Basic Statistical Concepts - Week 2 Flashcards
Scales of Measurement - (4 answers)
We can define MEASUREMENT as the assignment of symbols to events according to a set of rules. The particular set of rules used in assigning a symbol to the event in question is known as a SCALE OF MEASUREMENT.
The four scales of measurement that are of interest to psychologists:
- Nominal scale
- Ordinal scale
- Interval scale
- Ratio scale
Nominal scale - what are the other two names for it?
Is simply a set of category labels, identifying some things as Different OR the Same to other things (e.g., male/female, football jersey numbers).
Involves using numbers simply as codes for some attribute.
Is also known as CATEGORICAL or QUALITATIVE variables
Ordinal Scale
ORDERS a variable, however the DIFFERENCE BETWEEN the orders DOESN’T necessarily mean anything and are not necessarily equal.
e.g., We achieve an order of preference, however the actual difference between the ranks is Not known. A range of colours could be ranked in order of ‘Liking’.
The actual difference between ranks could be any magnitude.
Interval Scale
ORDERS a variable AND the DIFFERENCE between the orders HAS a legitimate meaning.
Assigns numbers to a characteristic, but in this case there is a strong mathematical relationship between the numbers, AS EACH INTERVAL IS EQUAL.
e.g., Temperature - the difference between 5 degrees and 10 degrees is the SAME as the difference between 10 degrees and 15 degrees.
Ratio Scale
ORDERS a variable, the DIFFERENCE between the orders HAS a legitimate meaning AND it has a true “Zero Point” (unlike temperature, where 0 is assigned arbitrarily).
e.g., There is NO such thing as No Temperature. It is, however, possible to have Zero weight, or Zero time.
There is a TRUE ZERO so that 4 is TWICE as much as 2.
Distance is a Ratio scale as 0 indicates No distance, and 10km is TWICE as much as 5km.
Inferential Statistics
Trying to reach conclusions that extend beyond the immediate data alone.
If…
Samples are used to make INFERENCES about characteristics of the general population.
Then…
INFERENTIAL STATISTICS is estimating the ACTUAL population parameters (or qualities).
Most of this unit deals with “how to do” inferential statistics.
How to establish ‘Cause-and-Effect’ relations and Why is it important?
Only when we Manipulate an IV and Control potential Extraneous variables are we able to infer a ‘cause-and-effect’ relationship.
It is important because mere observations, no matter how repeatable cannot tell you why those phenomenon occurred. Only when we give a cause-and-effect explanation do we begin to answer the WHY question.
Statistic
A summary measure from a sample is known as statistic, and it is used as an ESTIMATOR of the population parameter.
For the sample statistic to be a good estimator of the population parameter the sample must be REPRESENTATIVE of the population, and NOT be biased in any way.
The best way to achieve a REPRESENTATIVE SAMPLE is via RANDOM SAMPLING.
What are the 2 goals of Analysis?
- DESCRIPTIVE - simply to describe the sample (or sometimes the population), using techniques that ORGANISE and SUMMARISE the data.
- INFERENTIAL STATISTICS - to use sample statistics to make inferences about population parameters, or to use relationships found in a sample to make inferences about the relationships that exist in the population.
Inference
a conclusion reached on the basis of evidence and reasoning.
Internal Validity
Refers to the accuracy of any conclusions we draw about the Causal relationship between the IV and DV.
It is THREATENED to the extent that the observed relationship can be attributed to other things.
External Validity
Refers to the extent to which research conclusions can be generalised beyond the specific research context, that is, to different people, places, and times.
Discrete Variables
Nominal variables are sometimes referred to as Discrete variables, because they have FIXED VALUES, and it is not possible to have smaller values between them (e.g., you cannot really be .5 of a man).
Discrete Variables posses only a limited number of levels or states (e.g., male/female).
Continuous Variables
Posses many different levels (e.g., height)
Interval or Ratio variables are often continuous variables, because they can be broken up into any number of finer divisions.
Depending on how finely we measure the distance between two points there can be anything up to an infinite number of measures (e.g., 15km, 14.91km, 14.907km etc.)
However, Interval or Ratio variables can also be DISCRETE (e.g., number of children in the family).
Introduction to Research Design and Statistics - brief notes from Textbook
Despite the dread induced in students by research and, even worse, statistics, quantitative research design and analysis (statistics) are really based on very simple ideas.
If you understand the simple foundations you will not be so overwhelmed by the details, although you must be sure to retain CONCEPTS, as later ones build on earlier ones, and you will soon be overwhelmed if you forget what is learned at each step.
In essence, Quantitative research is about commonsense PATTERN RECOGNITION. One of the simplest forms of pattern recognition involves identifying what things occur together, so you can predict one on the basis of the other.
This in essence is what CORRELATION is about, but you need to be very careful with the interpretation. Just because things occur together does NOT necessarily mean that one causes the other. Perhaps the most fundamental issue of all is that of CAUSATION.
Usually, what we most want to identify are CAUSE and EFFECT relationships.
The commonsense way to do this is to EXPERIMENT.
If you think X might CAUSE Y, then MANIPULATE X and see if Y changes accordingly, and do this repeatedly so you can rule out CHANCE as an explanation.
