Research- Break Down Abstract Question & Needed Info

Experimental Research

controlled manipulation of the independent variables (IVs)

-must have multiple groups

Pretest-Posttest Control-Group

R O X O (Experimental group)

R O C O (Control group; i.e., no intervention)

-has to have more than one group

-used for both within- and between- subjects components

(Experimental)

Posttest-Only Control-Group

R X O (Experimental group)

R C O (Control group; or R Xn O for comparison group)

-cannot make pretest measurements

(Experimental)

Nonequivalent Control-Group

O X O (Experimental group)

O C O (Control group)

-only used when non-randomly assigned control group

-does not control factors you are not intentionally studying that may affect outcome

-threats to internal validity: history, maturation, attrition

(Experimental)

Single-Group Pretest-Posttest

O X O

-no control group

-no manipulation = not experimental

Time Series

O O O O Xn O O O O

- Multiple measurements pre and/or post treatment

- Used to establish a baseline when a control group is not available

- 1 group = non-experimental

Repeated Treatment

O X1 O X2 O X3 O

-same subjects measured under all levels of IV

-need "wash-out" period & counterbalance treatment order

(Experimental)

Multiple-factor Experimental

- design with two or more IVs

- look at interaction of two (or more) variables

- Interaction: the effect of the 2nd IV depends on the level of the 1st IV = result is different from simply adding the two effects together

- Look at efficacy of combined treatments

Non-Experimental Research Design

- does not involve manipulation of IV (s)

- ONLY ONE GROUP

- Must take things as they are

- Cannot make conclusions about cause-and-effect

Dimension- Time

Must include two words

1) retrospective vs prospective

-retrospective = data collected before research design

-prospective = data collected after research design & completion

2) Cross sectional or longitudinal

-cross sectional = data collected at one point of time

-longitudinal = data collected over period of time

Dimension - Degree of Control

Three Options

1) Non-experimental - one group

2) True Experimental - random assignment

3) Quasi- Experimental - without random assignment

Dimension - Purpose

3 options:

1) description - document the nature of phenomenon through systematic collection of data

2) analysis of relationship (s) - assess relationships between variables/measures

3) analysis of difference (s) - focus on whether groups or treatments are different

Potential Ethics Question Answers

- Personal Autonomy = Informed Consent

- Beneficence = obligation to help all participants & maximize possible benefits and minimize risks/harm

- Justice = Fairness in distribution of both burdens & benefits of research

Identify Independent Variable(s)

- What the IV is = what is manipulated

- Levels = how many IVs there are

- Within or between?

*within (one group getting all treatments) or between (different groups getting different treatments)

- Active or Attribute?

*Active (can place subject in any group) or attribute (subject has characteristic that places them in specific group - researcher can't sort subject)

Identify Dependent Variable (s)

- what the DV is = what the outcome is

- level of measure = nominal, ordinal, interval, ratio

*nominal- numbers are category

*ordinal- numbers can be ranked

*interval- has order and equal distance (+ and -)

*ratio- order, equal distance, true zero (+,-,x, and /)

Is there order, equal distance, and/or true zero = for NOIR classification

History (Threat)

(Internal Validity Threat)

Events unrelated to the treatment occur during the course of the study that may affect the outcome (DV)

Maturation (Threat)

(Internal Validity Threat)

- Changes in DV are simply a function of passage of time

- Subjects become more experienced, older, stronger, bored, etc.

Attrition (Threat)

(Internal Validity Threat)

- Loss of subjects before the study is completed

- Especially if a differential loss of subjects occur for specific reasons related to experimental situations

Testing(Threat)

(Internal Validity Threat)

-changes in DV simply as a function of testing (e.g. repeating, presence of the testing, etc.)

Instrumentation (Threat)

(Internal Validity Threat)

- Spontaneous changes in the measurement tools lead to changes in DV

Regression (Statistical) toward the mean (Threat)

(Internal Validity Threat)

Subjects "accidentally/uncharacteristically" demonstrate an extremely high or low score at pre-test...

Diffusion or imitation of treatments (Threat)

- social threat

- Subjects are aware of the interventions or sharing information among various treatment groups

- blind the subjects to control it

Compensatory equalization of treatments (Threat)

- Social Threat

- When the treatment is considered "desirable" by the researchers, they may provide compensatory actions/services to the control or less desirable groups

- blind the researcher to control it

Compensatory Rivalry or Resentful Demoralization (Threat)

- social threat

- When subjects perceive receiving less desirable treatment, they may try to compensate (e.g., exercise more) or become less motivated in their performance

-blind subjects to control it

Simple Random Sampling

(Probability)

- Every member has equal chance

Systematic sampling

(Probability)

- Select every nth point on the list from a random starting point

Stratified (Proportional) random sampling

(Probability)

-ratio of characteristics in sample match ratio of characteristics in population

EX: 30% men and 70% women ratio is found in both population and sample

Disproportional sampling

(Probability)

- When more samples from a specific characteristic is required

EX: minorities and non-minorities have equal representation in study

Cluster or multistage sampling

(probability)

- Successive random sampling of a series of units in population

EX: sample from 10 states, 5 hospitals per state, 10 employees per hospital

Convenience Sampling

(non-probability)

- Subjects chosen based on availability

EX: first come, first serve basis

Quota Sampling

(non-probability)

- stop recruiting volunteers with a specific characteristic once the proportion of such characteristic in the sample is met

Purposive Sampling

(non-probability)

Hand-pick subjects based on specific criteria desired by the researchers

Snowball Sampling

(non-probability)

Carried out in stages by asking subjects who have been tested to identify other potential subjects

EX: You get asked to bring a friend to a study

External Validity

Can the results be generalized to other persons, settings, or times?

Construct Validity

To what constructs can results be generalized?

Internal Validity

Is there evidence of a casual relationship between independent variables and dependent variables? Are changes in DV explained truly and only by level of IV?

Statistical Conclusion Validity

is there a relationship between the independent and dependent variables? Correct use of statistical analysis?