1 Jargon & Basic Concepts

Download Report

Transcript 1 Jargon & Basic Concepts

Jargon & Basic Concepts
Howell
Statistical Methods for Psychology
Questions
• Define and illustrate:
–
–
–
–
–
–
–
Population, Sample
Parameter, Statistic
Descriptive, inferential statistics
Random selection (sampling), assignment
Internal, External validity
Discrete, continuous variables
Scale types (nominal, ordinal, interval,
ratio)
Population vs. Sample
• Population – collection of all the
objects of interest to researcher (you).
– College students, students at USF
• Sample – subset of objects from the
population
–
–
–
–
Want a representative sample
Samples are relatively practical
Random samples have good properties
One person’s sample is another’s
population
Parameter vs. Statistic
• Parameter – numerical summary of
population
– E.g., mean, standard deviation
• Statistic – numerical summary of
sample
– E.g., mean, standard deviation
• Typically we compute statistics and
estimate parameters using statistics.
Descriptive vs. Inferential
• Descriptive statistics describe a sample
– How tall are these students?
• Inferential statistics use sample
statistics to make decisions about
populations.
– Is one method of instruction better than
another?
Random Select & Assign
• Random selection is a process of picking a sample
from a population so that each element has the same
probability of being sampled.
– E.g., lottery, every 3rd name from a list (this is
actually a systematic sample but it’s good)
• Random assignment is assignment to treatment so
that each element has an equal probability of being
assigned to each treatment.
– E.g., lottery, every other name, etc.
• Both are typically accomplished by lists (aka frames)
and computer generated numbers (e.g., SAS PROC
PLAN)
Internal, External Validity
• Internal validity - quality of inferences
about the study itself. Random
assignment, history, maturation, etc.
• External validity – quality of inferences
from the study to the larger domain of
interest. Representative sample of
participants, task relevance, behavioral
consequents, etc. Aka generalizability
of the results (but not generalizability
study).
Variable & Distribution
• Variable vs. constant
– Attribute either varies across objects or not
• Distribution: Collection of data
• Distribution: Array of scores
–
–
–
–
Height
Beck Depression Index
Rat bar press
Wonderlic
Discrete vs. Continuous
• Math
– Integer vs. real numbers
• Data
– Categorical vs. continuous (many valued,
ordered)
• Examples
Political party, job satisfaction, response
time, country of origin
Scale types
• Nominal, ordinal, interval, ratio
• Nominal – categories. No ordering;
mean has no connection to attributes
• Ordinal – rank order only
• Interval – rank order plus equal interval.
ratio of differences has meaning
• Ratio – rank order, equal intervals,
rational zero point. Ratio of numbers
has meaning.
Scale Types: Footrace review
Nominal
Ordinal
Interval
Ratio
ID number Rank order Time of
of finish
day of
finish
043
1
10:57 a.m.
Elapsed
time from
start
4 min
011
2
10.59 a.m.
6 min
136
3
11:01 a.m.
8 min
112
4
11:02 a.m.
9 min
086
5
11:04 a.m.
11 min
Review
Find a partner to work on this exercise.
Suppose you want to know whether one brand of tennis
shoe is better than another. You have about $10K from a
grant to study this. Describe a study you might conduct
to find out. What might be your population, sample,
independent and dependent variables? What statistics
might you want to compute? Never mind the actual
statistical test at this point. What data would you gather?
What might a critic say about the internal and external
validity of your study? What scale types are your IV and
DV?