Positivist

Positivist

1. Quantitative

   • Experimental
     • Basic Designs (b.w subject designs)
       • Pretest-Posttest Design
         • Establishes baseline scores; E.G. give test to both classes to test general knowledge, then administer test again with added content; If groups unequal @ pretest not good
       • Post-Test Only
         • No baseline scores; 2 groups; not compared right away; 1 group given some content while others given no content; both groups test to see how they're doing but bad b.c no baseline
       • Double Blinded
         • Often blind to researcher and subject a.k.a double blinded; eliminates bias
     • Advanced Designs
       • Factorial Design
         • 2 or more variables manipulated simultaneously E.G. tactile stimulation vs auditory stimulation with factor of time in new borns, and 6 diff of groups of babies within subject
       • Cross-Over Design (w.i subject design)
         • Exposed to more than 1 treatment; Only thing that is being randomized is which treatment first; a.k.a four diff treatments and everyone receives random order; Also called a Repeated Measure Design
           • Repeating constant testing (e.g. see if attitude is changing, so keep being tested over a year numerous times) same subject randomly selected; Act as their own control because researcher can see changes over time
         • Can be inappropriate for certain questions because of possible carryover affects (E.G. participant exposed to 2 diff drug treatments and may be influenced by second drug based on experience with the first)
     • 3 things must occur to be called an experiment
       • Manipulation, Control, Randomization
     • Advantages: most powerful design for testing hypothesis of cause and effect relationship, greater corroboration of IV and DV relationship r.t controlling properties (manipulation, control groups, randomization)
       • Disadvantages: Not all variables are able to be manipulated (e.g. human characteristics such as disease, or health habits; these can’t be randomly conferred on ppl); Variables exist that can technically, but not ethically be manipulated (e.g. want to determine if smokes cause lung cancer so you assign a group to smoking a.k.a force ppl to smoke which is not ethically correct); Hawthorne effect b.c knowledge of being studied causes people to change their behavior thereby obscuring the true effect of research variables; Sometimes a double Hawthorne effect can exist where nurses and patients both are aware of experiment that investigates the effect of new post-op procedures (thus a positive way to combat this is a “double-blind experiment”  Some limitations that make it difficult to apply to real-world problems
   • Quasi-Experimental
     • Similar to an experimental design in that participants receive a treatment/intervention (manipulation of an independent variable); However, lack either the randomization or control-group features of true experiments
     • Time Series Design
       • No control group or randomization (This one-group before–after design); participants act as own controls
       • involves collecting data over an extended time period, and introducing the treatment during that period; extended time perspective strengthens our ability to attribute change to the intervention
       • Suppose that a hospital was adopting a requirement that all its nurses accrue a certain number of continuing education credits before being eligible for a pro- motion. Administrators want to assess the effect of this mandate on turnover rate and number of promotions awarded. Let us assume there is no other hospital that can serve as a reasonable comparison for this study, so the only kind of com- parison that can be made is a before–after contrast. If the requirement were inaugurated in January, one could compare the turnover rate, for example, for the 3-month period before the new rule with the turnover rate for the subsequent 3-month period
     • Non-Equivelant Control Group (Before-After) Design
       • Same as pretest-posttest but NO randomization occurring
         • involves two or more groups of subjects observed before and after the implementation of an inter- vention. As an example, suppose we wanted to study the effect of primary nursing on nursing staff morale in an urban hospital. The new system of nursing care is being implemented throughout the hospital, so randomization of nurses is not possible. Therefore, we decide to collect comparison data from nurses in a similar hospital that is not instituting primary nursing. We gather data on staff morale in both hospitals before implementing the primary nursing system (the pretest) and again after its implementation (the posttest).
         • Weaker because, without randomization, it cannot be assumed that the experimental and compari- son groups are equivalent at the outset. The design is, nevertheless, strong because the collection of pretest data allows us to determine whether the groups had sim- ilar morale initially. If the comparison and experimental groups were similar at the pretest, we could be relatively confident that posttest differences in self- reported morale resulted from the intervention.
         • Term comparison group used instead of control group
     • Advantages: practicality, allows for some control, used when experiment not possible
       • Disadvantages: Cause and effect inferences cannot be made as easily as there are alternative explanations for observed results
   • Non-Experimental
     • Correlational Designs (a.k.a Ex-Post Facto)
       • Used for research problems that cannot be addressed with an experimental design because the independent variable cannot be manipulated for practical and/or ethical reasons, or for questions in which little is known about the phenomenon and where the purpose is descriptive.
         • Many research problems cannot be addressed with an experimental or quasi-experimental design. For example, suppose we were interested in studying the effect of widowhood on physical health. Our independent variable here is widow- hood versus nonwidowhood. Clearly, we cannot manipulate widowhood; people lose their spouses by a process that is neither random nor subject to control. Thus, we would have to proceed by taking the two groups (widows and nonwidows) as they naturally occur and comparing their physical well-being.
       • Research has been conducted after variation in the I.V has occurred; Purpose to study relationships among variables; investigators do not control the independent variable—the presumed causative factor—because it has already occurred. It is risky to draw cause-and-effect con- clusions in such a situation. A famous research dictum is relevant: correlation does not prove causation. That is, the mere existence of a relationship between variables does not warrant the conclusion that one variable caused the other, even if the relationship is strong.
       • Retrospective vs Prospective Slide 42
       • NOT ABOUT CAUSE AND EFFECT
     • Descriptive Designs
       • Survey research; Observe, describe, document phenomenon
     • Advantages: many interesting problems not amenable to experimentation; crucial role in nursing when experimentation is not possible; tends to be high in realism; efficient and effective means of collecting large amount of data
       • Disadvantages: inability to reveal causal relationships conclusively; susceptible to faulty interpretations due to pre-existing differences