# Research Methods Exam 2

## 55 terms

### Single Subject Designs

A quasi-experimental design in which we access a change in a dependent variable in a single research subject

### Usefulness of single-subject designs

-allow us to evaluate intervention over time
-provide immediate and practical feedback
-inexpensive
-capture the process of change over time
-do not use a control group (quasi-exp.)
-avoid ethical concerns about withholding interventions for the sake of an experiment

### Knowledge-Level Continuum

What is the preexisting knowledge in the problem area?

-Define
-Operationalize
-Select
-Establish
-Quantify
-Graphing

### Single-subject process (define)

Define the problem with the client-some aspect of the client's behavior, perception, feeling or attributes

### In a single subject design, how do we select the problem?

-most clearly experienced by the client
-has the most negative consequences
-has a high probability of success

### Single-Subject process (operationalize)

Operationalize the problem-identify indicators and establish goals

### Why is it useful to have multiple indicators when operationalizing a single-subject design?

-# of times take meds
-Duration of diet
-number of times a child cries or argues
-w/agency-# of times of grievances filed, staff turn-over

May need to proximate (intermediate) goals

### Single-subject process (select)

Select appropriate single subject design

### Single-subject process (establish)

Establish and measure a BASELINE (a measure of the client's functioning prior to intervention)
-possible data sources for measurement
-direct observation of client should be
-unobtrusive
-multiple
-undifferentiated, between baseline and treatment phases

Duration
Frequency
Magnitude

### Graphing

-X or horizontal axis shows time units of observations
-Y or vertical axis shows frequency of occurrences of behavior of interest

### Design relates to type of research question

Exploratory
Descriptive
Explanatory

A=Baseline
B=intervention phase
C=an intervention other than B

A-Baseline
B
BB1

### BB design

Time is divided into two phases
-B phase
-B1 phase

### Descriptive Case-Level Designs

-AB
-ABC or ABCD

Because you have a baseline measure (A) you can conclude, with some level of surety, that there is a relationship between the change and the intervention

### Explanatory Case Level Designs

-attempt to get at the difficult question of causality
-want to control for intervening variables
-reversal designs
-multiple baseline designs

### Explanatory Reversal

-ABA
-ABAB (withdrawal designs)
-BAB (oops, forgot initial baseline)
-BCBC (compare effectiveness of two interventions)

### Explanatory Multiple Baselines

-More than one case, vary timing of intervention. What type of threat to validity does this address?
-More than one setting, vary timing and setting for one client
-more than one problem, same intervention

### Survey Research

Data collection via gathering respondents by having them respond to questions
-standardized, systematic data collection from a systematically selected sample

### Value of survey research

-used for exploratory, descriptive and explanatory studies
-Best method for collecting original data from large populations, enhancing generalizabiity
-Good for measuring attitudes and orientations via a series of questions
-strong reliability

### Limitations of survey research

-can't measure thoughts, feeling, behaviors

### Before conducting survey

-what do I want to know?
-Who are the best informants or sources of info?
(who will be in sample? How? How to select?)
-How should I collect info?
-How should I analyze the data?

### Structuring written survey instrument

-Professional appearance
-title
-sequencing of questions
-group like questions
-first question should be interesting to respondents
-only ask questions yo need answers for that relate to study/research question
-Avoid double barrel questions (2Q's in one)

yes/no
multiple choice
t/f

### Pre-testing

First on colleagues
Potential users of data
Sub-sample of true sample

Make sure questions are understandable

### Implementing a mail survey

Cover letter:
-explain what survey is
-enclose SASE
-explain confidentiality
-explain who you are, what your study is

### 5 kins of surveys

Mail
Group
Phone (allows for follow-up/probe questions)
Interview (in person)
Electronic

### Reactivity

People act differently if they know they are being observed

### Contingency questions

dependent on something...

### Matrix questions

A grid that indicates how often something occurs.

ex. How often do you engage in the following sports?....

### Triangulate

get data from 3 different sources

### Experimental designs

A controlled method of observations (collecting data) in which the values of one or more independent variable(s) is/are changed to assess its causal effect on one or more independent variables.
-the intervention is the independent variable

### Experimental design strengths

they allow us to assess if a theory is correct

they allow us to isolate the cause and effect relationship, ruling out rival explanations

### Cause & Effect relationship

when all other things are equal, the variation in the IV will be followed by variation in DV
-cause must come first
-association: some connection between IV & DV
-non-spuriousness: the correlation between the IV & DV can't be explained by a 3rd variable

### Internal validity

refers to the degree of confidence that the study's results accurately state that one variable is or is not the cause of a change in another variable
-Helps us understand how much we can generalize the study's findings
-validity determines how well the research has controlled threats to internal validity

### External validity

the degree to which results of research are generalizable to a larger population or settings outside research setting

### Controlling rival hypotheses

-Controlling extraneous variables (extra, outside)
-Using correlated variation
-using analysis of covariance-when 2 groups are not equivalent
-Use a control group
-Randomly assign research participants to groups

### R

R=Random assignment to control or experimental groups

O=observation

### X

X= intervention (change in intervention status)

### Pre-experimental designs

One shot case study
-X O
(Intervention & Observation)

### One group, pre-test, post-test design

O X O

(observation, intervention, observation)

Assess DV before & after intervention to assess correlation (change in IV)

### Post-test only design with non-equivalent group

X O (group 1)
O (group 2)

(intervention, observation)
(just observation)
Compare 2 groups after experiment during which only one group had intervention

### True experimental designs

Used to control for threats to internal validity/rival explanations
-not always feasible, ethical
-components:
random assign to groups
introduction of intervention (change IV)
comparison of the amount of change between groups

### Characteristics of "ideal"/true experiments

-Time order (sequence) of IV must be established
-IV must be manipulated
x present vs. x absent
small amount of x vs. large amount of x
x vs. something else
-Relationship between IV & DV must be established

### pre-test/post-test control group design

Classic experimental design

R O1 X O2 (experimental group)
R O1 O2 (control group)

Controls most threats to internal validity because of R (random sample) and pre-test

### Post test only control group design

R X O (experimental group)
R O (control group)

assumes 2 groups similar without pre-test so any differences is due to intervention

measures not between pre and post tests, but between 2 groups after intervention provided to 1 group

### Multiple Experimental Groups with one Control Group design

R O X1 O (one intervention, but
R O X2 O each group has a
R O X3 O different amount/duration)
R O O (control group)

### Factorial Design - 2 IVs with 2 values (absence or presence)

Exp Grp 1: R X1 X2 O
2: R X1 O
3: R X2 O
Control: R O

### Quasi-experimental Designs

Used when it is unethical to withhold intervention or to create a true control group
-Lacks randomization (no R)

### Non equivalent control groups

Groups seem logically comparable
Called comparison (not control) groups

O1 X O1
O2 X O2

### Simple Time-Sequence Designs (Longitudinal case study)

Simple interrupted time series design

O1 O2 O3 O4 X O5 O6 O7 O8

Problem-need to rule out other explanations

### Multiple time series design-interrupted time series w/non-equivalent control group

O1 O2 O3 O4 O5 X O6 O7 O8 O9 O10
O1 O2 O3 O4 O5 O6 O7 O8 O9 O10

Create Set