RESEARCH Flashcards
emperical
collection of data
nonemperical
use existing info, lit/sys review of previous research
qualitative
understand human behavior/ person’s perspective
qualitative data collection
observation/interview
qualitative data analysis
themes/ descriptions, “verbal”
quantitative
finding facts/ data
quantitative data collection
measurements
quantitative data analysis
numerical comparisons, statistical analyses, “numerical”
which type of research can be experimental or non-experimental?
quantitative
basic research
obtaining knowledge
applied research
solving some immediate problem
experimental
experimenter “manipulates” one or more factors
true experimental
researcher creates conditions by manipulating factors
RANDOM ASSIGNMENT of participants to different experimental groups
quasi-experimental
researcher creates conditions by manipulating factors
NO RANDOM ASSIGNMENT
non-experimental
researcher investigates existing conditions; case studies, surveys, comparison or case control studies
beneficence
human subjects should NOT be harmed
BENEFITS should be maximized
respect for persons
participants should be treated as AUTONOMOUS agents & make OWN decisions
justice
equitable decisions on who will PARTICIPATE, benefits/ risks must be distributed fairly, fair procedures & outcomes in selection of subjects
hypothesis
predicted outcome
null hypothesis
assume that there will be NO significant results
alternate hypothesis
assume that there WILL be significant results
reject the null hypothesis
shows a SIGNIFICANT relationship/ difference
response researchers want
reject the null hypothesis
fail to reject the null hypothesis
NO significant results
directional hypothesis (difference study)
hypothesis stated in a way you expect one result to be better/ worse than the other
directional hypothesis (relational study)
hypothesis stated in a way you expect a specific relationship (positive/ negative)
non-directional hypothesis
hypothesis states there will be a difference or a relationship, but the difference can be either positive or negative (does not state direction)
face validity
judgement as to where the variables measure what they are SUPPOSED to (logical validity)
weakest form of validity
face validity
content validity
involves judgment, but more FORMAL than face validity
panel of expert reviewers
content validity
ecological validity
does this test apply to the REAL WORLD of the person being tested
criterion validity
comparison to an EXISTING test/ measurement; have a “standard” to compare against
concurrent validity
criterion
give participants “new” test and standard test CLOSE in time
predictive validity
criterion
give participants “new test and compare against some standard LATER in the future
construct validity
theories have constructs: set of behaviors w meaningful associations within the theory
to test construct validity, developers have
looked for patterns on certain test items
convergence
construct
happens if person performs SIMILARLY on test questions or if two different tests have SIMILAR content
divergence
construct
happens if person performs DIFFERENTLY on test questions that measure DIFFERENT constructs
interobserver reliability
DIFFERENT examiners/ observers testing the SAME person
participant test twice
interobserver reliability: similar results
= strong correlation
test-retest reliability
participants tested two different times; typically, SAME examiner
correlation coefficient
degree of reliability
parallel-form reliability
two DIFFERENT forms of the SAME construct or Ax
test form A and test form B, form A, then (at a later date) form B
parallel form reliability
split-half reliability
administer test ONCE to participants, SPLIT tests into 2 equivalent forms, get score for each section/ half
first half/ second half; odd number questions/ even number questions
split-half reliability
internal validity
outcome of study happened for hypothesized reasons
other factors that could have influenced results?
rule of influence of extraneous variables
external validity
can the study results be EXTENDED to the gen pop? to individuals other than the ones who were participants in the study?
generalizability can relate to populations, settings, treatment variables, measurement variables
survey research
non-experimental
obtain opinions, find individual experiences, obtain info from a LARGE number of participants
longitudinal research
non-experimental
individual or group follow OVER TIME
may be associated w other designs & experimental research
assess effects of Tx over longer time period
ex-post facto
non-experimental
independent variables have ALREADY occurred; studying the DEPENDENT variable
retrospective research design
Ex-Post Facto (non-experimental)
normative research
non-experimental
distribution of selected dependent variables across different age groups are observed and recorded
NO manipulation of independent variables
cannot draw cause/effect relationship
two types of normative research
cross sectional
longitudinal
post-test only randomized control group design
experimental
randomize groups, only one group gets Tx, both groups get a post-test
post-test only randomized treatment group design
experimental
randomize groups, groups get different treatments , both groups get a post-test
post-test only randomized treatment and control groups design
experimental
randomize groups, groups 1&2 get different treatments, 3rd group gets NO treatment, ALL groups get a a post-test
post-test only randomized treatment group design
experimental
randomize groups, 3 groups get different treatments, all groups get a post-test
pre-test - post-test randomized control group design
experimental
group 1 - observe/ measure, Tx, observe/measure
group 2 - observe/ measure, NO Tx, observe/ measure
pre-test - post-test randomized treatment group design
experimental
group 1 - observe/ measure, Tx 1, observe/ measure
group 2 - observe/ measure, Tx 2, observe/ measure
pre-test - post-test randomized treatment and control groups design
experimental
randomize groups,
group 1 - receive 1 Tx
group 2 - receive a diff Tx
group 3 - NO Tx (control)
ALL groups get a pre-test and a post-test
pre-test - post-test randomized treatment groups design
experimental
randomize groups, 3 groups each get a different Tx, ALL groups get a pre-test and a post-test
solomon randomized four-group design
experimental
two control groups: one receives pre/post tests and one receives post test ONLY
two experimental groups: one receives pre/post tests and one receives post test ONLY
switching replications design
experimental
random assignment, both groups will eventually get Tx
pre-test for both groups, group 1 receives Tx, both get post-test 1, group 2 receives Tx, both get post-test 2
factorial designs
investigators manipulates two or more variables
IV is a factor; each factor has two or more levels
factorial designs
2x2
2x3
2x2x2
threats to internal validity
history
statistical regression
mortality
maturation
instrumentation
selection
history
something that happened during course of study that could have impacted the study
statistical regression
if participants are tested over time, scores will tend to be around the mean
mortality
participants drop out before the end of the study
maturation
improvements over time d/t growth & development, esp important in research w children
instrumentation
physical instruments, human “instruments” that record data, observe behavior
selection
happens more often if selection is not random, would decrease ability to define a cause/effect relationship
nonequivalent control group design
quasi experimental
2 pre existing groups - 1 assigned to experimental condition and 1 assigned to control condition
repeated measures group design
quasi-experimental
1 group of participants, multiple measures/ observations
could be experimental or non experimental
population
all persons of interest for a particular study
sample
individuals who participate in a study - small portion of the larger population
random sampling
start with a list of potential participants and randomly choose the required number of participants - random # generator, etc
systematic sampling
start with a list of potential participants - establish sampling interval to select participants
stratified random sampling
researchers identify 1 or more strata
cluster sampling
researcher defines groups (clusters) and randomly chooses units/ participants from the cluster
sample size
increasing sample size should increase how well the sample represents the population; however, increasing sample size does not decrease systematic bias
nominal
naming, responses in categories, participants fit into only ONE category
ordinal
rank ordered, does not tell info about difference between rankings
interval
info on rank ordering AND how much they differ; does NOT have a true zero
ratio
same components as interval but HAS a true zero
mode
based on frequency of occurrence - the score that occurs most often
primary measure for nominal, but can be used with all
mode
median
number at the midpoint of all scores
put numbers/ measures in order
useful for ordinal, interval, and ratio
median
mean
average - sum all the scores then divide by total number of scores
most appropriate for interval and ratio levels
mean
extreme outliers will affect
mean more than median!!!
range
difference between minimum and maximum
interquartile range
difference between 75th quartile and 25th quartile
interquartile range is less impacted by
extreme scores than the range
standard deviation
for interval or ratio measures - measure of variability
dispersion of score around the mean
confidence interval
mean of the sample +/- margin of error
gives you a range
the larger the confidence interval, the larger the
RANGE and more likely the score falls within the range
α value
level of significance
p value
probability of rejecting the null hypothesis
p value has to be ___________ than chosen α value to be considered significantly significant
SMALLER
0.3-0.5
WEAK
0.51-0.85
MODERATE
0.86-0.95
STRONG
positive direction
both variables change in the same direction
negative direction
variables change in different directions
analyzing differences between 2 groups/ conditions
T-test
Z-test
analyzing differences for more than 2 groups/ conditions
ANOVA
analyzing differences between 2 samples
independent t-test
paired t-test
Z-test
ethnography goal
to describe a specific cultural environment
immerse in environment
grounded theory goal
data based theory of an observation (NOT a theory based one)
collect any type of data
code into themes
phenomenology goal
to get a feel of what “it” is like to participants
collect data - mainly interviews
conversation analysis goal
describe details of communicative behaviors in natural situation
collect data of participant interacting with someone
analyze sequence
single subject design
ONE subject
subject serves as his/her own control
quantitative, empirical, and objective
quasi-experimental: NO control group or randomly assigned subjects to Tx
Single Subject Design: Baseline Tx (AB)
visual changes in level, trend, and variability
ideal baseline: low variability, no improvement trend, at least five repeated measurements
pre-experimental
Single Subject Design: Tx-withdrawal design (ABA)
effect of single Tx (IV) on an outcome measure (DV)
STRONGER than AB design: provides second demonstration of Tx effect
Single Subject Design: Tx-replication design (ABAB)
extension of ABA
provies 3rd demonstration of Tx effect on outcome measure
weakness of ABA design
not all Tx are reversible
straightforward and among strongest single-subject designs
Tx-replication design (ABAB)
A researcher is conducting a study on cycles approach in elementary aged children with
phonological disorders. The study follows the students from kindergarten to third grade and
measures their improvement. A new school opened up in the area and many students in the
study will be transferring due to new district policies. The biggest threat to internal validity is:
Mortality
A study is investigating the effects of language intervention on students with learning disabilities.
The researchers have two groups: one with no intervention despite likely benefiting from it, and
a group who receives weekly intervention. The largest risk in this study would be:
beneficence
A study was completed in ranking the intelligibility of preschool aged children with speech sound
disorders. They were ranked from 1 (most intelligible) to 30 (least intelligible). There is no
information regarding the distance between rankings. Which form of data analysis is this?
Ordinal
A research study is looking at the effectiveness of a new speech-language therapy approach for
stuttering, the researchers hypothesize that there will be no significant difference in stuttering
severity between the treatment group and the control group. What type of hypothesis is this?
Null Hypothesis
n SLP graduate student is doing qualitative research on individuals with swallowing disorders.
They are conducting interviews to gain an understanding of what it’s like to deal with the effects
of altered diets and problems swallowing. What qualitative research method is this?
Phenomenology
Purpose of non-experimental research design:
observe/ describe features
determine relationships among variables
compare groups with different characteristics
In a study investigating the effectiveness of a new therapy approach for improving
speech and language in children with sensorineural hearing loss, the researcher has chosen an
alpha level of 0.05. If statistics are run comparing pre- and post-tests and findings show p = 0.15
the results are NOT statistically significant, and the null hypothesis should be accepted
An experimenter observed a +0.40 correlation between difficulty toilet training and
severity of stuttering. This correlation indicates that:
Difficulty with toilet training and stuttering may both be related to a third factor, or they may be casually related