Chapter 9- Quantitative Research Design Flashcards

1
Q

Causality

A

Etiology (causation)

Cause of health-related phenomena usually are not deterministic, but rather probabilistic- that is, the cause increase the probability that an effect will occur.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Counterfactual

A

What would happen to teh same people exposed to a causal factor if they simultaneously were not exposed to the causal factor?

an effect is the difference between waht actually did happen with the exposure and what would have happened without it.

A counterfactual clearly can never be realized but it is a good model to keep in mind in desiging a study ot answer cause-probing questions.

A central task for all cause-probing research is to create reasonable approximations to this physically impossible counterfactual.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Temporal causality

A

a cause must precede an effect in time.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Empirical relationship (causality)

A

a relationship between the presumed cause and the presumed effect must exist.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

No cofounders

A

The relationship cannot be explained as being caused by a third variable.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

coherence

A

involves having similar evidence from multiple sources

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

consistency

A

involves having similar levels of statistical relationships in several studies

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

biologic plausibility

A

evidence from lab or basic physiologic studies that a causal pathway is credible.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Experiment- typically called Randomized Controlled Trial- RCT

A

Researchers are active agents- not simply observers

isolating phenomena and controlling the conditions under when they occurred

Considered the GOLD STANDARD for yielding reliale evidence about causes and effects

RCTs offer the most convincing evidence about whether one variable has a causal effect on another

the goal in most RCTs is to have an identical intervention for all people in the treatment group.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

manipulation

A

the reserachers does something to at least some participants- there is some type of intervention

Experimenters manipulate the IV by administering a treatment (or intervention) to some people and withholding from others (C)., or by administering an alternative treatment to two or more groups.

Experimenters deliberately VARY the IV (the presumed causes) and observe the effect on the outcome (O).- which is sometimes referred to as an ENDPOINT in the medical literature

.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

control

A

the researcher introduces controls, including devising a counterfactual approximation- usually, a control group that does not receive the intervention

control group- refers to a group of participants whose performances on an outcome is used to evaluate the performance of the treatment group on the same outcome. Is a proxy for an ideal counterfactual.

may be an alternative interventions, standard methods of care, or a placebo

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

randominzation

A

the researcher assigns participants to a control or experimental condition on a random basis

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

intervention protocols

A

full nature of the intervention must be delineated in formal intervention protocols that spell out exactly what the treatment is.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Tailored interventions or patient-centered interventions (PCIs)

A

purpose is to ebhance treatment efficacy by taking people’s characteristics into account

Each person receives an intervention customized to certain characteristics, such as demographic traits or cognitive factors.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

placebo

A

a pseudointervention presumed to have no therapeutic value

used to control for the nonpharm. effects of drugs, such as extra attention (?)

Placebo effects- changes in he outcome attributable to the placebo condition, because of participants expectation of benefits or harm

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

attention control group

A

used by researchers when they want to rule out the possibility that intervention effects are caused by the special attention given to those receiving the intervention, rather than by the actual treatment itself.

The idea is to separate the “active ingredients” of the treatment from the “inactive

ingredients” of special attention.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

dose- response effects

A

different doses or intensities of treatment wherein all participants get some type of intervention, but the experimental group gets an intervention that is ricker, more intense, or longer.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

wait-list control group

A

with delayed treatment, the control group eventualy receives the full intervention after all outcomes are asseessed

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

comparative effectiveness research (CER)

A

strives to produce evidence that is especially useful for clinical-decision making

(Testing two competing interventions ?)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Randomization- Random assignment- random allocation

A

assigning participants to treatment conditions at random

participants have equal chance of being assigned to any group

if people are placed in groups randomly there is no systematic bias in the groups with respect to preintervention attributes that are potential confounders that could affect outcomes.

There is no guarentee that the groups will be equal.. the smaller teh sample group the less randomization.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Matching

A

assigning participants to control vs experimental groups, however trying to ensure each group has similar partipiant characterisitcs

Complicated to match on more than two or tree confounding simultaneously.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Basic randomization

A

Flipping a coin and assigned at random to one of the two groups. 50-50 chance of being assigned to intervention group.

No restrictions- sometimes called COMPLETE RANDOMIZATION

Large imbalances can occur, especially with small sample sizes.

Not recommended with sample sizes less than 200

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

Simple randomization

A

involves starting with known sample size, then prespecifying the proportion of subjects who will be randomly allocated to different treatment conditions.

Ex; we know 15 participants, so 5 will be assigned to 3 different groups- two intervention, one no intervention.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

Randomization vs. Random sampling

A

Randomization- is a signature of an experimental design. If participants are not randomly allocated to conditions then the design is not a true experiment

Random sampling- a method of selecting people for a study- is not a signature of an experiments. Most RCTs do not involve random sampling.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

allocation concealment

A

prevents those who enroll participants form knowing upcoming assignments, to avoid potential biases.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

baseline data

A

preintervention data on outcomes that should be collected before randomization to rule out any possibility that knowledge of the group assignment might distort baseline measurements.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

quasi- randomization

A

method of allocating participates in a manner that is not strictly randoms

ex: participants may be assigned to groups on an alternative basis (qother person) or based on whether their birthday is an odd or even #.

This is not a true method of randomization.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
28
Q

Statified randomization

A

in which randomization occurs seperately for distinct groups (e.g. males, females)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
29
Q

permuted block randomization

A

in which people are allocated t o groups in small randomly sized blocks to ensuer a balances distribution in each block

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
30
Q

urn randomization

A

in which group balance is continuously monitored and the allocation probability is adjusted when an imbalance occurs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
31
Q

randomized consent

A

in which randomization occurs prior to obtaining informed consent- aka Zelen design

WHY??

32
Q

partial randomization

A

in which only people without a strong treatment preference are randomized- sometimes called PARTIALLY RANDOMIZED PATIENT PREFERENCE (PRPP)

33
Q

cluster randomization

A

which involves randomly assigning clusters (e.g. hospitals) rather than people to different treatment groups

34
Q

Performanc bias

A

systematic differences in care provided to members of different groups of participants, apart from any intervention.

Those delivering an intervention might treat participants in groups differently

35
Q

Detection, or Ascertainment bias

A

concerns systematic differences between groups in how outcome variables are measure, verified, recorded
,
Adressed by blinding those who collect that outcome date, or in some case, those who analyze it

36
Q

Open study

A

blinding is NOT USED

37
Q

Blinding/Masking

A

often used in RCTS to prevent biases stemming from AWARENESS

involves concealing information from participants, data collectors, care providers, intervention agents, or data analysts to enhance objectivity and minimize EXPECTATION BIAS.

can also involved withholding information about study hypotheses or baseline performance on outcomes

38
Q

closed study

A

Blinding used.

39
Q

Single-blind study

A

blinding is used only with one group of people (e.g. study participants)

40
Q

double-blind study

A

two groups blind- those delivering the intervention and those receiving it

important for researchers to identify which groups were blinded

41
Q

R
O
X

A

R- randon assigment
O- outcome measurements
X- exposure to intervention

42
Q

Posttest only design

A

Orrrr AFTER ONLY DESIGN

because data on the outcome are collected only once- after randomization and completion of the intervention

43
Q

Pretest-posttest design

A

BEFORE-AFTER design

mixed design. analyses can examine both differences between groups and changes within groups over time

44
Q

Repeated meausres design

A

Pretest-Posttest design taht includes data collection at multiple postintervention points

45
Q

Factorial Design

A

Experimental design where two or more variables are manipulated simultaneously

Permit us to test not only main effects (effects from the manipulated variable) but also interaction effects (effects from combining treatments).

46
Q

Crossover Design

A

Involves exposing the same people to more than one condition

Within-subjects desisgn has the advantage of ensureing the highest possible equivalence among participants expose to different conditions

Participants must be randomnly assigned to different orderings of treatments.

Inappropriate for certain research questions because of possible CARRY-OVER EFFECT

47
Q

carry-over effect

A

when people are exposed to two different conditions they may be influenced in the second condition by their experience in the first one.

When concerned about carry-over effect, researchers will have a wahout period in between treatments (period of no treatment)

48
Q

Experimental Design- Strengths and Weaknesses

A

Strengths: gold standard for testing interventions because they yield strong evidence about intervention effectiveness. Researchers can feel confidence in which causal relationships can be inferred.
Controls imposed by manipulation, comparison and randomization, alternative explanations can be discredited.

Limitations: constraints that may make experimental designs impractical or impossible
sometimes criticized for artificiality- which partially stems from requirements for comparable treatments within randomized groups with strict adherence to protocols.

49
Q

Hawthorne effect

A

caused by people’s expectations.

Knowledge of being in a study (not just knowledge of being in a particular group) appears to have affected people’s behavior, obscuring the effect of the intervention.

50
Q

Quasi-experiments

A

Controlled trials without randomization

involve an intervention but they lack randomization

Some even lack control groups

A signature of quasi-experimental design is an intervention in the absence of randomization.

51
Q

Nonequivalent control group pretesst-posttest design

A

sometimes called controlled before-after design

involves two groups of participants, for whom outcomes are measure before and after intervention

52
Q

comparison group

A

in quasi-experimental design, in leui of control group.

lieu

53
Q

Nonequivalent control group posttest only design

A

no information about initial equivalence of participants.

much weaker quasi-experimental design

54
Q

historical comparison group

A

comparison data are gathered from other people before implementing the intervention.

Even when the people are from the same institutional setting, however, it is risky to assume the two groups are comparable, or that the environments are comparable except for the new intervention.

Still is possibility that something other than the intervention could account for observed differences in outcomes

55
Q

Time Series Design

A

Before-After comparison.

one-group pretest-posttest design is straightforward but has several weaknesses

Many factors that this design cannot control

Time series design (interrupted time series design) data are collected over an extended period during which an intrevention is introduced. This is to modifiy design so that some alternative explanations for outcome could be ruled out.

The extended time period strengths ability to attribute change to the intervention

O1, O1, O3, O4, X, O5, O6, O7, O8

many data points- 100 or more_ are recommended for a traditional analysis, and analysis tends to be complex

56
Q

Statistical process control (SPC)

A

used by nurse researchers to collect data sqeuntially over a period of time before and after implementing a practice change.

Time series design with SPC analyses are important for QUALITY IMPROVEMENT PROJECTS

57
Q

dose-response design

A

outcomes of those receiving different doses of an intervention (not as a result of randomization) are compared

58
Q

Quasi-Experimental Strengths and Weaknesses

A

Strengths: practical when an experimental design is not possible. strong quasi-exp. designs introduce some research control when full experimetnal rigor is not posisble
Do not involve random assignment, are liekly to be acceptable to a broader group of people.

Weaknesses: there is usually a RIVAL HYPOTHESIS competing with the interventions as explanations for the results (issue that relates to internal validity)

59
Q

Nonexperimental/Observational Research

A

When researchers do not intervente by manipulating the independent variable , the study is nonexperimental/observational

Most nursing studys are nonexperimental because most human characteristics cannot be manipulated.

Also, many variables cannot be manipulated ethically.

Explore causes-and-effect relationships when an experimental design is not possible

60
Q

Correlational designs

A

researchers study the effect of a potential cause that they cannot manipulate, a correlational design used to examine relationship between the variables.

correlation- relationship or association between two variables, a tendency for variation in one variable to be rel to variation in another.

correlation does not prove causation

61
Q

Retrospective Design

A

studies in which a phenomenon in the present is linked to a phenomena that occured in teh psat.

The researcher begisn with the dependent variable (the effect) and then examines whether it is correlated with one or more previously occuring independent variables (potential causes)

62
Q

case-control design

A

researchers begin with a group of people (lung cancer) and then another group that do not (control). Then look for differences between the two groups in antecedent circumstances or behaviors, such as smoking.

While designing a case-control study, researchers try to identify controls without the disease or condition who are as similar as possible to the cases on key confounding variables (age, gender).

Sometimes use matching or other techniques to control for confounding variables

63
Q

cross-sectional design

A

data on both the dependent and independent variables are collected at a single point in time

Many retrosepctive studies are cross sectional

64
Q

Prospective design- AKA Cohort Design

A

Correlational study where researchers start with a presumed cause and then go forward in time to the presumed effect.

Best design for Prognosis questions and for Etiology questions when randomization is impossible

More costly than retrospective studies, in part because prospective studies require at least two rounds of data collection

a Good prospective/cohort study the researchers will take steps to confirm that all participants are free from the effect (the disease) at the time the independent variable is measured, which may be difficult/expensive.

Considered stronger than retrospective studies

To note: prospective study are not necessarily longitudinal.

prospective means that information about a possible cause is obtained prior to information about an effect.

RCTs are inherently prospective- introduce the intervention and then determine it’s effect.

65
Q

Inception Cohort Design

A

Involves the study of a group assembled at acommon time earlier in a health disorder or exposure to a putative cause of an outcome (e.g. immediately after a TBI) and then followed thereafter to assess the outcomes

66
Q

exploratory prospective study

A

researcher measure a wide range of possible “causes” at one point in time. then examine an outcome of interest at a later point.

Can be more convincing then a retrospective study if it can be determined that the outcome was not present initially because time sequence is clear.

67
Q

natural experiments

A

a group exposed to a phenomenon with a potential health consequence is compared with a nonexposed group.

the researchers did not intervene, but natural experiements if people are affected essentially at random.

68
Q

path analysis

A

aka similar causal model technique?

procedures that allow researchers to test whether nonexperimental data conform sufficiently to the underlying model to justify causal interences.

path analytic studies can be done within the context of both cross-sectional and longitudinal designs, the latter providing a stronger basis for causal inferences becaues of the ability to verify time sequences.

69
Q

Descriptive Research

A

purpose of studies is to observe, describe, and document a situation as it naturally occurs.

Sometimes the starting point for hypotheses generation or theory development

70
Q

Descriptiive Correlational Studies

A

Aim is to describe relationships among variables rather than to support inferences of causality

71
Q

Univariate descriptive studies

A

are not focused on a single variable.

May involve multiple variables, but the primary purpose is to describe the status of each, not to study correlations among them

72
Q

Prevalence studies

A

From the field of epidemiology

are done to estimate the prevalence rate of some conditions at a particular point in time

rely on cross-sectional designs in which data are obtained from the population at risk of the condition

researchers take a “snapshot” of the population at risk to determine the extent to which the condition is present.

Prevalence rate (PR)

# of cases with condition/disease at a given point in time/ # in the population at risk of being a case x K
(K is the number of people for whom we want to have the rate established (per 100 or 1000 population)
73
Q

Incidence studies

A

of new cases with the condition/disease over a given period of time / # in population at risk of being a case (free of the condition at onset) x K

estimate the frequency of new cases

longitudinal designs are needed to estimate incidence because the researchers must first establish who is at risk of becoming a new caes- that is, who is free of the condition at the outset

74
Q

relative risk

A

Prevalence and incidence rates can be calculated for subgroups of the population (eg. men vs women)

estimated risk of “caseness” in one group compared to another

relative risk is computed by dividing the rate for one group by the rate for another.

Relative risk is an important index in assessing the contributions of risk factors to a disease or condition

75
Q

Correlational Research: strengths and weaknessess

A

weak in their ability to support causal influences

in correltational studies, researcher work with preexisiting groups that were not formed at rando, but rather through self selection.

Researcher cannot assume that groups being compared were similar befoer the occurence of the hypothesized caused- IE the IV