Glossary of Statistics- School Counseling

Start Studying! Add Cards ↓

Created by erikasalls

Independent Variables
Whats measured
The outcome
Depends on treatment(independent)

educated guess
null hypothesis
a hypothesis of no difference
Sampling methods

Random sampling
best to use
stratified sampling
arranged in some way (alphabetical or by number)
cluster sampling
Take by cluster
ex. classroom, or every 5th case ect.
purposive sampling
they have a purpose for the study
convenience sampling
convenient to use them
snowball sampling
get one and then they tell someone else who tells someone else and so on
multistage sampling
large areas targeted then smaller areas randomly
- then in that large sample, they randomly seclect a smaller area
Sample size
larger the sample size, less sample error
face validity
on the surface, does it appear to measure what it is supposed to measure- to those taking the instrument

Ex: students leaving the exam were asked if they felt the exam covered what they expected it to cover

does it elicit truthful responses

ex: I developed an inventory that measures how ofeten sutdents met with faculty to prepare for the Comprehensive Exam. I piloted it on my own class and found that students reported meeting with me three times more often that my records indicated

Experimental Approaches
Cause and Effect
- true experiment
-group experiment
-time series
- quasi-experimental or pre experimental

Research Designs: Time series
Time series Individual
Multiple baseline

individual behavior over time

Posttest only
Experimental group Treat PT
control group Control PT

-weakness, no pretest so you do not know where they began

One group pretest posttest
Experimental group PT TR PT

-no control group, nothing to compare it too

Factorial/independent groups comparison
Factor 1
x y
Factor 2


This factor compared to that factor

Mixed design
Pretest Posttest Control group
EXP Group PreT Tr PT
Control Group PreT control PT
Soloman four group
Perfect design

PreT exp Pret TR PT
Pret control PreT Cont. PT
Untest exp TR PT
untest cont Cont PT

Wait control
Immediate PreT T Post1 W Post2
Wait PreT W Post1 T Post2

pretest everybody

Risks to internal validity
practice effects
experimentor bias
demand social accept.
regression to mean

Risks: History
history of peoples roles

ex: In a nursing home and people are answering depression survey after nurse died. Obviously going to feel depressed because nurse took after them.

Risks: maturation
change over time

ex: drives 16-25 do drivers ed program for 16 years old and test at 25, is it because of program or maturation?

experimental only (fake)

are they changing because they know they are being studied
John Henry
control group works harder-knows they are being studied
experimenter bias
start doing what they think the experimenter wants
deman/social acceptability
respond most acceptable way
create own instrument to test
regression toward the mean
if you give a test and get big spread in scores, next time you give test mean will be almost identical but will be closer to the mean causing less variance
Risk to external validity
can we generalize the results to the population?
-population sample differences
-artificiality of research
- take participants out of normal environment and put in room with a 2 way mirror

high-low = range
mean of the squared deviations from the mean of the distribution
standard deviation
the square root of the mean of the squared deviations from the mean of a distribution reflects the typical deviation from the mean
have rank
no zero point
has absolute zero
how many scores fall in that interval
take frequency/ # of scores x 100= %
bar charts
- graphs the frequency
-compares something

line graphs
graphs cum. frequency

Pie Chart
graphs percentages
parts of a whole
Measures of central tendencies

Measures of variability
distance from mean for each score
Linear regression
predicting one variable over another
factor analysis
groups questions together to measure something
Assumes normal curve:
-anova (f test)

F test
compares 3 or more means
does not assume normal curve
-chi square
Type 1 Error
rejecting the null, accepting the alternative

-there is a difference, when in reality no difference

Type 2 error
Retain the null, reject the alternative

-saying there is no difference when really there is

Lower p value then..
more likely to commit type 2 error, less likely to commit type 1
increase sample size...
will decrease type 2 error
chi square
nominal data
compares two or more sets of data
assumes random sampling
no cell is less than 5 cases

Pearson R
-straight line relationship
-both variables are at least interval
-random sampling
-normal distribution in sample under 30

-1.00 Perfect neg
-.60 strong neg
-.30 mod neg
-.10 weak neg
0 no correlation
.10 weak pos
.30 moderate pos
.60 strong pos
1.00 perfet pos

Factors that affect reliability
Test length
objectivity of scoring
variability of groups
difficulty of items

Risks to test validity
History: influence outside events
maturation: normal changes
testing threat: pretesting influences post testing
instrumentation: alternate forms may not be equivalent
statistical regression interaction: combination of the above

Research Design
-General idea: know more about the research
-review the literature: go out and review what we already know
-measurement: subjects, who is it were studying
-subjects: who is it were studying
-variables: what are we studying
-group assignment: random, non random
-analysis plans: analyze data
gathering data: experiment
-Treatment & Controls:
-Data Analysis: analyze data that was collected
-report results

Ethics for Research
Subjects rights:
-No harm to subjects
-right to privacy
-data confidential
informed consent
-purpose for study
-what is being done
-potential benefit
-potential harm
-right to withdraw

Add Cards

You must Login or Register to add cards