Nursing Research 63 377 Dr Wally J Bartfay
Nursing Research 63 -377 Dr. Wally J. Bartfay “If you want something done, ask someone who’s busy!” (Wally J. Bartfay, 2003)
Review Quiz l l l (1) _____ are scientific investigations that make observations & collect data according to explicit criteria (2) True or “classic” experiments have 3 essential criteria (list them): (i) (iii)
Types of Experimental Designs l l l (1) True or Classic experiments (pretestposttest control group design) (2) Solomon four-group design (similar to true experiment but has 2 additional groups, an experimental after-group and a control aftergroup) (3) After-only design (also called post-test only control group design, has 2 groups like true experiment but no pretesting occurs)
True or Classic Experiment Randomization Of Subjects Experimental Group Control Group Pretest Experimental Treatment No Experimental Treatment Posttest
Solomon four-group design Randomization Of Subjects Experimental Group Control Group Pretest No pretest Experimental Treatment No experimental Treatment Posttest
After-only (post-test) experimental design Randomization Of Subjects Experimental Group Control Group No Pretest Experimental Treatment No Experimental Treatment Posttest
Non-experimental Research Designs: Relationship/ difference Studies l l l (1) Correlational studies: examines relationship between 2 or more variables (eg. , serum ferritin levels & AMI) (2) Developmental studies (time perspective) (a) Cross-sectional: specific relationships examined at one point in present time (effects of age on attitudes towards smoking in public places) (b) Cohort studies: subject groups are compared based on specific characteristics (e. g. , smokers vs. non-smokers) (c) Longitudinal & prospective studies: data collected on same subjects at different time points in the future (subjects serve as their own controls, e. g. , Framingham Heart Study) (d) Retrospective: attempt to link present events with past events (e. g. , incidence of prostate CA in chimney sweepers in London in the 19 Century)
Methodological Research l l Is development & evaluation of datacollection instruments, scales, or techniques (e. g. , SF-26 for QOL, Caregiver Burden Scale) Psychometrics deals with theory & development behind these (e. g. , constructs/ concepts like anxiety, stress, caring)
Methodological Research: Basic Steps l l (1) Clearly define the construct/ concept or behaviour to be measured (2) Formulate the items for the instrument/ tool to be used (3) Develop clear instructions for users & respondents (age & educ. level appropriate) (4) Test the tool’s/ instruments reliability & validity
Meta-Analysis l l l Not a design per se, but a research method Takes results of several studies in a specific area & synthesizes their findings to draw conclusions regarding the state of knowledge in a defined area & indications for future research (e. g. , all studies examining exposure to asbestos & development of lung CA) Can be used to synthesize both experimental & nonexperimental studies
Secondary Analysis l l Not a research design but a form of research in which the researcher takes previously collected data from one study & reanalyzes data for secondary purpose Original study may be experimental or non-experimental E. g. , original study examined stress & immune response in mothers with LBW infants. Secondary analysis done on data from healthy mothers to describe whether effects of nutrition & physical activity in postpartum women according to 4 wt. categories (1) underweight; (2) normal weight; (3) overweight, and (4) obese
Research Population l l l Is a well-defined set that has certain specified properties It can be composed of people, animals, objects or events Target populations: is the entire set of cases about which the researcher would like to make generations (e. g. , nursing students, 1 st time mothers, pt’s with COPD)
Samples & Elements l l Sample is a set of elements that make up the population Element is the most basic unit about which information is collected (e. g. , the person, place or object)
Sampling l l l Is a process of selecting a portion or subset of the designated population to represent the entire population Its purpose is to increase the efficiency of a research study When done correctly, the researcher can draw inferences & make generations about the target population
Major Sampling Types l l l (1) Representative sample: is one whose key characteristics closely approximate those of the population (e. g. , 70% of population in a study of child -rearing practices were employed full-time, sample should be same %) (2) Nonprobabilty: elements are chosen by nonrandom sampling (findings are less generalizable) (3) Probability: uses some form of random selection when the sample units are chosen
Major Sampling Types l l (4) Quoto sampling: a form of nonprobability sampling in which knowledge about the population of interest is used to build some representativeness into the sample (uses strata) (5) Purposive sampling: researcher selects subjects who are considered to be typical of the population (CA pt’s who have undergone bone marrow transplant due to leukemia)
Sampling Basics Step 1: Identify target population Step 2: Delineate the accessible population Step 3: Develop a sampling plan Step 4: Obtain approval from REB
Confounding variables l l l Occurs when there is an extrinsic factor that is associated with the predictor variable & a cause of the outcome variable E. g. , cigarette smoking is a likely confounder in ETOH-induced CHF Simplest strategy to deal with it is to include specific criteria for inclusion/ exclusion
Inclusion & exclusion criteria: l l Also known as “delimitation” criteria Lists characteristics of subjects that must be present or not present (e. g. , certain age group, gender, specific Dx or Rx etc) Criteria that specify your target population in the study E. g. , only non-smokers included in study looking at effects of ETOH-induced CHF
Specification l l Here the researcher “specifies” what criteria will be included & excluded & helps to control for potential confounding effects E. g. , only non-smokers included in study looking at effects of ETOH-induced CHF
Matching l l l Used often in case-control studies Involves selecting for each case a control with the same value of the confounding variable E. g. , In a study of ETOH as a predictor of CHF, the ETOH drinking case (subject with CHF) who smoked 1 pack cig. QD would be compared with a ETOH drinking case (subject with CHF) who DID NOT smoke 1 pack cig. QD
Advantages to matching: l l (1) Is an effective way to prevent confounding by constitutional factors like age & gender that are often strong determinants of outcome (2) Used to control confounders that can’t be measured & controlled in any other way (e. g. , twins matched to examine effects of environments vs. genetics) (3) Increases precision of comparisons between groups and thus the power of the study (4) May be used for sampling convenience, to narrow down an otherwise impossibly large number of potential controls
Disadvantages of matching: l l l (1) Often requires additional time & expense to identify a match for each subject (2) B/C matching is also a sampling strategy, the decision to match is made at beginning of study & is irreversible (3) Possibility of “over-matching” exists, when the matching variable is not a confounder b/c it isn’t associated with the outcome at all
Coping with confounders in the analysis phase: Stratification l l assures that only cases & controls (or exposed & unexposed subjects) with similar levels of a potential confounding variable are compared Involves segregating the subjects into “strata” (subgroups) according to the level of a potential confounder & then examining the relationship between the predictor & outcome separately in each stratum Principal disadvantage is that only limited number of variables can be controlled for simultaneously E. g. , Confounders for MI include BP, cholesterol profiles, cig. Smoking, BMI, ETOH intake (to stratify on these 5 variables, & if only 3 strata each, would require 35 strata = 243
Coping with confounders in the analysis phase: Adjustment l l Statistical techniques that “model” the nature of the associations among the variables in order to separate out the effect of the confounder Has advantage of multivariate adjustment techniques which can control for the influence of many confounders simultaneosly using “continuous variables” (e. g. , age)
Have a great week… l Sing, laugh and be merry…life as a student is great in comparison to the hostile reality of the work world….
- Slides: 26