Exam 2 Flashcards
blueprint or detailed plan for conducting of study
research design
provide the basis for research design
Purpose
Review of Literature
Framework
4 types of quantitative research designs
Descriptive
Correlational
Quasi-experimental
Experimental
involves examining a group of subjects simultaneously in various stages of development, levels of education, severity of illness, or stages of recovery to describe change in a phenomenon across stage
cross-sectional design
involves collecting data from the same subjects at different points in time and might also be referred to as repeated measures
longitudinal design
- There is a cause-and-effect relationship between the variables
- The simplest view is one independent variable causing a change in one dependent variable
Causality
x (independent variable) causes y (change in dependent variable)
- There is a cause-and-effect relationship between interrelating variables
- There are multiple independent variables causing a change in the dependent variable
Multicausality
Length of hospital stay is related to patient dx, age, pre surgical condition, complications postop
- Probability address relative rather than absolute causality
- Variations in variables occur
- Researcher recognize that a particular cause will probably result in a specific effect
Probability
- The slanting of findings away from the truth
- Distorts the findings
- Research designs should be developed to reduce the likelihood of this or to control for it
Bias
Example: some of the subjects for the study might be taken from a unit of the hospital in which the patients are participating in another study involving high quality nursing care or one nurse, selecting patients for the study, might assign the patients who are most interest in the study to the experimental group (p. 195)
Potential causes of Bias in designs
- Researchers
- Components of the environment and/or setting
- Individual subjects and/or sample
- How groups were formed
- Measurement tools
- Data collection process
- Data and duration of study (maturation)
- Statistical tests and analysis interpretation
the timing of data collection is described as
retrospective or prospective
retrospective
looking backward
prospective
looking forward
interventional/experimental research must be
prospective
- Implemented throughout the design
- Improved accuracy of findings
- Greatest in experimental research
Control
Ex.:
- random selection and assignment
- control the duration of the education program
- control the methods of teaching and teachers
- limit the characteristics of subject (e.g., diagnosis, age, type of surgery, incidence of complication)
- Implementation of a treatment or intervention
- The independent variable is controlled
- Must be careful to avoid introduction of bias into the study
- Usually done only in quasi-experimental and experimental designs
Manipulation
examines the effect of a particular intervention on selected outcomes
causality
distortion of study findings that are slanted or deviated from the true to expected
bias
the power to direct or manipulate factors to achieve a desired outcome
- this is greater in experimental that quasi-experimental designs
control
the recognition that several interrelating variables can be involved in causing a particular outcome
- the presence of multiple causes fro an effect
multicausality
address relative rather than absolute causality
probability
a form of conrtrol vernally use in quasi-experimental and experimental studies
manipulation
validity is focuses on determining if study findings are accurate
(IV -> DV)?
internal validity
threats to internal validity
- participant selection
- participant attrition
- history
- maturation
validity is concerned with the extent to which study findings can be generalized beyond the sample used in the study
external validity
threats to external validity
- participant selection (people)
- setting (place)
- history (time)
Over 40% of the potential study participants approached declined to participate because they did not want to follow a structured low-calorie diet. This interaction of the selection of participants and the study intervention is an example of what type of threat to design validity?
external validity (attrition)
The study participants were allowed to select either the exercise (intervention) group or the no exercise (comparison) group in a study examining the effects of exercise on weight, BMI, and blood pressure values, which is a threat to what type of design validity?
internal validity (participant selection and assignment to group concerns)
- most commonly used design
- examines characteristics of a single sample
- identifies phenomenon, variables, and conceptual and operational definitions and describes definitions
typical descriptive design
clarification -> measurement -> description -> interpretation
typical descriptive design
Examines differences in variables in two or more groups that occur naturally in a setting
- Results obtained from these analyses are frequently not generalizable to a population
comparative descriptive design
describe relationships between/among variables
descriptive correlational design
predict relationships between/among variables
predictive correlational design
test theoretically proposed relationships
model testing design
essential elements of experiments
- Random assignment of subjects to groups
- Precisely defined independent variable or intervention
- Researcher-controlled manipulation of independent variable
- Researcher control of experimental situation and setting
- Inclusion of a control/comparison group in the study
- Clearly identified sampling criteria
- Carefully measured dependent or outcome variables
- groups in comparative descriptive studies
- control group
- comparison group
- equivalent vs nonequivalent groups
study groups
- Untreated control group design with pretest and posttest
- Nonequivalent dependent variables design
- Removed-treatment design with pretest and posttest
Quasi-experimental design
quasi-experimental posttest-only design with a comparison group
Experimental group -> intervention (manipulation of independent variable) -> posttest (measurement of dependent variable)
Nonequivalent comparison group -> posttest
Intervention - often ex post facto (may not be well defined)
Experimental group - those who receive the intervention and the posttest
Comparison group - not randomly selected - tend to be those who naturally in the situation do not receive the intervention
Approach to analysis:
- comparison of posttest scores of experimental and comparison group
- comparison of posttest scores with norms
Uncontrolled threats to validity:
- no link between intervention and change
- no pretest
- selection
quasi-experimental pretest and posttest design with a comparison group
Experimental group -> pretest (measurement of dependent variables) -> intervention (manipulation of independent variable) -> posttest (measurement of dependent variable)
Nonequivalent comparison group -> pretest -> posttest
Intervention - experimental group (comparison group not treated or receives standard or routine care)
Comparison group - not randomly selected
Approach to analysis:
- examine difference between comparison and experimental pretest scores
- examine difference between pretest and posttest
- examine difference between comparison and experimental posttest scores
Uncontrolled threats to validity:
- selection-maturation
- instrumentation
- differential statistical regression
- interaction of selection and history
experimental pretest-posttest control group design
Randomized experimental group -> pretest (measurement of dependent variable) -> intervention (manipulation of independent variable) -> posttest (measurement of dependent variable)
Randomized control group -> pretest -> posttest
Intervention: under control of researcher
Approach to analysis:
- comparison of pretest and posttest scores
- comparison of control and experimental groups
- comparison of pretest/posttest differences between samples
Uncontrolled threats to validity:
- testing
- instrumentation
- mortality
- restricted generalizability as control increases
- the design uses large number of subjects to test a treatment’s effect and compare results with a control group who did not receive the treatment
- the subjects come from a reference population
- randomization of subjects is essential
- usually, multiple geographic locations are used
randomized controlled trial (RCT)
elements of a strong design
- Controlling environment: selection of study setting
- Controlling equivalence of subjects and groups
- Controlling treatment (Tx)
- Controlling measurement
- Controlling extraneous variables
which type of study is considered strongest for testing effectiveness of an intervention?
randomized controlled trial (RCT)
A sample of 100 first-time mothers was studies to examine the relationships among the variables of hours of sleep, stress level, anxiety level, and depression 1 month after the birth of their infants.
What is the most appropriate research design or study?
correlational design
A study was conducted to examine the effect of vitamins on weight gain in a sample of infants diagnosed with failure to thrive at 2 months after their birth. The sample of 60 infants was obtained from three pediatricians’ offices; 30 infants were randomly assigned to the intervention group and 30 to the comparison group. The infants were weighed before and after the implementation of the vitamin intervention, which lasted for 6 months. The intervention was implemented using a structured protocol.
What is the most appropriate research design or study?
Quasi-experimental pretest-posttest design with comparison group
A study was conducted to examine the effectiveness of a new drug to treat hypertension in adults. The study had a large sample of convenience and included all patients with hypertension in five primary-care clinics in two different cities in Texas. The participants were randomly assigned to either the intervention group or comparison group. The drug intervention was highly controlled to ensure accurate delivery of the medication and blood pressures (BPs) were precisely at the start of the study and 3 months later.
What is the most appropriate research design or study?
Randomized controlled trials
defines the selected group of people or elements from which data are collected for a study
a sample
sample
selected with sampling method
concepts related to sampling
- Selecting a group of people, events, behaviors, or other elements with which to conduct a study
- Sampling
- Sampling plan
- Members of the sample can be called the subjects or participants
selection of a subset of a population to represent the whole population
sampling
sampling method; defines the selection process
sampling plan
a particular group of individuals or elements who are the focus of the research
population
the portion of the target population to which the researcher has reasonable access
accessible population
an entire set of individuals or elements who meet the sampling criteria
target population
individual units of the population and sample
elements
- Extending the finding from the sample under study to the larger population
- The extent is influenced by the quality of the study and the consistency of the study’s findings
generalization
characteristics that the subject or element must possess to be part of the target population
inclusion criteria
Examples: In a study of patients who have dementia, a researcher wishes to examine the effects of moderate exercise on patients’ abilities to perform self-care. The researcher decides to use subjects between 70 and 80 years of age who have been diagnosed with dementia for less than 1 year.
characteristics that can cause a person or element to be excluded from the target population
exclusion criteria
two ways to define sampling criteria
homogeneous and heterogeneous samples
researchers may narrowly define the sampling -> researchers may have difficulty obtaining an adequately sized sample from the accessible population, which can limit the generalization of findings
homogeneous sample
a sample in which subjects have a broad range of values being studied, which increases the representativeness of the sample and the ability to generalize from the accessible population to the target population
heterogeneous sample
a researcher uses a sample whose members have characteristics similar to those of the population from which it is drawn
representative sample
the sample, access population, and target population are alike in as many ways as possible
representativeness
difference between the population mean and the mean of the sample
sampling error
the expected difference in values that occurs when different subjects from the same sample are examined
random variation
- difference is random because some values will be higher and other lower than the average population values
consequence of selecting subjects who measurement values differ in some specific way from those of the population
systematic variation (bias)
- these values do not vary randomly around the population mean
percentage of subjects who declined to participate in the study
refusal rate
- 80 subjects approached and 4 refused
- 4/80 = 0.05 = 5% refusal rate
percentage of subjects who consented to be in the study
acceptance rate
- 80 subjects approached and 76 accepted
- 76/80 = 0.95 = 95% acceptance rate
withdrawal/drop or loss of subjects from a study
sample attrition
number of subjects withdrawing / number of study subjects x 100
attrition rate
- researcher begins with 250 subjects and 50 drop out before the study is concluded
50/250x100 = 20%
number of subjects who remain in and complete a study
sample retention
a listing of every member of the population, using the sampling criteria to define membership in the population
sampling frame
subjects are selected from the sampling frame
outline strategies used to obtain a sample for a plan
sampling plan
probability vs nonprobability
- simple random sampling
- stratified random sampling
- cluster sampling
- systematic sampling
equal chance
probability (random) sampling
- purposive sampling = judgmental or selective
- network or snowball sampling
- theoretical sampling
- convenience sampling
- quota sampling
this sampling method is least likely to produce findings that are generalizable to a large population
non-probability (non-random) sampling
the size of difference between the groups or the strength of the relationship between two variables
effect size
small effect size
- 0.3
medium effect size
0.3 - 0.5
large effect size
> 0.5
sample size in quantitative studies
- Effect size
- Type of quantitative study conducted (descriptive study tends to require a larger sample size that the others)
- Number of variables
- Measurement sensitivity
- Data analysis techniques
occurs when additional sampling provides no new information, only redundancy of previous collected data
saturation of study data
qualitative
sample size in qualitative research
- Saturation of study data
- Scope of the study
- Nature of the topic
- Quality of the data
- Study design
direct measures
concrete things, such as O2 sat, temp, BP, weight, demographic variables
indirect measures = indicator of concepts
abstract concepts such as pain, depression, coping, self-care, self-esteem, anxiety, feelings
multiple measurement of the concept of pain
Physiologic - pulse, BP, respiratory
- concept of pain
Observation (rubbing and guarding the area, facial grimacing, crying)
Pain scale (faces)
levels of measurement
- Nominal
- Ordinal
- Interval
- Ratio
nominal or categorical level of measurement
e.g., nationality
qualitative
named variables
ordinal level of measurement (e.g., level of conflict)
quantitative
named + ordered variables
interval level of measurement (e.g., temp)
quantitative
named + ordered + proportionate interval between variables
ratio level of measurement (e.g., group size)
quantitative
named + ordered + proportional interval between variables + can accommodate absolute zero
difference between the true measure and what is actually measure
measurement error
measurement errors with direct measures
- a weight scale may be inaccurate for 0.5 lb
- uncalibrated BP equipment
- a tape measure may not be held at exactly the same tension in measuring the waist of each patient
measurement errors with indirect measures
element of being measured cannot be seen directly
the variation in measurement is in the same direction
systematic error
Examples:
A paper and pencil rating scale designed to measure hope may actually also be measuring perceived support.
When measuring subjects’ weight, a scale that shows weights that are 2 pounds over the true weights
the difference is without pattern
random error
Examples:
The person completing a paper and pencil scale may accidentally mark the wrong column
The person entering the data into a computer may punch the wrong key
concerned with how consistently the measurement technique measures the concept of interest
reliability
reliability expressed as a correlation coefficient (r)
- 00 is perfect reliability
0. 00 is no reliability
the lowest acceptable coefficient for a well-developed measurement tool is
0.80
concerned with the consistency of repeated measures or test-retest reliability
stability
focused on comparing two versions of the same instrument (alternate forms reliability) or two observes (interrupter reliability) measuring the same event
equivalence
parallel-forms reliability or inter-rater reliability
addresses the correlation of various items within the instrument or internal consistency
- determined by split-half reliability or Cronbach’s alpha coefficient
homogeneity
internal consistency reliability
a determination of how well the instrument reflects the abstract concept being examined
validity
Example: The CES-D was developed to measure the depression of patients in mental health setting. Will the same scale be valid as a measure of the depression of cancer patients? Researcher determine this by pilot-testing the scale to examine the validity of the instrument in a new population.
comparable to validity in that it addresses the extent to which the instrument measure what it is supposed to in a study
accuracy
comparable to reliability
- the degree of consistency or reproducibility of measurements made with physiological instruments
precision
sources of error in physiological measures can be grouped into 5 categories
- Environment: temperature
- User: person using the equipment
- Subject: capacity
- Equipment: calibration
- Interpretation: misinterpretation
sensitivity = probability of disease = a/(a+c) = true
positive rate
specificity = probability of no disease = d (b+d) = true
negative rate
the process of acquiring subjects and collecting study data
data collection
the steps of data collection are
specific to each study and depend on the research design and measurement techniques
during the data collection process, researchers
- train the data collectors
- recruit study participants
- implement the study intervention
- collect data in a consistent way
- protect the integrity of the study
determines the method of selecting participants
the study design
why is recruiting the number of subjects planned critical
because data analysis and interpretation depend on having an adequate sample size
statistical analysis focuses on
- Identification of all of the statistical tests mentioned in the report
- Determining whether those statistical tests were the correct ones to use
- Deciding whether the researchers interpreted the tests properly
- Projecting the true usefulness of the findings in the real world
stages in data analysis
- Prepare data for analysis
- Describe the sample
- Test reliability of measurement methods
- Conduct exploratory analysis
- Conduct confirmatory analysis guided by hypotheses, questions, or objectives
theory addressing statistical analysis from the perspective of the extent of a relationship or the probability of accurately predicting an event
probability theory
if probability is 0.23, then p = 0.23
- there is a 23% chance that it will occur
- probability is usually expected to be p < 0.05 or p < 0.01
- probability theory is mathematics-based (it addresses the likeliness of an occurrence)
theory based on assumptions associated with the theoretical normal curve
(used in testing for differences between groups, with the expectation that all the groups are members of the same population)
decision theory
- the expectation is expressed as a null hypothesis, and the level of significance (alpha) is often set at 0.05 before data collection
- If something happens less than 5% of the time, at most, then p
a theoretical frequency distribution of all possible values in a population
- A normal curve is a theoretical frequency distribution of all possible values in a population.
In a normal distribution curve, the mode, median, and mean are equal. - It is theoretical in that no naturally occurring population perfectly fits the curve. However, for a data set of 30 or more values, most distributions approximate the normal curve—a few small values, many values in the middle, and a few large values.
- Levels of significance and probability are based on the logic of the normal curve.
- In research, the probability that any data score will be within a certain range of a mean value is calculated based on the theory of the normal curve.
- For a normal curve, about 68% of variable values are within one standard deviation of the mean, and about 95% of the values are within two standard deviations of the mean.
two-tailed test of significance =
the analysis of nondirectional hypothesis
one-tailed test of significance =
the analysis of directional hypothesis
Tailedness
- One-tailed statistics are uniformly more powerful than two-tailed test
- Two-tailed test of significance = the analysis of nondirectional hypothesis
- One-tailed test of significance = the analysis of directional hypothesis
- When a researcher tests a hypothesis that values will exceed a certain value, the researcher is interested only in one side of the curve—the “more than” side. The hypothesis is directional, and this type of hypothesis is called “one-tailed.”
- If the researcher tests a hypothesis that values will “differ from” the usual values (either larger or smaller), the hypothesis is nondirectional, and it is called “two-tailed.”
H0 =
there is no difference between intervention and control group/among variables
type I error (a)
occurs when the null hypothesis is rejected when it is true (false positive)
(e.g., when the results indicate that there is a significant difference, when really there is not)
type II error (ß)
occurs when the null hypothesis is regarded as true but is in fact false (false negative)
(e.g., the results indicate there is no significant difference, when in reality, there is a difference)
the risk of a type II error can be determined using
power analysis (1-ß)
4 parameters of a power analysis
- the level of significance (a=0.05)
- sample size
- power (minimum acceptable power is = 0.80/80%)
- effect size (>0.5 = large)
probability that a statistical test will detect a significant difference if one exists
(the probability of correctly rejecting H0 = 1-ß)
Power
- At the level of power desired by the researcher (usually 0.80), and for a set level of significance, ideal sample size can be determined, if effect size is known from the previous research studies.
- Effect size = size of the effect
describe or summarize the sample and variables (summarize the data)
descriptive statistics
aka summary statistics
- in any study in which the data are numerical, data analysis begins with descriptive statistics
- in simple descriptive studies, analysis may be limited to descriptive statistics
- their purpose is to determine predominant and average values, as well as sameness and differentness
descriptive statistics consist of
- frequency distributions
- percentages
- measures of central tendency
- measures of dispersion
- standardized scores
their purpose is to describe the most typical values of a data set and the amount of dispersion values
frequency distributions
- ungrouped frequency distributions
- groups frequency distributions
- percentage distributions
measures of central tendency
- mean
- median
- mode
measures of dispersion
- range
- variance
- standard deviation
- confidence interval
- standardized scores
- scatterplots
3 types of descriptive statistics
- Frequency statistics
- Measures of central tendency
- Measures of dispersion
show the relative frequency of various values or groups of values within a data set
frequency distribution
show the frequency of all individual values in the data set
ungrouped frequency distributions
Example: data are presented in raw, uncounted form
1: /
2: /////
3: ///
4: /
5: //
display the frequency of ranges of values
grouped frequency distributions
Example: data are pre-grouped into categories
- Ages 20 to 39: 14
- Ages 40 to 59: 43
- Ages 60 to 79: 26
- Ages 80 to 100: 4
indicate the percentage of values in each group
percentage distribution (frequency distribution)
Example: Salaries: 41.7% Maintenance: 8.3% Equipment: 16.7% Fixed costs: 8.3% Supplies: 25%
the numerical value or score that occurs with greatest frequency
mode
used for nominal data (the most commonly occurring value)
- there may be more than one mode
- outliers do not affect the mode
the midpoint or the score at the exact center of the ungrouped frequency distribution (the 50th percentile)
median
used for ordinal data (outliers do not affect the median value)
the sum of the scores divided by the number of scores being summed
mean
used only for ratio/interval data (the average of all values)
- outliers do affect the mean value
provide the average, middle, or most common value
measures of central tendency
in a perfectly normal distribution, the mean is also the median and is also the mode
indicate the amount of differentness there is min a data set
measures of dispersion
the spread between the highest value and the lowest
range (measure of dispersion)
- Is obtained by subtracting the lowest score from the highest score
- Uses only the two extreme scores
- very crude measure and sensitive to outliers
Point spread is an example of range - it may be expressed as “from… to” or as the numerical difference
indicates the spread or dispersion of the scores
variance
measure of dispersion
indicates the average difference between each data point and the mean of a data set
standard deviation (SD)
- the square root of the variance
- mean = average value
- SD = average difference score
(measure of dispersion)
expresses raw scores as deviations from the mean, so that scores across different instruments can be compared (e.g., GPA vs SAT)
standardized scores (measure of dispersion)
- Raw scores that cannot be compared and are transformed into standardized scores
- Provides a way to compare scores in a similar process
e.g., children’s BMI
Z-score
common standardized score
a visual representation of data, on a scaled graph, with two axes
- used to display matched values
(e. g., ht vs wt)
scatterplot (measure of dispersion)
- have two scales: horizontal axis (X) and vertical axis (Y)
- illustrate a relationship between two variables
unless drawn to scale, a scatterplot is only fairly good at displaying dispersion, and provides no quantification
infer or address objectives, questions, and hypothesis
- used to test an actual or implied hypothesis, emanating from the research purpose
inferential statistics
assists in:
- identifying relationships
- examining predictions
- determining group differences in studies
those testing relationship and difference are the primary ones
types of inferential statistics
Examining relationship
- pearson product-moment correlation
- factor analysis
Predicting outcomes
- regression analysis
Examining differences
- chi-square
- t-test
- analysis of variance (ANOVA)
used to identify the existence of a relationship between or among variables, as well as discovering the direction of the relationship and its magnitude
correlational tests
tests for the presence of a relationship between two variables
pearson product-moment correlation
Results:
- Nature of the relationship (positive or negative)
- Magnitude of the relationship (-1 to +1)
- Testing the significance of a correlation coefficient
- Does not identify direction of a relationship (one variable does not cause the other)
- Are symmetrical
Pearson product-moment correlation (r) determines presence, direction, and amount of relationship between two ratio-level or interval-level variables.
- Values range from -1.00 (perfect negative, or inverse) through +1.00 (perfect positive).
- In a positive relationship, as values of one variable increase, so do the values of the other variable. As values of one variable decrease, so do the values of the other variable.
- In a negative relationship, as values of one variable increase, the values of the other variable decrease.
- A value of 0 indicates no relationship, as do values very near 0. The value +0.0001 does not signify a weak positive relationship: it means that there is no relationship.
In general:
- Values between -0.3 and +0.3 indicate weak correlations
- Values between -0.5 and -0.3, or between 0.3 and 0.5 indicate moderate correlations
- Values less than -0.5 or greater than 0.5 indicate strong correlations
the value of r2 is
an estimate of the explained variance
- it is the amount of difference in the value of one variable that is related to a change in the other
- the value is near 100% when one variable changes only when the other changes, and in perfect proportion (this is rare)
used when one wishes to predict the value of one variable based on the value of one or more other variables
regression analysis (predicting outcomes)
- used to predict the value of one variable (the outcome or dependent variable), using one or more other variables as “predictors” (the independent variables)
- prediction is not the evidence of causation
tests for significant differences between two samples (means)
- most commonly used test of differences
t-Test (examining differences)
example: t = 4.169 (p < 0.05)
tests for differences between variances
- more flexible than other analyses in that it can examine from two or more
analysis of variance (ANOVA)
(examining differences)
- if there are more than 2 groups under study, it is not possible to determine where the significant differences are
- post hoc tests are used to determine the location of differences
- is much like the t-test but examines variances instead of means, again to determine whether the groups are the same
tests for differences between expected frequencies if groups are alike and frequencies are actually observes in the data
- used with nominal or ordinal data
chi-square test (examining differences)
- indicate that there is a significant difference between some of the cells in the table
- the difference may be between only two of the cells, or there may be differences among all of the cells
- will NOT tell you which cells are different
example: x2 = 4.98, df = 2, p = 0.05
- the chi-square test of independence essentially determines whether two or more groups differ
- the frequency observed values is compared with values that would have occurred by chance, were all conditions equal
5 types of results associated with quasi and experimental studies
- Significant results that agree with those predicted by the researcher
- Nonsignificant results
- Significant results that are opposite those predicted by the researcher
- Mixed results
- Unexpected results
