Professional Documents
Culture Documents
Ways of knowing
Types of Research
Experimental (most conclusive of methods)
Researcher tries different treatments (independent variable) to see their
effects (dependent variable)
In simple experiments compare 2 methods and try to control all
extraneous variables that might affect outcome
Need control over assignment to treatment and control groups (to make
sure they are equivalent)
Correlational Research
Looks at existing relationships between 2 or more variables to make
better predictions
Survey Research
Historical Research
Study past, often using existing documents, to reconstruct what
happened
Establishing truth of documents is essential
Quantitative v. Qualitative
Quantitative (numbers)
Facts/feelings separate
World is single reality
Researcher removed
Established research design
Experiment prototype
Generalization emphasized
Meta-Analysis
Locate all the studies on a topic and synthesize results using statistical
techniques (average the results)
Critical Analysis of Research (some say all research is flawed)
Question of reality (are only individual perceptions of it)
Question of communication (words are subjective)
Question of values (no objectivity only social constructs)
Question of unstated assumptions (researchers dont clarify assumptions
variable)
Dependent: the effect (aka outcome variable)
Extraneous: uncontrolled IVs
Disadvantages
Subjects must know the purpose of the study, possible benefits/harm; participation is
voluntary and they can w/draw without penalty any time
When deception used subjects they should be okay with it after (and they can refuse use
of their data)
Sampling
Any difference between population and sample is random and small (called
random sampling error)
Convenience sampling
Purposive sampling
Sampling
Measurement
1=poor to 5= excellent)
Interview schedules (complete scales as interview takes place; use
precoding; beware of dishonesty)
Tally sheets (for counting/recording frequency of behavior, remarks,
activities, etc.)
Flow charts (to record interactions in a room)
Anecdotal records (need to be specific and factual)
Time/Motion logs (record what took place and when)
Item Formats
Selection items or closed response (T/F; Yes/No; Right/Wrong; Multiple
choice)
Supply items or open ended (short answer; essay)
Unobtrusive measures (no intrusion into event usually direct observation
and recording)
Types of Scores
Raw scores (initial score or count obtainedw/out context)
Derived scores (raw scores translated to meaningful usage with
standardized process)
Age/Grade equivalence; Percentile ranks; Standard scores (how far a score is from a given
reference point, i.e. z and T scores);
Which to use depends on the purpose; usually standard scores used
Measurement Scales
Nominal (in name only)
Numbers are only name tags, they have no mathematical value (gender: 1=male and 2=
female OR race: 1= Blk, 2=Wht, 3=other)
Numbers show relative position, but not quantity (grade level, finishing place in a race)
Numbers show quantity in equal intervals, but an arbitrary zero (can have negative numbers;
degrees C or F)
Numbers show quantity with base of zero where zero means the construct is absent
Higher levels more precisecollect data at highest level possible; some statistics
only work with higher level data
Test/Retest give the same test (of enduring trait) to the same
people at two times and correlate the scores
Experimental Research
Weak Designs
One Shot Case Study (X O)
(O X O)
X1 O
Need control for diff subj characteristics X2 O
Static Group Pretest/Posttest Design (adds a pretest)
R X1 O
(Random assign to trtmt/cntrl, then posttest) R
O
Randomized Pretest/Posttest Control Group R O X1 O
(controls history, maturation, etc.)
R O X2 O
Match pairs on factors that influence DV then randomly assign to treatment or control
(subjects limited by no match elimination)
Statistical matching can be done using predicted scores
Survey Research
Types
Cross sectional provide a snapshot in time
Longitudinal collect data at different points in time to study changes
over time
Trend study - random sample each year on same topic
Cohort study - sample from same cohort members year after year
Panel study - same individuals surveyed year after year (mortality a
problem over long time periods)
Chapter 10
Descriptive Statistics
(Tools to summarize data)
Descriptive statistics describe many scores with just one or two indices
(such as mean or median)
Sample of a pop is described w/ indices called statistics
Entire pop is described w/ indices called parameters
Types of data (words or numbers)
Quantitative data scales measure how much (test scores, amount of
money spent, etc.
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Quantitative Data)
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Quantitative Data)
Smaller SD means less spread out, larger one means more spread out
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Quantitative Data)
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Quantitative Data)
A-4/5)
The probability for getting higher or lower than any given score can then be
calculated
T-scores are often used because negative z scores awkward (all T-scores
are positive)
Multiply z times 10, then add 50 (p. 212 Table 10.15)
Standard test scores often given with T-scores and percents above/below the
given score
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Quantitative Data)
Chapter 10 - Continued
Descriptive Statistics
(Summarizing Categorical Data)
Frequency tables
Give percents for ease in interpreting
Crossbreak or crosstabulations for relationships (IV goes on
the side, then give row percents)
Bar charts and pie charts used
Bars for ordered categories
Pies for unordered categories
Chapter 11
Inferential Statistics
Inferential Statistics
(The logic of inferential statistics)
Sampling error
Samples differ from their parent populations (no two samples are the same)
Chapter 11 - Continued
Inferential Statistics
(The logic of inferential statistics)
Once we can estimate the mean and SD of the sampling distribution can determine how
likely it is that a particular sample mean came from that population
i.e. Mean of pop=100, SD=10 and draw a sample with a mean of 110, yes could be from
that popbut if draw a sample with a mean of 140, most likely NOT from that popsince
is +4SEM from the mean (almost zero probability)
Express means as z scores; a z score move that 2SEM is going to occur less
than 5% of the time (2.5% each side)
Chapter 11 - Continued
Inferential Statistics
(The logic of inferential statistics)
Chapter 11 - Continued
Inferential Statistics
(The logic of inferential statistics)
Chapter 11 - Continued
Inferential Statistics
(Hypothesis Testing)
Inferential Statistics
(Hypothesis Testing)
Research hypothesis is what you really think is going on; opposite of the
null
Example of hypothesis test
H0 (null) is that mean1=mean2, meaning the mean scores are equal OR the
Chapter 11 - Continued
Inferential Statistics
Chapter 11 - Continued
Inferential Statistics
(Hypothesis Testing)
Inferential Statistics
(Inference Techniques)
Parametric tests (for quantitative I/R data from normal distributions of sample
size 30+)
Nonparametric tests (for categorical data and I/R from non-normal pops or
small samples)
Chapter 12
Statistics in Perspective
Approaches to research
Either 2 or more groups compared OR variables in 1 group studied AND
data are either categorical or quantitative
OR all three
Interpretation improves with experienceneed to know when something
statistically significant is not practically significant
Chapter 12 - continued
Statistics in Perspective
Chapter 12 - continued
Statistics in Perspective
Recap
Use graphics and numbers
Pay attention to outliers
Pay attention to magnitude of differences
Use inference tests for generalizing purposes and examine sampling
Use multiple techniques and CIs