1
Q
Data Types - Nominal Data
A
- Uses numbers as labels
- Denotes group membership
- Can be called Categorical Data
- Does not reflect magnitude or order in any way
E.g. gender: 1 = woman, 2 = man, 3 = non-binary, 4 = another gender
2
Q
Nominal date
A
- Common in psychological research
- Often used in experimental design
- Can be presented as frequency or percentage
3
Q
Data Types Ordinal Data
A
- Also called Rank Data
- Used to rank order, so has magnitude
- Does not tell us how different the rankings are
e.g. positions in a race 1st, 2nd, 3rd but not individual times - cannot use ordinal data to make inferences
4
Q
Data Types - Interval Data
A
- Also called Continuous Data
- Does not have a real Zero Point
- No way of assessing absence of a construct this way
e.g. Although Celcius temperature has a zero value it can continue beyond into negative figures. Temperature gets colder, it does not dissappear
5
Q
Zero Point
A
- Point on a scale that denotes zero
- From here, positive and negative readings can be made
- The point from which progress can be charted.
6
Q
Data Types - Ratio Data
A
- Also called continuous Data
- has a highest level of measurement
- Similar to interval data equal in scale
But in addition to order intervals might be different i.e. times in a race - Has equal magnitude and an Absolute Zero
e.g. weight data, bunch of apples equals a kilo, as we remove apples weight goes down but when last apple is removed we get zero
7
Q
Types of Data
A
- Nominal Data
- Ordinal Data
- Interval Data
- Ratio Data
Different statistical tests such as test of Correlation has miminum requirement for Interval Data.
Many psychological measures area hybrid of Interval and Ratio Data
8
Q
Correlation
A
- Correlation is key when looking at measurement
- Basis for reliability, validity, factor analysis
- Allows for assessment of individuals
- Individual variations are important factors of biological and social beings
9
Q
Correlational Psychologists
A
- their goal is to predict variation within a treatement
- demand uniform treatment for each case
10
Q
Explain Within Groups Variability
A
- Look for an association with the within groups variability of another variable
- Scatter Plot charts with x & y axis can measure two variables
e.g. marks on a test within the group of a classroom
11
Q
Quantify Relationships Between Variables
A
- Measures magnititude (strength) and direction (Positive or negative)
- Is it a strong or weak relationship
- Are there other variables at work
- The more the dots look like a line the stronger the relationship.
e.g. income vs Happiness
12
Q
Scatterplots
A
Linearity can described in two ways:
* Magnitude.
- i.e., strength.
* Direction.
- Positive or negative.
13
Q
Null Hypothesis Significance Test
A
- Also NHST
- Tests if sampling error is responsible for correlation
- Tied to sample size - the bigger sample the better likleyhood that difference is significant
- How likely is our result if the variables are unrelated in the real world
14
Q
Effect Size
A
- How practically meaningful a result is is
- regardless of its significance
15
Q
NHST - Formula
A
- H0: r = 0
- Variables are independent.
- r > 0 is due to sample error.
- H1: r ≠ 0 (two tailed); r > 1 or < 1 (one tailed).
- Variables are correlated
16
Q
Cohen’s Effect Size
A
Cohen’s (1988) Conventions for Psychological Research:
1. Large: > .50
2. Moderate: Approximately .30
3. Small: Approximately .10
Look at magnitude/size of the correlation
* Anything above 0.05 is a large correlation or effect
* Around 0.32 is moderate correlation
* between 0.12 and .3 is a small correlation
17
Q
Power Tables
A
- Important to consider when evaluating effects
- Need to assess if your measure has been properly tested
- This measures your effect based on sample size.
- The smaller the anticipated effect the larger the sample size you require
18
Q
Reliability Coefficient
A
- Must not apply Power Tables too rigidly
- r = .50 would be a poor measure
- Would not use the scale
- Correlation Coefficients need to be put into context
19
Q
Rosenthal 1995 - Correlation Coefficient
A
- Physicians identify most important medical breakthrough.
- Most popular was drugg cyclosporine- reduces organ donation rejection
- The relationship between patient survival and use of drug showed r = .15.
- This small correlation actually saved thousands of lives
- Needs to be interpreted in context
20
Q
Factors Affecting Magnitude of Correlation
A
- Reliability of Measures
- Outliers
- Restriction of Range
21
Q
Reliability of Measures
A
- The extent of a measure producing stable & consistent scores over time
- Not influenced by random errors
- Unreliable measures reduce the size of the correlation
22
Q
Non Linearity
A
where there is not a straight-line or direct relationship between variables
23
Q
Restriction of Range
A
- Variable has less variability in a sample than in the full population.
- Restricted range can be present for observed variables or in other variables that were not measured.
- Scores outside of a cenratin range can change the size of the effect
24
Q
Pearson’s R
A
- Most common statistic for correlation
- Indexes the degree of linearity only
- Not related to the slope of the line. if there is one.
- Measure of the linear relationship between two variables
- r = 0 Not necessarily absence of relationship
25
Pearson's R Magnitude
* Should not be confused with direction
* +.05 is the same magnitude as -.05
26
Bivariate Correlation
* Determine the existence of relationships between two different variables
* Correlation does not equal Causation
* Two correlated variables does not mean they cause one another
* Global Warming does not decrease the number of pirates
27
Causation
For causation we need:
* Covariation
* Temporal Ordering _ arrangement of events in time
* No Confounding - another variable that affects covariation
28
Covariation
Bivariate Correlation can only tell you about covariation
29
Reasons for a Correlational Relationship
* A direct causal relatioship exists
- Not possible to establish through correlation alone
* Non Causal Associations
- Reverse Causation
- Spuriousness - A covariation that appears to be causal but is not
- Measurement Association
* Complex causal associations
- Indirect Causation - Mediation
- Interactive Causation - Moderation
30
Normal Distribution
* The way scientists assume the majority of distributions look like
* Assuming normal distribution allows more complex analysis and inferences
* Mean, Median and Mode are all the same
* If actual scores are too different from normal then inferences become flawed
* Always check for normal distribution before inferring
31
Non Normal Distributions
* Postive and Negative Skew have longer tails to the left or right
* Mean mode & Median
32
Standardisation Groups
* When we have one score we compare it to the other scores in the group
* Single scores are not really useful
* This relies on large groups to strengthen normality
* Standardisation could be specific or general
33
Measuring Physical Characteristics
* So long as the measurements are accurate will give the same true measurement
* Like a ruler with mm cm metres
* Not arbitrary or guessing
34
Measuring Psychological Responses
* Cannot be directly observed
* Inferred from things we can observe
- Actions
- Behaviour
- Responses to self report scales
35
Test Scores
* Usually expressed by numbers
* Used to describe and draw conclusions
* Make inferences based on their outcome
36
Measurement
* Assigning numbers or symbols to characteristics
* Dictated by rules that represent magnitude
37
Measurement
* Assigning numbers or symbols to characteristics
* Dictated by rules that represent magnitude
38
Scale
* A set of numbers with the same properties
* Model empirical properties the objects that they are assigned
e.g.inches and centimeters
39
Sample Space
The values that a variable can take on.
e.g. Gender: male, female, nonbinary
or
e.g. Age in Years: 0, 1, 2 etc
40
Discrete Scale
* A sample that can be counted
* Numbers between spaces are not allowed
e.g. cannot have 3.4 vists to hospital
41
Categorical Scale
Divides sample into categories
e.g. Freshman, Sophomore, Junior, Senior
42
Quantitative Scale
Discrete scales because they can be countable
Can have numbers between spaces
e.g. 3.45 kgs of flour
43
Continuous Scale
* Values can be any real number in a scale.
* Can have fractions and decimals
* In theory can have irrational numbers
* Best to round these to certain decimal places to make them easier to work with
* Usually always contain errors
44
Error
* Refers to collective influence of factors beyond those specific to measure
* There can be many sources of error
* Usually connected to uncertainty of measuring rather than mistakes
e.g. decimals are thought of as Real Numbers but a score of 25 might actually be 24.98793 or 25.2134
45
NOIR
N - Nominal Scale
O - Ordinal Scale
I - Interval Scale
R - Rational Scale
46
Nominal Scales
* Simplest form of Measurement
* Involve Classification or Categorisation of distinguishing features
* Mutually exlusive categories
* Demongraphic factors can be categorised so are nominal
* Indicate quantities but can be Unique Identifiers
* Not always quantified
47
Nominal Scales
* Simplest form of Measurement
* Involve Classification or Categorisation of distinguishing features
* Mutually exlusive categories
* Demongraphic factors can be categorised so are nominal
* Indicate quantities but can be Unique Identifiers
* Not always quantified
48
Ordinal Scales
* Assign people to categories
* Have clear distinct order
e.g. how often do you show thrill seeking behaviour never, sometimes, often
or
e.g. I am a thrill seeker Strongly agree, agree, disagree, Strongly disagree
* rankings and percentiles are ordinal
* Alfred Binet developed Intellegence Tests said they were ordinal as they both measure and classify people by intelligence
49
Rokeach Value Survey
* An Ordinal Scale that lists personal values
e.g 1. Empathy - most important
2. Care - second most important
3. Respect - Third most important
* Values are put in order according to importance 1 - 10
* It is assumed there is *some* importance of each values
* Values have no Absolute Zero - this creates issues
e.g you cannot rank average 1. Empathy importance and 3. Respect Importance to acheive 2. Care importance
50
Interval Scales
* Have similar features to nominal and ordinal
* Have meaningful distances between each unit - A ruler
* Each unit is the same as the others - mm, cm, metre
51
Absolute Zero
* indicates absence of the unit being measured
e.g. 20 apples minus 20 apples = 0 apples
52
Ratio Scales
* Ratio Scales have true Absolute Zero
* Countable quantity that indicates absence of the unit
* represent the magnitude of the unit being measured
* Can be compared as proportions like 25 kilos is half of 50 kilos
* In psychology can be used in neurological functioning
53
Measurement Scales in Psychology
* Ordinal measurement is used most in psychology
* Tests that indicate the amount of intelligence, aptitude or traits etc
* Data is measured and converted to make them manageable or understandable
* Use graphs or tables to demonstrate ordinal and nominal level data
54
Frequency Distribution
* Scores are listed alongside the number of times each score occured
* Often refferd as *Simple Frequency Distribution* indicates data has not been grouped
* Is simple to read and usually represent actual test scores
* *Grouped Frequency Distribution* is used to summarise data and uses class intervals
* This summary may result in lack of detail
55
Ways to Display Frequency Data
* Graphs
* Histograms
* Bar Graphs
* Frequency Polygons
56
Measure of Central Tendency
* Indicates average or middle point between extreme
* Can be defined in different ways
* Usually defined by ** Arithmetic Mean** or the average
* Mean might not be used if there are extreme scores
* Could also use the Median and the Mode
57
What are Interviews in Psychology
* A tool in the biopsychosocial assessment kit
* Utilised in research & practice
* Ways of collecting information
58
How is data collected in psychology
* Interviews
* Questionnaires
* Survyers
* These are supplemental, often working hand in hand
59
Intake Interview
* Used for Clinical Purposes to find out where someone is at and building rapport
* Typically covers
* Demographic data
* Reason for referral
* Past medical history
* Present medical condition
* Familial medical history
* Past psychological history
* Past history with medical or psychological professionals
* Current psychological conditions
60
Mental Status Examination (MSE)
* Can be an Intake interview but can be separate
* Describes appearance, behaviour, consciousness and alertness, motor/speech, mood and affect, thoughts and attitudes and cognition
* Alertness, language, memory, constructional ability, and abstract reasoning are the most clinically relevant
* Also can be used to monitor progress over time
61
Different Types of Interviews
* Structure
* Unstructured
* Semi-Structured
62
Structured Interview
* Questions prepared in advance
* Mainly Closed Questions
* Read out exactly as written
* Schedule for Affective Disorders and Schizophrenia (SADS)
63
Schedule for Affective Disorders and Schizophrenia (SADS)
Detects symptoms of schizophrenia, major depression, bipolar and anxiety
64
Unstructured Interview
* No specific predetermined
* May have certain topics in mind
* Guided conversation or discovery interview
* Questions can change as we go
* Ask follow up questions
65
Semi Structured Interview
* Some structure and some flexibility
* Combination of open & closed questions
* Structured Clinincal Interview for DSM-5 (SCID)
