all Flashcards
The empirical method
Empirical = through experience
Stages:
- Gathering data directly through out external senses
- Patterns and relationships within the data
- Problems:
o Without background theory is difficult
o Assumptions may influence -> not neutral
Planning research
- Research question
- Variables
- Samples
- Design
- Analysis
Quantitative research
numerical basis (counting, categories, measuring)
Qualitative research
meaning, experiences, descriptions, verbal reports
Observation
- Both for qualitative and quantitative research
- Technique within traditional research design or its own research
- Advantages:
o Immediate data on real behaviour rather than possibly distorted self-reports
o Gather data on behaviour
o In field settings data is unconstrained - Disadvantages:
o People’s behaviour affected by awareness of being observed
o Time consuming
o Identify cause and effect is difficult
Variables
- Independent variable (IV): Variable which the experimenter manipulates in an experiment -> direct effect on dependent variable
- Dependent variable (DV): Variable that is assumed to be directly affected by changes in independent variable
- Confounding variable: Variable that is uncontrolled
- Extraneous variable: any variable that could affect the results
True experiment:
- Manipulates the IV
- Holds all other variables constant (Inc. random allocation of conditions/ levels of the IV)
- Measures any changes in the DV
Levels and Factors
- IVs are terms of levels – different conditions are different levels
- Title of the IV covers the dimension (e.g. temperature levels: hot/cold)
- Multiple IVs are called factors – experiment have a factorial design (e.g. temperature and noise, both with at least two levels)
Control groups
- Baseline condition in the IV can be achieved by using control groups/ placebo group
- Placebo group might be using double-blind design
Strengths of experiments
- Can isolate cause and effect because IV and extraneous variables are controlled
- Alternative explanations and effects can be investigated (reproduction of the experiment)
- Can control many extraneous influences
Critique of experiments
- No unique view from participants unless interview after the experiment
- Reactive effects occur because participants know they are in an experiment
- Limited cause variables must be tight
- Results with false credibility
Experimental designs
- Independent sample design
o One group of participants in experimental condition
o Different group of participants in controlled condition
o Participant variables can confound the experimental results - Repeated measures design
o Same group of people is tested on different experimental conditions (i.e. levels of the IV)
o Repeated on each participant under the different conditions of the IV
o Multiple testing is not repeated measures
o Order effects
Effects from the order people participate
Counterbalancing in order, conditions, stimulus - Matched-pairs design
o All participants on come measure
Highest two scores in A and B
Second two highest score in A and B… - Single participant and small n design
o Useful investigation of cognitive deficits associated with specific medical conditions
o Useful when very long-terms training is required
Flied and laboratory
- Field studies are in natural environment -> capture natural behaviour
- Difference between field and lab is sometimes difficult to determine
Sampling for quantitative research
Sampling for research
- Population sample
- Rational
- Target population or sampling frame – specific description of the population of interest
Probability based sampling methods
- Equal probability selection method: producing sample; equal probability of being selected
- Needs random selection
- Use rng generators
o Simple random sample
Equal chance of selection
o Systematic random sample
Select every n-th case starting number randomly
o Stratified sampling
Large population with many subgroups
Identify relevant subgroups
Sample randomly
Related method cluster sampling (choosing an entire cluster frome each strata, e.g. an entire class from each faculty)
Non-probability-based sampling methods
- Quota sampling – define subgroups and collect data until the quota is fulfilled
- Self-selecting sample – often used in observations, field studies and when participants are volunteers
- Opportunity or convenience sample – test whoever is there
Sample size
- Small sample is more likely to be biased
- Sample size relates to the statistics for the analysis
- A larger sample size increases the power of a statistical test – more likely to detect effects
Quantitative data analysis
Need of statistics in research
- Statistics are a means to obtaining a clear, objective and fair conclusion
- Research is usually conducted on small samples
o Statistics relflect information about entire population - Helps to understand research results
- Summarize and describe our findings in an economical way (descriptive statistics)
- Make decision whether our results support our claims/ hypotheses (inferential statistics)
Levels of measurement
- Categorical and measured data
- Mostly, IV is categorical and DV is measured
- Data measured on:
o Nominal scale
E.g. two groups/ categories
Distinct categories
o Ordinal scale
Ordinal numbers represent the rank position in a group
Position but not distance between elemts
o Interval scale
Equal units
o Ratio scale
Intercal scale plus a absolute zero point - Higher level -> more information
- Level of measurement is determined by the type of measurement
- Level of measurement determines what kind of statistical test you can conduct
- Change is possible but only downwards
o Interval >ordinal > nominal
Descriptive Statistics
- Data can be described in terms of central tendency and dispersion
- Mean
o Arithmetic mean is used for interval/ratio data
o Sum of x / n
o Advantages:
Powerful statistics to estimate the population parameter from our sample, sensitive and accurate
o Disadvantages:
Easily distorted by outliers
Discrete values are difficult to interpret
- Median
o Used for data on ordinal level o Central value of a set of data o Not disturbed by extreme values o Does not account for exact distances o Median of 2,3,5,98,112 is 5 o Odd number median is (N+1)/2 o Even number take sum two central numbers / 2
- Mode
o Value that appeared most often
o Used for nominal scale value, mostly for discrete measurement scales
o Multiple mode possible
Measures of dispersion
- Describe the spread or distribution of the data around a central value
- The range
o Distance between top and bottom values
o Distorted by extreme values and is urepresentative of the distribution of values within the range
o Highest – lowest value
- Interquartile range
o Central grouping of values in a set
o Used for ordinal data
o Represents the calues that cut off the bottome and top 25% of values
o Not sensitive to outliers, but inaccurate when there are large class intervals
- Mean deviation
o Deviation value is the amount by which a particular value deviates from the mean
- Standard deviation and variance
o Most important measure of dispersion
