
Simultaneous evaluation; Agreement btw 2 or more people who rate/grade sessions
Internal Consistency
Split & Cronbach's Alpha
Split Half
Split down the middle; Compare one half to the other; EX: 100 Question test - Compare 50? To other 50?
Cronbach's Alpha
Average of all split half correlations; Remove bias; (EX: .9 or higher = Good); High stakes tests
Modalities of Measurement
Self-Report; Physiological; Behavioral
Self-Report
Direct Questions (Pain level)
Physiological
Objective, Expensive; May not provide valid measure of construct; (EX: Blood pressure, heart rate, MRI)
Behavioral
State vs. Trait; Behaviors are NOT stable; (Scales, Words Recalled, IQ) (EX: IQ vs. Hunger)
4 Ethical Issues:
(1) Do NOT Harm
(2) Informed Consent
(3) Confidentiality
(4) Deception
Do NOT Harm
Ex: Prison Studies (Guards vs. Prisoners); Milgrim Study (Electric Shock)
Informed Consent
Know all information about a study to make rational decisions;
(1) To be apart of the study or not
(2) To cancel or quit at any time
(3) Understand what is expected of them in the study
Confidentiality
Everything Protected & Safe & Secure; All Demographics, Responses and other information are secure
Deception
Could Modify Behaviors of Participants if they knew true reason for the study;Behavior Modification
Debriefing
Informing the participants of the true purpose of the study AFTER the study is over.
2 Types of Deception
Passive & Active
Passive
Withhold Information (Test on Memory)
Active
False or Misleading Information (Confederate)
Confederate
Use of actors in studies
(Ex: Ashe Study: Match the lines; All actors with 1 participant)
Plagarism
Using someone else's work & publish as their own; Unethical representation of someone else's work as their own
Types of Bias
(1) Parsomony
(2) Plausible Rival Hythothesis
(3) Findings VS. Conclusions
Parsomony
Occum's Razor to Research
Plausible Rival Hythothesis
Results of any simple experiment may be result of influences other than IV; *Eliminate All Other Factors; Demonstrate effect multiple times to be true; (Ex: Significant findings could be due to chance or bias)
Demonstrate effect multiple times to be true
Findings
Descriptive feature of study; (What you got from the study)
Conclusions
Hypothesis (Expansions of Content)
Single Blind Study
Participant unaware of Study
Double Blind Study
Participant & Experimenter Unaware of Study
4 Elements of Experiments
(1) Manipulation
(2) Measurement
(3) Comparison
(4)
Manipulation
Researcher manipulates IV to change the value; Levels of the Independent Variable
Measurement
(Dependent) measure effects of change; Cause
Comparison
Scores/Treatment condition compared to another;
(a) Experiment:
(b) Control:
Experiment
Receive treatment or different level of treatment
Control
Do not receive treatment/ Placebo
Control
IV & DV; Control all other Variable Separate from IV & DV; Extraneous/ Confounding Variables (Do not affect IV & DV)
Control for Confounding Variables
(1) Match Design
(2) Random Assignment
(3) Confound
Match Design
(Stratified Sample) Match so both groups are equal
Random Assignment
Randomly assigned to 1 group/ other (Divide group evenly)
Confound
2 Variables are related to another unknown variable (Random, Manipulation, & Comparison )
Pre-test
Determine pre-existing knowledge before the study; When cannot control for experiment (Quasi-experiment)
Control Group
Something to compare the experimental group against
Random Sample
How you drew the people/population to study
Stem & Leaf Plot
Group Scores; Know the Information; Weakness: Do not know exact #'s (Lost Info)
Interpolation
Group Frequency Table; Weakness: Lose Specificity
Histogram
Interval/Ratio Data (Continuous); Real Limits
Bar Graph
Ordinal; Nominal
Probability Density Function
Normal Distribution; At mouth of the curve of Histogram -- With More Score Points; (For All Score Points)
Mean
Add all scores (divide) by Sample Size; Ordinal; Ratio/Interval (No Nominal); Most representation not always suitable
Median
Middle; 50% (Percentile)
Mode
Most Frequent Numbers
Symmetrical Distribution
Mean, Median are the same; 2 Different Modes
Uniform Distribution
Mean, Median are the Same; No Mode; No Peak; Random Sampling
Intercortical Range
Between 75% & 25% (percentile); Closer to Middle (Middle 50% - percentile)
Semi-Intercortical Range
Take intercortical (75% -25%) & Divided by 2
Variance
Average (squared)2 difference from mean
Standard Deviation
Average Distance from Mean; Most scores within Range (Mean)
Degrees of Freedom
# of Scores in observation; Not all options are free to Vary (cannot change)
Z Scores
Relative standing in the group; Score in Standard deviation units; Positive above the mean, Negative below the mean
Standardizing of Z Scores
Negative Numbers (Half positive, Half negative); Difficult to interpret negative score (May not be correct interpretation)
Unit Normal Table
3 ways to look up probability; Proportion in body, tail; Z Score greater than 1
Distribution of Sample Means
Collection of Sample Means of ALL possible random samples for a particular size; Normally distributed - Because Normal (same method for individuals)
Standard Error of the Mean
R O X O
R O O
Difference is the X; Treatment Made a Difference
History
Specific (global) events occurring btw 1st & 2nd measurement
Maturation
Respondents are function of passage of time
(a) Change over time
(b) Growing older, hungrier, more tired, etc.
Testing
Effects of taking a test upon the scores of the 2nd test; Only introducing testing threat (EX: Pre-test vs. Post-test)
Instrumentation
Changes in measurements due to changes in calibration of measuring instrument or in observers/scorers (Different rater & rater style; Change measures from Pre-test to Post-test)
Statistical Regression
Groups selected on basis of extreme scores; Extremes natural rush toward means (Regression toward the mean)
Selection
Biases of respondents for comparison groups; 2 Groups - Not equivalent before study begins
Experimental Mortality
Differential loss of respondents from comparison group; Die or Drop out of the study/ experiment
Selection-Maturation Interaction
Multiple-group quasi-experimental designs is confounded w/mistaken for the effect of experimental variable; Not equivalent before study; 1 group matures quicker than the other
Reactive/Interaction Effect of Testing
Pretest might increase or decrease the respondent's sensitivity or responsiveness to experimental variable; Know what study is about/ other group present; Cannot generalize to the population
Interaction Effect of Selection /(Treatment) Experimental Variable
Groups not equivalent prior to study; Respond to treatment differently; Hard to recruit sample
Reactive Effects of Experimental Arrangements
Conditions of study different from real world; People know in study and React differently (Being Watched)
Multiple Treatments Interference
Multiple treatments are applied to same respondent; Prior treatments are not erasable; Repeated measures design (ABAB); Because repeated (measure if works) OR Different treatment unsure which one worked
Temporal Sequence
Intervention first then measurement (Cause than Effect)
Reactivity
Modify natural behaviors to natural settings
Good Subject Role
Aware of purpose of study & try to support hypothesis (Ex: Respond in unnatural ways)
Apprehensive Subject Role
Overly concerned about being evaluated; Worried about answers & how they respond; Dishonest because uncomfortable
Negative Subject Role
Not Nice; Know study and want to be oppositional/ disprove the study or hypothesis
Faithful Subject Role
Know hypothesis, but avoid acting on it; Do not want to screw up the study; Honest, apathetic & Not affected
Inferential
Generalizing from sample to population; How general public will react based on sample
Effect Size
Size of the difference rather than confounding with sample size
Hypothesis Testing
4 Steps:
(1) State Hypothesis
(2) Set alpha level & criteria for decision
(3) Parametric Test (Assumptions)
(4) Make Decision
State Hypothesis
Ho (Null Hypothesis)
H1: Alternate Hypthesis
Null Hypothesis
(Ho) No Effect = IV had no affect over DV
Alternate Hypothese
(H1) IV does have affect over DV
Set alpha level & criteria for decision
Alpha Level
Alpha
What would warrant rejection of Ho; Define probability value for what is unlikely
Alpha Level
To confidence or error probability; Not always failure of treatment (by chance)
1 Tail Test
ex: 5% chance on 1 end (all on 1 side)
2 Tail Test
split on either side of region; examines positive and negative effects (Critical region in tails)
Parametric Test
Assumptions:
(1) Normality
(2) Homogeneity
(3)Interval/Ratio
(4) IV Observations
(5) Random Sample
Normality
Sample or Hypothesis are normally distributed
Homogeneity of Variance
Multiple groups; Pulled Variance (Same Population)
Independent Observations
Each person is responding independently; Other participants do not influence another
Random Sampling
Infer from sample to population (Must be Representative)
Make Decision
(1) Reject Null
(2) Fail to Reject the Null
Reject the Null
Support the Hypothesis; If sample statistics falls in critical region; Parsimony
Fail to Reject the Null
May be left out some portion; Always Alternate Evidence
Parsimony
Without further evidence, No effect
Decision Accuracy Table
Z-Test
Standard area with fixed (same area) = fixed parameter; (Find Z - See if it falls in/out of the tails)
Critical Region
Single T-Test
Approximate population variance with sample variance
T Statistic
t(df) = (t-value), p<(p-value)
Between Samples
(Independent Variable) Same population; Control Groups [Ex: Male vs. Female; Treatment vs. Control]
Within Samples
(Dependent Variables) Same participants; Change for sample group; [Measure Anxiety , then Hypnosis]; Measure Impact of Treatment
Matched Subject Design
(Equivalent Groups) Same Qualities; [1 treatment, 1 control group; People that compare in characteristics]
Independent T-Test
Ho = u1 - u2 = 0;
(1) Find df
(2) Find Sp
(3) Find t Value
(4) Define Critical Region
(5) Reject/Fail to Reject
Dependent T-Test
Evaluate 2 set of scores (Impact of treatment);
(1) Ho = u = 0
(2) Find df
(3) Find Sx
(4) Find T Value
(5) Define Critical Region
(6) Reject/Fail to Reject
ANOVA (1 Way)
Evaluate mean difference between 2 populations; Purpose: Split different types of variance
1-Way Anova Steps
Ho= u1 = u2 = ... = u15
(1) Level of Significance & Critical Level
(2) Complete Source Table
(3) Reject/Fail Null
[F(df1,df2) = F, p (>,<) .05/.01, R2 = (value)
Anova Source Table
Correlation Statistics
Relationship between 2 variables; No Causal; [Greek - p; Latin - r]; (For validity & reliability)
3 Characteristics of Correlation
(1) Form (Linear)
(2) Direction of Relationship (Positive/Negative)
(3) Degree of Relationship (Strength)
Correlation - Degree of Relationship
Closer to 0 (bad; no relationships); Closer to 1 (stronger relationship)
Pearson Correlation
Linear relationships; Interval/Ratio; Quantifies relationship between 2 variables in a scatterplot
Regression
Finding best-fitting straight line for set of data ("Fit Line"); Predict "y" when only know "x" value; Higher the correlation, higher the prediction
Residual
Left over" [Same as the deviate from the mean]
Restriction of Range
Do not know the full extent of the relationship; Make predictions, Do not have all the information;
(Limits the correlation; Do not go the full range)
Outlier
Extreme score compared to average data; Remove it - Mess up data