Research Methods I

target population

largest group, to whom you intend to apply result of the study

accessible population

* group must be accessible to you
* collect your sample from this group
* representative of target pop

sample

group that participate in study

probability sampling

allows everyone from accessible population to have equal chance of being selected

examples of probability sampling (4)

* simple random sample
* systemic sample
* stratified random sample
* cluster sample

non probability samples

do not give everyone from the accessible population an equal likelihood of being selected

which type of sampling used more common in health care

nonprobability

which type of sampling reduce error and bias

probability

examples of non probability (4)

* convenience sample
* quota
* purposive
* snowball

convenience sample

Participants chosen based on availability to be included

* consecutive sample involves recruiting everyone
* volunteers may be recruited by announcing the study and requesting those interested to participate

quota sample

non-random stratified sampling

* quota established from different strata to match the sample to the population

purposive sample

researchers purposely select participants based on characteristics desired for the sample

snowball sample

used for rare or sensitive topics

* after initial group is selected to participate, they are asked to identify others who meet criteria, who are then invited to participate

simple random sample

all indiv are included in the pool as a random selection occurs

systematic sample

when there is non ordered list available, every nth indiv is selected

stratified random sample

to improve representativeness (decrease sampling error), strata are created that relate to the accessible population

* random sample within each strata

cluster sample

when accessible population is large, it is grouped into cluster

* random selection of cluster is made and then population of the cluster, or a random selection of those within the cluster is sampled

the flaw in bias is ….

systematic

misclassificaion bias

participants are not correctly classified as having or not having a quality, based on poor operational definition

recall bias

participants recall (or do not) historical facts based on wanting to identify a relationship (or not)

reporting bias

participants in one group systematically underreport/overreport an exposure or outcome

sampling bias

the sample is no representative of the accessible population for a systematic reason (not by chance)

allocation bias

in nonrandom samples, when individuals with certain qualities are represented differently in groups

examples of primary research

* research conducted with patients
* observational research
* experimental research

observational research

* describes group of indiv on different variables
* may be based on an individual, a single group, or more than one person
* some separate descriptive research out as a sep category, while other include it as observational

experimental research

experimental designs used to compare two or more different conditions/interventions to assess efficacy

examples of secondary research

* research is conducted on the results of primary research
* researchers not working with patients
* synthesis research

synthesis research

combines results of various other studies to provide stronger evidence

examples of observational study design (4)

* case studies
* case report
* case series
* cross-section studies
* case-control studies
* cohort

examples of experimental study design (1)

randomized control trials

what does pyramid of evidence tell us

1. how common
2. validity

what type of evidence best in treating patients

systematic review and meta-analysis

observational studies

* primary study design
* used commonly in epidemiology
* helpful discover new info
* researcher doesnt have large amount of control
* doesnt determine cause and effect

which study mostly used in epidemiology

observation

case studies (2 types)

\*lowest on evidence pyramid

* case report (indiv)
* report of single patient with unique phenomenon
* case series (group)
* report of a group of patients all with same or similar unique phenomenon

what is the least valid approach to determining cause because of competing variables and lack of control

case studies

why are case studies useful

1. understand new and unusual presentations of conditions we already know about
2. identify new treatments for both existing and new conditions
3. discover new conditions that were previously unknown
4. begin to develop hypotheses for testing with other study designs

case studies objective

describe individuals with disease

case studies population

all included indiv with same disease

when is case study used

patient or group is available with no comparable group

caution with using case study

lack of generalizability

statistical measures of case study

descriptive statistics

cross-sectional studies

observational study that measures the association of an exposure to the presence or absence of a disease at a particular point in time

* capture moment in time, nothing historical or future

surveys are examples of

cross-sectional studies

objective of cross-sectional studies

describe exposure or disease status of a population at a moment in time

population of cross sectional studies

study participants must be representative of source population

when is cross section study used

time or budget limited

what to be careful of with cross sectional study

sample is not representative of pop

statistical measure of cross-sectional study

prevalence, linear regression (r)

case-control study

observational study when we compare

* cases- indiv with the phenomenon of interest, and
* controls- indiv without phenomenon of interest
* gives retrospective
* Investigate previous exposure or risk factors

\*study design is retrospective

case matching

individual (case) and examine controls and try to match in as many dif ways as possible

Ex: patient is F, native american, 60s, diabetes. what is risk factor smoking and COPD, look for control similar age, ethnicity

objective of case control studies

comparing exposure histories of people with and people without a disease

population of case control studies

cases and controls must be similar, aside from disease state

when is case control study used

disease is uncommon, long induction period, or little is known but a source of cases is available

what to watch out for with case control studies

recall bias

statistical measure of case control studies

odds ratio

cohort studies

* can be prospective or retrospective
* observational

prospective study vs retrospective

begins after exposure /risk/intervention and follows groups forward in time

\
looks for groups with conditions, and then looks even further back to establish a baseline

difference btw cohort and case control

cohort studies groups people by risk/intervention and look for disease, where case control group by disease and look for previous risk/exposure

objective of cohort

compare rate of disease overtime in people with or without particular exposure

population of cohort study

participants must be similar except for exposure status, and not have the disease at the start

when is cohort study used

exposure is somewhat common and there is group available

what do you watch out for with cohort study

loss to follow up (prospective) and missing records (retrospective)

statistical measure of cohort

relative risk ratio

randomized control trials

experimental designs require an intervention, control, and assignment to group

two key elements of randomized control trials (RTC)

key elements are randomization and control

T/F control group is always placebo

F

randomization

participants assigned to groups by chance- the researchers do not select who will be assigned to what group

T/F blinding and randomization are the same

F- there can be randomization without blinding, and there can be blinding without randomization

what allows for cause and effect relationship

randomization

* equal likelihood confounding variables spread across groups

key concepts in RTC (4)

1. rigorous selection of subjects
2. specific protocol and measurement
3. specific measurements
4. blinding

objective of RTC study

compare outcomes btw those assigned to an intervention and those assigned to a control

population of RTC study

similar participants assigned to each group

when is RTC study used

to assess for cause/effect

what to watch out for with RTC

non compliance

statistical measure of RTC

efficacy, risk

what does CONSORT diagram help determine in RTC study

how patients move through a study

ex: how many pt lost in study?

systematic review

secondary research

* study question and inclusion/exclusion criteria defined
* extracted primary study results are raw data for this primary research
* extensive search of literature

objective of systematic review

synthesize existing knowledge

population of systematic review

published literature

when is systematic review used

to compare findings of previously conducted studies on a specific topic

what to watch out for with systematic review

publication bias

statistical measure of systematic review

none, but may report measures from included studies

meta analysis

type of systematic review

uses only studies with quantitative results

summary statistic from each included study is included that is then analyzed with other statistics from primary studies

objective of meta anlysis

synthesize existing knowledge

population of meta analysis

published literature

when is meta analysis used

to compare findings of previously conducted studies on a specific topic using pooled statistics

what to watch out for with meta analysis

publication bias, studies that cannot be fairly compared

statistical measures of meta analysis

summary measure

experimenter effect (internal validity threat)

researcher could treat participants differently if they knew what group the participants were in therefor affecting the participants behavior

observer bias (internal validity threat)

when the researcher knows the hypothesis and variables of the study and has a biased view because they know what they are looking for

researcher attribute (internal validity threat)

how the characteristics of the researcher can affect the participants (example how they look, gender)

hawthorne effect (internal validity threat)

participants responses change because they know they are being observed, similar to the social desirability effect

testing effect (internal validity threat)

taking a pretest before the experiment can influence the participants views which confounds the experimental results internal validity that the conclusions drawn from experimental results accurately reflect the experiment

maturation (internal validity threat)

people and their surroundings are continually changing, and such changes can effect the experiment

experimental mortality (internal validity threat)

people drop out of experiment before it is complete (like if there was a smoking psa experiment, the kids who already smoke might just leave)

selection bias (internal validity threat)

comparisons between two groups of participants means nothing unless these groups are essentially the same at the beginning of the study (random sampling helps us ensure they will be)

intersubject bias (internal validity threat)

when participants from the control group and the experimental group have accesses to one another and can share information

compensatory rivalry

those in the control group may try to compensate for lack of stimulus and work harder that normal

demoralization

feelings of being denied in the control group may result in them just giving up