Online Module Flashcards
Construct 3
- Can’t be measured directly
- Underlies observable behaviour
- WHY? Attribute that can’t be touched/felt
EXAMPLE = happiness, intelligence
Operational definition (2 parts):
1) The PROCESS Of MEASURING an unobservable construct
3) INDIRECTLY
Eg. happiness + no. of smiles in 5 minutes // Intelligence + IQ test
Operationalisation
process of FINDING the way to measure a construct
Measure or operational definition? - thought process 3
1) What do we WANT to measure?
2) What are we ACTUALLY measuring?
3) No match = OD
Match = measure
EX 1 Total score on a questionnaire measuring symptoms of anxiety
EX2 Weighing scales to measure weight
Measure or operational definition?
EX1 = Operational definition
- What do we WANT to measure? Anxiety
- What are we ACTUALLY measuring? Total score
- No match, therefore OD
EX 2= measure
- What do we WANT to measure? Weight
- What are we ACTUALLY measuring? Weight
- Match, therefore measure!!
Psychology - what is it based on?
- Psychology is EVIDENCE BASED
- CLAIMS are always made (eg. yoga helps with happiness) - but we need evidence!
Research process 8
1) Research Question + Hypothesis
2) Variables and Measures:
3) Study Design
4) Recruitment
- Participants are recruited.
5) Data Collection
6) Analysis
7) Conclusions
8) Critical Consideration
- The results are considered, and the next step in the research process is identified. The process then repeats when a new study is undertaken.
Research question - definition
DEFINITION
- broad ideas that ask about either association, difference or causation
UNDERSTANDING
Hypothesis - definition, understanding
DEFINITION
- logical, specific, testable, refutable and predictive statements about what will happen in a psychological research study
- (BASED ON THE RESEARCH QUESTION)
Research q - what is it, how is it formed? 4 , example
HOW MAY IT BE FORMED?
1) Personal interests
2) Observations
3) Practical problems (eg. unsolved health issues in society)
4) Theories
WHAT IS IT?
- OVERARCHING QUESTION
EXAMPLE?
- If happiness is the area of interest, potential research questions could include:
-> Is happiness related to sunshine?
-> Are art students happier than science students?
-> Does happiness influence academic performance?
3 types of research questions + examples
TAKE HAPPINESS AS THE AREA OF INTEREST
1) ASSOCIATION : q explores relationship between diff variables
Eg. Is there a relationship between happiness and sunshine?
2) DIFFERENCE: q thinks about differences between groups or conditions - related to interest
Eg. Are art students happier than science students?
Eg. are children happier in public or private schools?
3) PREDICTION: q aims to predict outcomes based on certain variables.
Eg. Can we predict academic performance based on students’ levels of happiness?
After formulating the research question, what should be done before proceeding to the next step of the research process?
LITERATURE SEARCH
Eg. peer-reviewed articles, books, reports, and other credible sources.
- ensures you don’t duplicate research, logical next steps - contribute smth meaningful to the field!
Hypothesis - explanation
WHAT IS IT?
- its a PREDICTION related to your research questions, but has to fulfil the LSTRP criteria
LOGICAL
- Should follow logically from the literature review.
SPECIFIC
- Narrowly focused compared to a research question.
TESTABLE
- Must be able to be measured and observed
REFUTABLE
- Should be phrased so that it could be supported or refuted
PREDICTIVE
- Should predict that something WILL happen (ie. not that nothing will happen)
EXAMPLE
- “In a group of university students, higher levels of reported happiness will be related to higher average daily duration of exposure to sunlight”
(both ** can be measured + observed!)
Difference between research process + scientific method
SCIENTIFIC METHOD =
- generalised
- applicable to all scientific research
RESEARCH PROCESS =
- more detailed + specific psychological research in clinical setting
Hypothesis vs prediction
Hypothesis: “Clinical treatment X reduces depression levels.”
Prediction: “Patients receiving treatment X will show a 30% decrease in depression scores on the Beck Depression Inventory after 8 weeks of treatment.”
- HYPOTHESIS = PREDICTION - PREDICTION IS ACTUALLY SECURING THE SPECIFICS, APPLY TO REAL LIFE SITUATION
Scientific Method - GENERAL STEPS 5
1) Observation → relationship = induction
2) Associated variables → Hypothesis
3) Hypothesis → prediction = deduction
4) Collect data (research)
5) Support, refute, refine hypothesis
Scientific Method - STEP 1
1) Observation → relationship = induction
- Observation = ppl swear in pain
- Induction = using a SMALL SET OF OBSERVATIONS to form a general statement about a LARGER SET of possible observations
- Relatonship = take 2 people to generalise relationship between swearing + pain
Scientific Method - STEP 2
2) Associated variables → Hypothesis
- Associated variables = conditions that change/have diff values for diff individuals
EG. swearing, type of pain, social setting, age, gender, height, economy - Hypothesis = use associated variables to HELP form a hypothesis if needed
-> swearing reduces pain (simple)
-> swearing more common in private settings
Scientific Method - STEP 3
3) Hypothesis → prediction = deduction
- Hypothesis = said above
- Deduction = GENERAL STATEMENT to reach conclusion about SPECIFIC EXAMPLES
- Prediction = applying hypotheses to a REAL LIFE SITUATION, smth OBSERVABLE
- eg. more swearing = high tolerance to pain
3 principles of the scientific method
1) STRUCTURED + SYSTEMATIC OBSERVATION
- Ie. the observations are structured in a way that the results either support or do not support the hypothesis (only one way or the other)
2) SCIENCE IS PUBLIC
- Has to be able to be replicated
3) SCIENCE IS OBJECTIVE
- Observations are structured so bias and belief of researchers do not influence the study
Variable definition
- characteristic or condition that can vary amongst people
Continuous variable - definition
- allow fractional values to be assigned when they are measured
eg. height, weight, CAN HAVE A DECIMAL
Discrete variable - definition
- only allow whole values to be assigned when measured
eg. number of people in a family, can’t have a half number of people!
Variable - intangible / concrete
- Intangible = (e.g., computer literacy)
= Concrete = height, weight, or age
Independent + dependent variable - causal relationship
Independent Variable:
- thought to be CAUSAL
Dependent Variable:
- hypothesized to be AFFECTED by the independent variable.
Example: In the hypothesis “sunlight affects mood,”
- sunlight = independent variable
- happiness = dependent variable.
What are scales of measurement?
- I have variables in a study
(eg. temperature, type of environment, family size) - For easy data analysis, I want to ASSIGN NUMBERS to my variables
- then i will PUT THEM ON A SCALE
- the different types of scales can TELL US different things
4 scales of measurement
1) Nominal
2) Ordinal
3) Interval
4) Ratio
Nominal scale of measurement
- CATEGORIES are assigned numerical codes
- SCALE ITSELF that doesn’t really mean anything!!
- eg.
HR = 1
Marketing = 2
Finance = 3
OR
Female = 1
Male = 2
Other = 3
Ordinal scale of measurement
WHAT DOES IT TELL US?
- Numbers are put onto a scale, and this indicates ORDER
- BUT the scale does not reveal the magnitude of difference between points.
EG. Finishing positions in a race (1st, 2nd, 3rd) - doesn’t tell you the time between the runners, BUT shows order
OR
Time spent in the sun:
A little = 1
Average = 2
A lot = 3
- we can’t quantify the difference between scale 1 + 2, BUT in order
Interval scale of measurement
WHAT DOES THE SCALE TELL US?
- EQUAL units of measurement SEPARATING each score on the scale
- BUT there is no real zero point (zero does not mean the complete absence of something).
Interval scale of measurement. EXAMPLES
1) scales could just be LITERAL NUMBERS
- but must be EQUAL in distance + no 0 point
eg. Temperature in Celsius; 10, 15,20,25 AND 0°C doesn’t mean no heat.
Ratio scale of measurement
WHAT DOES THE SCALE TELL US?
- EQUAL units of measurement SEPARATING each score on the scale
- BUT there IS a real zero point (zero does mean the complete absence of something).
Ratio scale of measurement - 2 examples
EXAMPLE 1 - LITERAL LIST OF NUMBERS
- Distance in centimeters 0cm, 5cm, 10cm, 15cm = equal intervals!
- 0cm means the objects are touching - ABSENCE of distance
- Age: (1 = “0–17”, 2 = “18–29”, 3 = 30-54”)
What scale of measurement?
ORDINAL SCALE
- Here intervals between the scales are NOT equal (we can get random age from 1 and 2, and this won’t be the same for 3 and 4)
- BUT order is correct
What is research design?
How we FORMAT + STRUCTURE a research project
- In terms of:
- What participants are asked to do
- The way participants are grouped
- The way data is collected
- How the project progresses over time
What might affect research design? 5
- Research question
- Hypothesis
- Variables used
- Measures used
- Plans for data analysis
Types of research design 5
- Descriptive
- Correlational
- Non-experimental
- Experimental
- Quasi-experimental
How to group the research designs? 3
1) OBSERVATION ONLY
- Descriptive
2) CORRELATION (NOT cause and effect)
- Correlational
- Non-experimental
3) CAUSE AND EFFECT
- Experimental
- Quasi-experimental
Experimental manipulation - definition, when is it present
- Introducing a control treatment
- Only present for EXPERIMENTAL designs (cause + effect)
- NOT present in both correlation designs
Naturalistic observation
- NO experimental manipulation done
Descriptive research design -what is it, example
WHAT IS IT?
- Just observations
EXAMPLE?
- Investigate typical height of people in class using measuring tape
Correlational vs Non-experimental research design - similarities + differences
Correlational
- SIMILARITIES = Finding correlations, NOT cause + effect
- DIFFERENCES = One group, 2 variables
* Eg. how does exercise impact happiness?
* Per person, 2 data has to be collected → exercise in a week + happiness quiz
Non-experimental
- SIMILARITIES = Finding correlations, NOT cause + effect
- DIFFERENCES = Two groups, 1 variable
* Eg. are science students smarter than art students?
* For each group, 1 piece of data = IQ test
Correlational vs Non-experimental - why can’t they show cause and effect?
- Eg. How does exercise impact happiness?
-> This doesn’t consider other factors like sleep, friendship, alcohol in the experiment
-> Hence furthest we can go is talk about correlation
Experimental vs Quasi-experimental - similarities + differences
Experimental
- SIMILARITIES = Finding cause and effect
- DIFFERENCE = 2 CORE FEATURES
-> 1) Manipulation of independent variable → to create a control treatment → eg. group with depression, group without
-> 2 ) Controlled experimental conditions → random assignment of participants to different groups
Quasi-experimental
- SIMILARITIES = Finding cause and effect
- DIFFERENCE = less controlled
-> Same manipulation of control treatment, but no random assignment!!
Experimental research design - example
- Most effective to see if medicine works
- One group = treatment given
- Second group = treatment not given
- This acts as a control treatment - able to really see if the medicine works or not
OR…
- Want to see if depression (IV) has a direct link to happiness
- One group = depression
- Other group = no depression
- Check happiness level using survey
- Random assignment further eliminates control for BIAS / might impact results!
Quasi experimental research design - example
- Same as experimental
- BUT participants may have been assigned to groups on the basis of a characteristic difference/shared quality
- This is NOT assigning them to groups randomly
- SO…may have other factors associated with independent variables NOT CONTROLLED!
Internal validity - definition
- Degree of confidence we have in the direct relationship between the independent variable in a research study
- (IE. causation and not just correlation)
Internal validity - how is this done?
- Through RANDOM ASSIGNMENT of participants to different groups
- Ie. don’t share all of the same characteristics
Approaching exam qs - what research design is being used?
- Are we observing, testing for correlation, or cause + effect?
- If correlation = is it 1 group + 2 variables (classic correlation) OR 2 groups + 1 variable
- If cause + effect = is there random assignment of groups or not?
Approaching exam qs - naturalistic observation in which research design?
- Correlation = no control treatments = YES, NATURALISTIC OBSERVATION
- Causation = yes, control treatments = NO, NOT NATURALISTIC OBSERVATION → EXPERIMENTAL MANIPULATION
Population vs sample
POPULATION
- Everyone of relevance to a research question
BUT…
- usually not possible to investigate an entire population so…
SAMPLE
- A sample is a group of people drawn from the population to participate in a research project.
Population - example
- If the research investigates the prevalence of anxiety in emergency workers, the population would be all emergency workers.
- If the research investigates cognitive features of schizophrenia, the population would be everyone who meets the criteria for a diagnosis of schizophrenia.
SAMPLING - 2 TYPES
- PROBABILITY SAMPLING
- the probability of selecting an individual CAN be determined.
- Population characteristics are known - NON-PROBABILITY SAMPLING
- the probability of selecting a given individual CANNOT be known.
- Population characteristics are NOT known
Probability sampling - how can we “determine” the probability of choosing someone?
- List Everyone: Write all 500 names on slips of paper (or assign numbers 1–500).
- Random Selection: Put the slips in a hat, mix, and draw 10 OR Use a computer to randomly pick 10 numbers between 1 and 500.
- We know the probability of choosing everyone is EQUAL
- BUT we can’t do this all the time (eg. can’t have people at a school write their names)
Probability sampling - PROS + CONS
PROS
- No selection bias (probability of choosing everyone in the population is EQUAL)
CONS
- Doesn’t guarantee the sample will be representative
- Eg. all boys in sample, no girls
Main example of non-probability sampling
CONVENIENCE SAMPLING
- The sample consists of individuals who are easy or convenient to recruit.
Convenience sampling - examples
- Example: First-year research experience programs
- Recruiting participants through hospitals for cancer research questions
Convenience sampling - pros + cons
PROS
- Easy, efficient
CONS
- Very prone to SELECTION BIAS
- Eg. You might accidentally recruit mostly 18-year-olds (easier to find on campus), ignoring older students.
How do we control sample bias?
USE QUOTAS
- Definition: Pre-set limits for how many people you recruit from each subgroup (e.g., age, gender).
- Goal: Mimic the natural distribution of groups in the population.
NOTE - there are many ways to do this though, quotas are not mandatory, but recommended
Quotas in research design - examples
Imagine your POPULATION of interest has this age split:
- 70% under 25
- 20% aged 26–50
- 10% over 50
Quota Sampling:
- In your sample you will set target numbers for each age group to match these percentages in the population!
- Example: For a sample of 100 students:
-> 70 participants under 25.
-> 20 participants aged 26–50.
-> 10 participants over 50.
Quotas in research design - how does it reduce sample bias?
Without Quotas:
- You might accidentally recruit mostly 18-year-olds (easier to find on campus), ignoring older students.
With Quotas:
- SO…improves generalisability (by reflecting OG population) + inclusion of all age groups
Exam q thinking - main goal of a sample?
- We take a sample then make INFERENCES on the whole population
- SO…goal of the sample is to always REPRESENT/MIMIC the overall population dynamic!!!
What if sampling bias couldn’t be controlled?
- DO NOT just get rid of the research
- Instead, just be TRANSPARENT and make this clear in your report
Who governs psychological research in Aus?
- NHMRC (national health and medical research council)
- They publish recommendations in the “National statement on ethical conduct in human research”
Ethical principles in psychological research? 5
1) Merit
2) Integrity
3) Justice
4) Beneficence
5) Respect
Merit
Research has benefit to humanity
Either in 2 forms:
- Providing knowledge to society
- Tangible (eg. improve social welfare)
Integrity
- Honesty - no manipulation of results / hypothesis
- Ie. anything that ruins public knowledge + understanding
Justice
EQUITY
- In terms of inclusion + exclusion of research participants
- No burden placed on one participant group
- Benefits of participation distributed fairly
PARTICIPANT TREATMENT
- Participants not exploited
- Money paid not proportional to time + effort - eg. paying 10 dollars for a 10 hour study
- Remuneration - eg. paying 300 dollars for a 15 minute task - pressuring participation
Beneficence
- Maximising benefits
- Minimising risks and harm
- INFORMED CONSENT
So…
- Must make clear of risks
- Not ethical if information about POTENTIAL risk/harm is not mentioned!!
Respect
- Individual’s culture, beliefs, welfare
- Respect privacy + confidentiality
- BUT asking qs on private, personal sensitive info is not considered unethical! Just need informed consent
Can research just be done?
NO
- Research proposal must be shown to an ETHICS ADVISORY COMMITTEE
- They will decide if the research can proceed using the 5 ethical principles
What does the research proposal include?
- Background of project
- Methodology
- How potential risks will be reduced
- Measures used (eg. surveys)
- Explanation of participant tasks
- Documentation
What does the research proposal include? - documentation 3
1) Plain language statement - clearly what the research is about in language w/o jargon
2) Consent form
3) Debriefing statement - after the research. Esp if there was deception involved
Bar graphs VS Histograms
1) APPEARANCE
Histograms = bars joined together/touch
Bar graphs = bars separated
2) X-AXIS
Histograms = continuous variable, range of numbers (eg. age, height) OR just one number (eg. score of 1,2,3,4)
Bar graphs = discrete variable, categories (eg. fruit, colours)
3) Y-AXIS
Frequency
4) EXAMPLE
Histograms = survey total 100 people. Want to find how many people in age groups 20-30, 40-50 etc.
Bar graphs = survey total 100 people. Want to find how many people like cats, dogs, birds etc.
Frequency distribution
- Way of organising data
- Tell us the number of times (frequency) a value occurs!
FORMALLY…
- Groups the no. of individuals
- Located at each point on the scale of measurement
How can we organise raw data?
- Impossible to make inductions from raw data
- Need to organise it into frequency distribution!
3 ways to represent frequency distribution
1) Histogram
2) Frequency table
3) Box plot
Histogram - 3 different shapes, what they represent
1) NORMAL DISTRIBUTION
- Looks like a bell-curve
2) POSITIVELY SKEWED
- Most data values take on LOWER values
- SO…tail at the end / right
3) NEGATIVELY SKEWED
- Most data values take on HIGHER values
- SO…tail at the front / left
What information can a frequency table include?
- Includes RELATIVE FREQUENCY for each category
- Can also present frequency as a percentage (shows proportion to total observations)
- Includes CUMULATIVE frequency for each category (shows how observations accumulate to total of 100%)
Box plots - 5 PARTS
1) LOWER WHISKER + LOWER PART = Bottom VALUE of 25%
2) BOX = Middle 50%
3) MIDDLE LINE = Median
4) UPPER WHISKER + UPPER PART = Top VALUE of 25%
5) OUTLIERS = represented by dots outside the whiskers.
(whisker = lines w/ caps at the end)
Top means top in terms of dependent variable
Box plots - what does it actually represent?
- Let’s say there are 20 students doing a test
- 25% = 5 students
- SO…for lower quartile, find the 5 LOWEST test scores (Eg. 15,17,20,30), then plot this onto the graph
Box plots - why is there skew if it’s always quartiles (ie. 25%)? Ie. why is the bottom line longer than the upper line?
- The quartile is always FIXED at 25% (eg. 5 people)
- BUT the RANGE of the lower values vary
- E.g. bottom 25% covers a low range of scores of 15-30, while top 25% covers large range of scores 60-100
SO…rmb box plots show SPREAD of values, not the actual frequency
Box plot - positive skew?
- Bottom whisker shorter than the top whisker
- Rmb we only know SPREAD, not FREQUENCY
- Bottom whisker means lower values not as spread out
- So here we look at SPREAD!
Scales of measurement - summarised 4
1) Nominal = scale no. represents discrete data (eg. 1=female, 2=male)
2) Ordinal = scale no. doesn’t show magnitude of interval - SAME AS NOMINAL, BUT shows order (eg. who came first + last in a race, 1st, 2nd, 3rd // 1=support, 2=moderate, 3=no support)
3) Interval = intervals between data is equal BUT no 0 point (eg. 10,15,20 degrees, 0 degrees doesn’t mean no temp)
4) Ratio = intervals between data is equal AND has a 0 point (eg. 0,5,10,15cm, 0 does mean absence of distance)
(Nominal + ordinal tends to have discrete data REPRESENTED BY numbers)
(Interval + ratio tends to be ACTUAL DATA - ie. no need to be represented)
Central tendency - definition
What is most TYPICAL or REPRESENTIVE of data
Central tendency - 3 measures
Mean (ie. average)
Median
Mode
Mean - how to calculate?
- Sum of all data points
- Divide by no. of data points
Eg. 1,2,3
Mean = 1+2+3 / 3
Median - how to calculate?
- Arrange data in rank order, then find the middle numnber
- OR Identify the 50% percentile of data (because 50% of data is above the median, 50% is below the median)
Limitations of calculating the mean?
When data is SKEWED or there are OUTLIERS
- Eg. workers earn 20 dollars per hour, but the boss earns 80 dollars per hour
- Including outliers in calculations would make the mean NO LONGER REPRESENTATIVE
SO…data for the mean must be NORMALLY DISTRIBUTED!!
Benefits of calculating the median?
- Outliers do not have an impact. As it is based on RANK ordering
- SO…suitable/can be used for skewed data w/ outliers (unlike mean)
Each category of data - what method of central tendency is best?
1) NOMINAL = discrete data, so MODE
2) ORDINAL = discrete data but ordered, so MEDIAN
3) INTERVAL = actual data, no representation by scale, so mean
4) RATIO = actual data, no representation by scale, so mean
BUT…if data is not normally distributed for the mean, MEDIAN will replace it, closer to show the representative
What is Q1 or Q3 of data?
- Q1 is the 25th PERCENTILE
- Eg. if data is test scores, Q1 is the score 25% of students scored less than this
- If Q3, 25% of students scored greater than this
AND
- It is the MEDIAN of the LOWER HALF of the data
What is the median in terms of quartile?
- It is Q2, ie it is 50th percentile
- 50% of data is less than the median, 50% of data is greater than the median
Finding median - odd number of data (eg. 7,2,9)
1) Rank data points (2,7,9)
2) Median POSITION = n+1 / 2
3) So…Median position = 4/2 = 2nd position
4) So…Median = 7
Finding median - even number of data (eg. 4,1,7,2)
1) Rank data points (1,2,4,7)
2) Median POSITIONS = n/2, n/2 + 1
3) So…Median position = 4/2 = 2nd position, 3rd position
4) So…Median = 2+4 / 2 = 3
Variability? Low and high
- How scores are SPREAD throughout distribution of data
- Refers to how data differ or not
- Eg. Low variability = scores are very close to the mean
- Eg. High variability = scores are spread out, short+ large distance to the mean
Variability? What does it say about the mean
- Low variability = mean is a good representation
- High variability = mean is a bad representation
SO…central tendency alone is not enough to tell us about the data
3 ways we measure Variability?
1) Range
2) Interquartile range
3) Standard deviation
Range? - definition, how to calculate
Difference of the largest and smallkest piece of data
Eg. 1,2,3,4,5,6,7,8,9,10
Range= 10-1=9
Interquartile range?
Distance between Q1 (25th percentile) and Q3 (75th percentile)
Finding IQR of even/odd number - summarised
- EVEN = split data into half with an imaginary line. Find the median of the lower + upper half
- ODD = remove/ignore median. Find the median of the lower + upper half
Finding IQR of even number 2,4,7,12,9,10
1) Order data = 2,4,7,9,10,12
2) Split data into half using an imaginary line = 2,4,7 // 9,10,12.
3) Calculate half by n/2
Find median of both halves. This will be Q1 and Q3
4) Q1 = 4, Q3 = 10
5) IQR = 10-4 = 6
Finding IQR of odd number 3,5,7,11,9
1) Order data = 3,5,7,9,11
2) Find the median (n+1 / 2) , and cross it out, REMOVE it!
3) Find median of both halves. This will be Q1 and Q3
4) Q1 = 3+5/2 = 4
5) Q3 = 9+11/2 = 10
6) IQR = 10-4 = 6
Standard deviation - definition
Average distance of data from the mean
Standard deviation - general gist of steps
1) Find distance of EACH data to the mean
2) Add all this up
3) Divide it by number of data points
Standard deviation - actual 5 steps
1) Find distance of EACH data to the mean = called DEVIATION SCORES
- Eg. if the mean is 5 and raw score is 7, DS = 2
- Eg. if the mean is 5 and raw score is 3, DS = -2
2) Negative distances is an ISSUE. When we total it up, it will cancel out. Hence we must SQUARE deviation scores
- (-2)^2 = 4.
3) NOW we can add all the SQUARED deviation scores up.
- The sum is called SUM OF SQUARES (SS)
4) Now divide this by number of data, BUT -1
SS/n-1
- n-1 represents degrees of freedom.
- This corrects for underestimation bias
- SS/n-1 is called VARIANCE
5) Standard deviation
- BUT this is all squared
- So… SD = root of variance
Deviation score - definition
Measure of raw data’s distance to the mean
Sum of squares - definition
- Sum of
- SQUARED deviation scores
- In a distribution
Variance - definition
- Sum of squares
- Divided by one less than the sample size
- SS/n-1
OR…squared standard deviation
Standard deviation - definition using specific terms
The AVERAGE deviation score in a distribution
If median was used to calculate central tendency, what measure of variability should be used?
INTERQUARTILE RANGE
- Because median and IQR are both not affected by outliers or unequally distributed data
- Because data is ranked
Ie. fit the same data conditions!!
If mean was used to calculate central tendency, what measure of variability should be used?
STANDARD DEVIATION
- Both work best when data is normally distributed + no outliers
Critical thinking?
An active process of scrunity
Critical thinking of research paper - what might it involve? 4
Basiclaly look at the entire research process circle we studied above
1) Whole research methodology
2) Deductive argumentation AND Inductive argumentation
3) Internal valdiity
4) External validity
How does deductive + inductive argumentation go together?
Deductive argumentation to derive a hypothesis AND Inductive argumentation to go from research findings to a conclusion
- forms the entire research process
Internal valdiity
Internal valdiity =looking for things that threaten confidence in IV-DV relationsihp
External validity
looking for things to see how generalisable the findings are to the real world + various populations
Logical reasoning - process in psychology?
2 ASPECTS
1) Premises
2) Conclusion
- We start with PREMISES (statements)
- Premises are meant to support the CONCLUSION
2 argument structures in psychological research?
1) Deductive argument
2) Inductive argument
Deductive argument
1) PREMISE
- Broad
- Form of broad theory
2) CONCLUSION
- Specific
- Form of specific hypothesis
Deductive argument - relevance to psychological research?
- This method is used to DERIVE a hypothesis (from broad theory)
- It doesn’t matter if the hypothesis doesn’t hold up, because the logic may be valid, however in REAL LIFE it doesn’t prove to be true!
Deductive argument - location in research paper?
- As it forms the hypothesis
- Tends to be found in the INTRODUCTION SECTION of research reports
How would we describe a “good” or “bad” deductive argument?
VALID/INVALID
- Premises must provide ABSOLUTE SUPPORT for the conclusion
How would we describe a “good” or “bad” deductive argument? - how do we determine validity? 3
- The premises DO NOT need to be true
1) We focus more on the STRUCTURE of the argument
2) Think - although the premises are true, is there a possibility of the conclusion being false?
3) Is there a GAP in the logic? - Ie. is something ignored or not considered?
Premise 1 = all australians love vegemite // Premise 2 = Francis is auatralian // conclusion = francis loves vegmite - what type of argument structure, good or not?
DEDUCTIVE ARGUMENT
- Premise = Taking broad statements
- Conclsuion = forming a specific hypothesis
VALID
1) Examine it in terms of STRUCTURE
Premise 1 = all students are happy // Premise 2 = all adults are happy // conclusion = all students are adults - what type of argument structure, good or not?
DEDUCTIVE ARGUMENT
- Premise = Taking broad statements
- Conclsuion = forming a specific hypothesis
INVALID
- Examine it in terms of STRUCTURE
- Assumes that if both groups possess the same quality, the groups are the same
- This is not true - possessing the same quality foesn’t automatically mean you are the same entity
Premise 1 = sunlight leads to happiness // Premise 2 = people who sleep during the day get no sunlight exposure // conclusion = group of night shift workers will report being unhappy - what type of argument structure, good or not?
DEDUCTIVE ARGUMENT
- Premise = Taking broad statements
- Conclsuion = forming a specific hypothesis
INVALID
- Examine it in terms of GAPS in the logic
- Doesn’t consider that things other than sunlight can lead to happiness in workers + wrong to assume night shift workers don’t sleep during the day
Premises in deductive arguments do not need to be true. If they are true, and it logically results in the conclusion, what do we call the aergument
Not just a VALID deductive argument
- It is called a SOUND deductive argument
Inductive argument
1) PREMISE
- Specific
- Tends to be an OBSERVATION/interpretation after conducting research
2) CONCLUSION
- Broad
How would we describe a “good” or “bad” inductive argument?
STRONG OR WEAK
- This is because data collected from research studies do not provide ABSOLUTE SUPPORT for the conclusion
- All data can do is either prove the conclusion to be MORE or LESS LIKELY to be true
How would we determine a “good” or “bad” inductive argument? 2
- TOTALLY different to deductive argument
- Deductive we look at logical structure etc - ABSOLUTE support
- Here we look at the quality of case study
->1) Sample size
-> 2) Representation / generalisbility
Premise 1 = 2012 study of 1500 psyc students found 95% enjoy studying// Premise 2 = 2017 study of 1600 psyc students found 90% enjoy studying // conclusion = likely that most psyc students enjoy studying - what type of argument structure, good or not?
INDUCTIVE ARGUMENT
- Premise = specific observation
- Conclusion = broad conclusion/observation
STRONG
- Good sample size
- Same findings in both studies
- LIKELY that conclusion is the case
Premise 1 = psyc student Liam likes the sunset // Premise 2 = psyc student sam likes the sunset // conclusion = all psyc students likes the sunset - what type of argument structure, good or not?
INDUCTIVE ARGUMENT
- Premise = specific observation
- Conclsuion = broad conclusion/observation
WEAK
- Bad sample size of only 2
- UNLIKELY that conclusion is the case
Premise 1 = 5 people diagnosed with schizophrenia, anxiety was associated with hearing voices. // Premise 2 = a person diagnosed with schizophrenia reported experiencing anxiety // conclusion = Hearing voices is likely to cause anxiety in people who have schizophrenia.
1) INDUCTIVE ARGUMENT
- Premise = specific observation
- Conclsuion = broad conclusion/observation
2) WEAK
- Super small sample size
- Premise 1 is based on correlation, not causal relationship (cause + effect) → broad GENERALISATION
- UNLIKELY that conclusion is the case
