Research Modules

Construct

psychological concept

• eg. happiness can be measured by observing smth like smiling which we think is indicative of happiness

Operational definition

Operationalisation is how we make a construct measurable (indirectly)

• generally numerous ways

Research question + types

broad question/idea/problem we want to investigate - “eg. does happiness influence academic performance?”

• Association question

• Difference question

• Prediction (example above)/ causation

These questions come from detailed literature reviews

Use research question to form hypothesis

• should create a logical argument from literature review

• must be narrow and specific

• must be testable with observable/testable outcomes

• should be refutable - data should either be able to support + refute hypothesis

• should state that smth will happen - should never hypothesise that “no diff” - must be a positive statement

Variables

characteristic or condition that can vary amongst ppl

• can be classified as IV or DV

• continuous variable: allow decimals + fractions (distance, weight, time)

• discrete variable: seperate indivisible categories/ whole values

Types of data

• nominal

• ordinal

• interval

• ratio

Nominal

categories without order

• eg. different mental health disorders

Ordinal

categories with an order/hierarchy

• eg. groups of intelligence levels (smart, average, dumb)

Interval data

• numerical

• No true 0 value

• Equal interval between each number

  • eg. 1 degree to 2 degrees to 3 degrees

Ratio data

• numerical

• Does have a true zero

  • eg. height

Types of variables

discrete and continuous

Discrete

divisible

• 1,2,3 - no 1.5 etc.

• eg. number of people

Continuous

not divisible cleanly

• eg. height and weight - can be 10.5kg or 169.2cm

• doesn’t have to be an option that ppl can pick but if averaged out, if it’s a meaningful number it’s fine

  • eg. people rating self esteem on scale 1-10, although 4.5 is not an option, it’s a meaningful value whereas averaging people 3.5 ppl is not meaningful

Study design

Descriptive research design

no experimentation - merely observe + record naturally occuring instances

• eg. number of ppl sick in a class as percentage of class

Correlational research design

relationship bw two variables - would have to provide data for variable 1 and 2

• eg. exercise + happiness - its about association NOT causation

Experimental research design

1. manipulation of IV

2. carefully controlled experimental conditions enhance internal validity
   - decreases the likelihood of other explanations for results
   eg. random assignment of treatment vs no treatment to determine whether treatment effect - rule out biases

Quasi-experimental research design

less stringent controlled conditions but still trying to determine causation

• NO RANDOM ALLOCATION - eg. we use depressed ppl - we’re not causing the depression and manipulating IV - we’re using whats aready out there

Non-experimental research design

demonstrate relationship bw variables but don’t try to establish cause + effect

• 2 or more groups of ppl but one variable - eg. arts vs science students intelligence = faculty + intelligence but no causation

Population

everyone of relevance to a research study

• group of ppl from population in research = sample

• to be able to generalise findings from sample to pop sample must be representative

Probability sampling

simple random sampling: everyone in pop has equal chance of participating in study

• often not possible

Non probability sampling

unknown population characteristics

Convenience sampling

accessing people who are easily accessible

• likely to be biased so can set quotas for subgroups that need to be met in sample

Ethics

There’s a statement from the aus gvt (NHMRC)

• research must has merit: benefits to knowledge, welfare, scientifically sound

  • literature review will occur to ensure merit

• Integrity: honest reporting

• justice: inclusion + exclusion of people is fair and no burden placed on one type of population, equitable access to benefits

• beneficience: benefits outweigh/justify risks

• respect: must respect their cutlure, autonomy, beliefs and allow them to make decisions regarding

Frequency distribution graph: histogram

• less ppl gave extreme responses

• normally gives bell curve if netural = normal distribution

• positively skewed = skewed to left

• negatively skewed = skewed to right

Frequency tables: cumulative frequency

Box plot

lines are whiskers

Central tendency

what is most representative in a distribution/ what is most typical

Mean

the average

• use with interval/ ratio data and normal distributions

Median

middle score

• use with ordinal data and skewed distributions or when outliers are present

Mode

the most frequent score

• use with discrete and nominal data

Variability

refers to how scores in a distribution differ or not

Measures of variability

• range

• inter-quartile range

• standard deviation

Range

max score - min score

Interquartile range

1. rank data

2. split in halves and then get two middle values of each half and minus

can be used when mean used to measure central tendency

Standard deviation

the average amount that scores differ from the mean

Calculating standard deviation includes 3 steps

1. sum of squares (deviation scores squared added together)

2. variance

3. standard deviation

If we calculated all the deviation scores in distribution (how much each score deviated from mean)

would add to zero bc pos and neg cancel each other out

• thats why we square them

Sum of squares

sum up the squared deviation scores (the top part of average for example)

Variance caclulation

this is dividing sum of squares by degrees of freedom to correct underestimation bias if we just used sample size

SS/n-1

Standard deviation final step

Take the square root of the variance

• So square root(SS/n-1)

Logical reasoning

number of statements that are premises

Deductive arguments

start with broad/general premise to specific solution

valid =conclusion if guaranteed if all premises true

Inductive arguments

START FROM SPECIFIC

DOES NOT PROVIDE EXHAUSTIVE/ABSOLUTE SUPPORT FOR CONCLUSION

• provides probable support

• strong = premise provides good support

On JASP - what diff symbols mean

Ruler = ratio/interval

Bar chart = ordinal

3 circles = nominal

What to use on JASP for diff data

Radio and Interval: Histogram, Boxplot

Ordinal and Norminal: Bar chart, frequency table