Experimental Research Designs

Question 1

What are experimental designs?

Answer 1

Experimental designs entail the manipulation of a variable (IV) to test the effects of the manipulation on some outcome (DV).

Needs to have manipulation.

Question 2

What is causal research?

Answer 2

Causal research can be defined as a research method that is used to determine the cause and effect relationship between two variables.

Conclusive research.

Question 3

What are the two objectives of causal research?

Answer 3

1) Isolate etiological causes of HB and disease outcomes

2) Determine whether interventions are effective in achieving intended outcomes

Question 4

What are the three necessary conditions for causation?

Describe them.

Answer 4

1) Relationship condition
Variable A and variable B must be related.

2) Temporal antecedence condition
Proper time order must be established

3) Lack of alternative explanation condition
Relationship between A and B must not be due to third confounding variable

Question 5

What are defining characteristics of experimental research design?

Answer 5

Manipulation of IV and control over extraneous variables.

Question 6

What is the IV?

Answer 6

The variables chosen by the INVESTIGATOR to determine its effect on the DV.

Is MANIPULATED by the investigator.

Question 7

What is the DV??

Answer 7

The variable(s) that is being tested.

Question 8

What are extraneous variables?

Answer 8

Not directly related to the hypothesis being tested but may have an effect on DV.

Question 9

How do extraneous variable threaten internal validity?

Answer 9

Extraneous variables may provide alternative explanations or rival hypotheses that threaten internal validity.

Question 10

What is internal validity?

Answer 10

The extent to which our study allows us to answer the question we want to answer

Extent to which study establishes cause and effect RL between IV and RV

Ability of a design to test hypothesis it was intended to test.

Question 11

What is external validity?

Answer 11

The extent to which findings of the research study can be generalized to the population under study.

Question 12

What is efficaciousness versus effectiveness?

Answer 12

Efficacy:
Having an effect.
Performance of an intervention under ideal conditions.

Effective:
How well it performs in real world.

Question 13

What is a one-factor versus two-factor design?

Answer 13

One factor has only one thing being treated (no control).

Question 14

What are the 9 threats to internal validity?

Answer 14

Hawthorne effect
Diffusion
History
Maturation
Testing
Instrumentation
Statistical regression 
Differential selection
Differential attrition

Question 15

What is the Hawthorne effect?

Answer 15

The alteration of BH by subjects of a study due to their awareness of being observed (need to have control group to avoid this).

Threat in one-group and two-group without placebo.

Question 16

What is diffusion?

Answer 16

Distinction between intervention and control group are not as separate as they should be (results might be due to contamination of control group); maybe people from same household cannot be in the same study - can try to control with study design.

Threat to two-group or multi-group design.

Question 17

What is history threat?

Answer 17

Any event that might have occurred between observations outside of trial.

Threat to one-group design.

Question 18

What is maturation threat?

Answer 18

Results might be due to PPTs growing older, wiser, stronger, etc. between observations.

If history and maturation are threats, include control group.

Threat to one-group design.

Question 19

What is testing threat?

Answer 19

The results are due to the number of times responses were measured (PPTs have answered questions before that might effect how they will answer questions in the future)- people might just get better with a test.

Control group needed as you would expect testing effect for both groups so any differences between the groups would be due to other reasons.

One group threat.

Question 20

What is instrumentation threat?

Answer 20

Results might be due to a change in the measuring of instrument between observation.

One group threat.

Question 21

What is statistical regression threat?

Answer 21

Regression to the mean, results might be due to the selection of PPTs based on extreme scores, and when measured again, scores might move back to an average level.

One group threat.

Question 22

What is differential selection threat?

Answer 22

Results might be due to differences that existed between PPTs in groups before the treatment occurred); with random assignment hope that people in two groups are similar to each other as much as possible (large sample this is less likely to be threat as more likely to equalize out).

Two group or multi-group threat.

Question 23

What is differential attrition/mortality?

Answer 23

Results might be due to the systematic differential loss of PPTs from each group; intensive intervention group is more likely to stay in the study (lose more people in control group).

Two or multi-group threat.

Question 24

What is error variance?

Answer 24

Variability in the DV that cannot be attributed to IV and instead to extraneous variables.

Measurement error or residual error. Random stuff that affects our measurements.

Time of day, noise in BG, etc.

Question 25

What is sampling bias?

Answer 25

Occurs when sample is not representative of the population.

Question 26

What is random assignment?

Answer 26

A method that uses chance to assign participants into two or more respective groups.

Question 27

What are the types of TRUE experimental designs?

Answer 27

Posttest control group design.

Pretest posttest control group design.

Matched-pairs group design.

Question 28

What is a posttest control group design?

Answer 28

Between-subjects design.

Two groups that are randomly assigned are compared on levels of DV following manipulation of IV.

Post-tests are givens to determine if a difference between groups exists.

Question 29

What are strengths of posttest control group design?

Answer 29

Quick and inexpensive.

Random.

Question 30

What are weaknesses of posttest control group?

Answer 30

No pretest so no baseline per person and no knowledge of past differences between groups.

Question 31

What is a pretest posttest control group design?

Answer 31

Includes pretest to assess differences between groups.

Between subjects design.

Random assignment. Two groups are compared on levels of DV after manipulation of IV.

Can be considered RCT.

Question 32

What is a matched-pairs group design?

Answer 32

Between-subjects design with randomization.

Subjects matched by characteristics to form pairs. One is placed in control and the other in intervention.

Used when characteristic is strongly associated with DV and may not get equally distributed across groups randomly.

Question 33

What are strengths of matched-pairs group design?

Answer 33

Significant reduction in potential confounding effect of matched characteristic.

Allows for control of differential selection on characteristic that would otherwise affect dependent variable.

Question 34

What is weakness of matched-pairs group design?

Answer 34

Time-consuming.

Not possible to truly match all PPTs in one group with a pair in the other.

Question 35

What is a within-subjects or repeated measures design?

Answer 35

Subjects serve as their own controls are are exposed to all levels of IV.

Question 36

What are some quasi-experimental designs?

Answer 36

Within-subjects or repeated measures design NEED TO CHECK.

Non-equivalent control group design.

Interrupted time series design.

Question 37

What is a quasi-experimental design?

Answer 37

Manipulation of IV but no randomization.

Question 38

What are strengths of within-subjects design?

Answer 38

Reduces error variance

Question 39

What are weaknesses of within-subjects design?

Answer 39

Carryover effects such as:

Learning, fatigue, habituation, sensitization

Question 40

What is a cross-over trial?

Answer 40

Specific form of within-subjects design with randomization.

Question 41

What is the benefit of a cross-over trial?

Answer 41

Reduces carryover effects.

Question 42

When is a crossover trial appropriate?

Answer 42

When IV effects are short and reversible.

We know effects are not due to passage of time, learning, or other effects.

Question 43

What is a smart design?

Answer 43

A study design that has to do with optimization of treatments.

Multiple points of randomization.

Can see effects of randomization and see when and for whom interventions are appropriate.

Question 44

What is random assignment?

Answer 44

Units are assigned to conditions purely by chance and each unit has a nonzero probability of being assigned to a particular condition.

Question 45

What is random sampling?

Answer 45

Random selection of study subjects from a population so estimates derived from sample can be taken as an approximation of estimates in the population.

Question 46

What is a randomized experiment?

Answer 46

Also known as a true experiment, a randomized experiment whereby the study design entails randomly assigning study subjects to groups.

Question 47

Describe simple randomization.

Answer 47

Units assigned to conditions by chance.

Question 48

Describe block randomization?

Answer 48

Main point is to balance number of PPTs in each treatment group.

Select size of block.
Assign PPTs in that order.
Repeat.

Question 49

Describe stratified randomization.

Answer 49

Used to achieve balance in groups based on covariates of interest.

Separate blocks for each combination of covariates generate and subjects assigned to the appropriate block.

Divide into subgroups (strata) and perform separate randomization within each.

Need to know participants’ characteristics for strata before randomization.

Question 50

Describe covariate adaptive randomization.

Answer 50

Seeks to ensure balance throughout study.

Minimize PX differences in characteristics across treatment groups.

Often done through minimization.

A new participant is sequentially assigned to a particular treatment group by taking into account the specific covariates and previous assignments of participants.

Question 51

Describe a non-equivalent control group design.

Answer 51

A group is matched as closely as possible to the experimental group and used as a comparison group.

Question 52

Describe an interrupted time series design.

Answer 52

Strongest quasi-experimental approach for evaluating longitudinal effects of interventions.

Collect data at multiple instances over time before and after treatment.

Key time point: when intervention delivered.

Treatment effect is demonstrated if the pattern of post-treatment responses differs significantly form the pattern of pretreatment responses (e.g., change in level or slope).