Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
196 Cards in this Set
- Front
- Back
Sampling: Purpose
|
• Gain information from a small group so that findings can be generalized to the larger population
• Sample must represent the larger population • Must have a clear rationale for sampling techniques to ensure correct selection of subjects |
|
• Population-
|
– entire set of subjects, objects, events or elements being studied
– Possess specific attributes |
|
• Sample
|
– Small subset of the population
– Must be carefully selected so that it represents population |
|
• Target population
|
– Population used for a study
– The entire set of elements about which the researcher would like to make generalizations |
|
• Accessible population
|
– The portion of the target population that is readily available to the researcher and represents the target population as closely as possible.
|
|
Sampling
|
• Process of selecting individuals for a study in such a way that individuals represent the larger group from which they were selected.
• Allows researcher to draw inferences & make generalizations about the population without examining every element in the population |
|
• Probability sampling
|
– Every subject, object or element in population has an equal chance or probability of being chosen
|
|
• Nonprobability sampling
|
– Not selected randomly
• Probability of inclusion and degree to which the sample represents the population are unknown |
|
Probability sampling: Simple random
Example |
– Example: sample included patients aged 20- 80 years having a CABG for the first time. Sample of 40 patients who experienced SVT & 40 patients who did not experience SVT was randomly selected from a list obtained from the medical records department using the Classification of Disease code- the list was not in any systematic chronologic or alphabetic order.
|
|
Probability sampling: Simple random
|
• Every subject has an equal & independent chance of being chosen
• Use of Table of Random Numbers – Can be generated by a computer – Assign all potential subjects a number – Close eyes & point to a number on the Table – This tells which subject has been selected. • Disadvantages of Simple Random Sampling – Time consuming – May not be possible to obtain complete list of target population |
|
Probability Sampling: Stratified
Random Sampling |
– Selecting a sample to identify sub-groups of the population that are represented in the
sample. – Reduces possibility that the sample might be unrepresentative of the population – Achieves a greater degree of representativeness with each subgroup (strata) |
|
Probability Sampling: Cluster
|
• Successive random sampling of units
• First unit to be sampled is a large grouping or cluster • Then draw from a smaller cluster • Example- a sample of nursing students |
|
Probability Sampling: Systematic
|
Subjects or elements are selected from a list by taking every kth individual
• Is not strictly probability sampling because not all members have an equal chance of being selected • But is considered random especially if list is randomly ordered |
|
Probability Sampling: Overall Comments
|
• Every subject has equal chance of being chosen
• Allows researcher to estimate the magnitude of the sampling error – Difference between the population values and sample values • Is expensive, time consuming & inconvenient |
|
Nonprobability
|
• Chance plays no role in determining the sample
• Limits the ability to generalize |
|
Nonprobability: Convenience
|
• Collection of data from subjects or objects readily available or easily accessible to the researcher
• Subjects not selected from a larger population • Data are collected from whoever is available & meets study criteria |
|
Nonprobability: Convenience
|
• Advantage: easy, saves time & amp; money
• Disadvantages: sampling bias, the sample used may not represent the population • Type of Convenience called “snowball” – First subjects in study are asked to refer researcher to other people who meet the eligibility criteria – so the sample grows like a “snowball” |
|
Nonprobability: Quota
|
• Researcher identifies strata of the population & amp; determines the proportions of the elements needed from the various segments of the population
• All segments are represented in the sample in the proportions in which they occur in the population |
|
Nonprobability: Purposive
|
• Also called judgmental sampling
• Based on belief that are searcher’s knowledge about the population can be used to hand pick the cases to be included in the sample • Uses:to develop an instrument (pretest) – When researcher wants experts in the field • Disadvantages: no external objective method for assessing typicalness of subjects |
|
Nonprobability Sampling
|
• Acceptable for pilot, exploratory or in- depth qualitative research
• Presents problems for quantitative research – Not every element of population has a chance of being selected for the sample – Some segment of the population will be underrepresented |
|
Sample Size (Quantitative Research)
|
• Standard rule: use the largest sample possible
• Two critical questions for determining adequacy of sample – How representative is the sample relative to the target population? – To whom does the researcher wish to generalize the results of the study? |
|
Sample Size (Quantitative Research)
Factors |
– Type of sampling procedure used
– Type of sample estimation formula being used – Degree of precision required – How many attributes under investigation – Relative frequency of occurrence of phenomenon of interest in population – Projected cost of utilizing a particular strategy |
|
Sample Size: Power Analysis
|
• A statistical method to calculate the exact number of subjects needed
• Based on effect size: concerned with the strength of the relationship(s) among research variables or the impact made by the independent variable. |
|
Sample Size: Homogeneity
|
• The degree to which objects or subjects are similar
|
|
Sample Size: Effect size
|
• Effect size is concerned with the strength of the relationships among research variables
|
|
Sample Size: Attrition
|
• Loss of subjects over time
• Increases if time lag between data collection points is great, population is mobile, high-risk |
|
Sampling Questions
• A(n) _______ is a subset of the units that comprise the population. |
Sample
|
|
Sampling Questions
• The main criterion for evaluating a sample is its __________________. |
Representative
|
|
Sampling Questions
• A sample would be considered ______ if it systematically overrepresented or underrepresented a segment of the population. |
Biased
|
|
Sampling Questions
• If a population is completely ___________ with respect to key attributes, then any sample is as good as any other. |
Homogenous
|
|
Sampling Questions
• Quota samples are essentially convenience samples from selected ______ of the population. |
Strata
|
|
Sampling Questions
• The most basic type of probability sampling is referred to as _______________ |
Simple Random
|
|
Sampling Questions
• When disproportionate sampling is used, an adjustment procedure known as ____________ is normally used to estimate population values. |
Weighting
|
|
Sampling Questions
• In systematic samples, the distance between selected elements is referred to as the ____________. |
Sampling interval
|
|
Sampling Questions
• Differences between population values and sample values are referred to as ________________. |
Sampling Error
|
|
Sampling Questions
• As the size of the sample __________ the probability of drawing a deviant sample diminishes. |
Increases
|
|
Sampling Questions
• If a researcher wanted to draw a systematic sample of 100 from a population of 3000, the sampling interval would be ___. |
30
|
|
Measurement
|
• Consists of rules for assignment number of objects to represent quantities of attributes OR
• Process of assigning numbers to variables • Because “whatever exists in some amount can be measured” |
|
Measurement: Rules (1)
|
• A process must be used to assign numbers
• Rules for measuring variables for research studies must be invented – Under what conditions – What method (survey, observation etc) – Numeric values to be used |
|
Measurement: Rules (2)
|
• Example: survey for parent’s opinions of sex education in school
• Strongly disagree • Agree • Slightly agree • Undecided • Slightly disagree • Disagree • Strongly disagree • Would want to quantify the responses–how? |
|
Measurement: Advantages
|
• Removes the guess work from gathering information
• Objectivity: can be independently verified by others • Produces precise information • Language of communication |
|
Errors of Measurement
|
• If instrument is not accurate then the measures it produces contains a certain degree of error
• Obtained score = true score + error |
|
Errors of Measurement Disadvantage
|
• Errors of measurement are problematic because they represent an unknown quantity and they are variable
|
|
Errors of Measurement
Factors contributing to errors of measurement: |
– Situational contaminants- scores can be affected by
the conditions under which they are produced – Transitory personal factors-scores of an individual may be influenced by temporary personal states – Response-set biases- number of relatively enduring characteristics of the respondents that can interfere with accurate measures of the target attribute. |
|
Errors of Measurement
• Administration variations |
alterations in the methods of collecting data from one subject to the next
|
|
Errors of Measurement
Instrument clarity |
if directions for obtaining measures are vague or poorly understood, then scores may be affected
|
|
Errors of Measurement
Item sampling |
errors are sometimes introduced as a result of the sampling of items used to measure an attribute
|
|
Errors of Measurement
Instrument format |
technical characteristics of an instrument can influence the obtained measurement
– Open ended questions may produce different information than closed–ended questions |
|
Instruments used in Measurement
|
• Devices used to record data obtained from the subject
• Interviews, direct observations, mechanical equipment etc |
|
Reliability
|
• The degree of consistency with which an instrument measures the attribute it is supposed to be measuring
• Reliability = stability, consistency, dependability of a measuring tool • The less variation an instrument produces – the higher its reliability |
|
Reliability: Types
Stability |
– Stability- extent to which same results are obtained on repeated administrations of the instrument
• Test-retest procedure- administration of the same test to a sample of individuals on two occasions then compare the scores obtained • Compute a reliability coefficient |
|
Reliability coefficient
|
– Looks for a relationship between two phenomena
Positive Correlations 1.0 Negative Correlation -1.0 |
|
Reliability Advantages
|
– Test-retest method is easy
– Can be used with self- report, observational & physiologic measures – Used with traits that are relatively enduring such as personality, abilities or physical attributes such as height |
|
Reliability Disadvantages
|
– Many traits do change over time- regardless of the measure eg (behaviors, attitudes etc)
– Second administration may be influenced by memory of first administration – Subjects may change as a result of taking the test the first time – Responses could be haphazard if subject is bored or impatient with instrument |
|
Reliability: Internal Consistency
|
• Addresses the correlation of various items within the instrument – also called homogeneity
• All parts of the scale are measuring the same characteristic • Most widely used method for reliability • Requires only one test administration |
|
Reliability: Internal Consistency
• Split-half technique |
– The items composing a test are split into two groups and scored independently
– Score on the two half-tests are used to compute a correlation coefficient |
|
Reliability: Internal Consistency
• Split-half technique Advantages |
– Easy to use
– Eliminates most of disadvantages of test- retest |
|
Reliability: Internal Consistency
• Split-half technique Disadvantages |
– Reliability estimates can be obtained by using different splits (odd/even; first/second half)
|
|
Reliability: Internal Consistency
• Chronbach’s alpha or Kuder-Richardson 20 (KR 20) |
• Chronbach’s alpha or Kuder-Richardson 20 (KR 20)
• Produce a reliability coefficient that can be interpreted as the other correlation coefficients • Yields values of 0 to 1.0 • Better than split-half because it computes all possible splits to estimate homogeneity |
|
Reliability: Internal Consistency
• Equivalence |
– Comparing two versions of same instrument or two observers (inter-rater reliability) measuring the same event
– Goal is to determine the consistency or equivalence of the instruments in yielding measurements of the same traits in the same people |
|
Reliability: Internal Consistency
• Interrater reliability |
• is estimated by having two or more trained observers watching some event simultaneously & independently recording the relevant variables according to a predetermined plan or coding system
•Results are used to compute an index of equivalence or agreement •Can use a correlation coefficient to compare results |
|
Reliability: Internal Consistency
• Interpretation of Reliability Coefficients |
– Reliability is the proportion of true variability to the total obtained variability
r= VT Vo |
|
Validity
|
• Refers to the degree to which an instrument measures what it is supposed to be measuring.
• Accuracy of the measure • If an instrument is designed to measure hopelessness-how can researcher be sure that it is measuring hopelessness? What if the instrument is really measuring depression? |
|
Validity: Methods to Determine
• Face validity |
– Whether the instrument looks as though it is measuring the appropriate critical variable (or construct).
– Not the strongest way to determine validity |
|
Validity: Methods to Determine
• Content validity |
– How well the instrument represents the characteristic to be assessed
– Used for affective measures (feeling, emotions, psychological traits) & cognitive measures (knowledge) – Based on judgment of experts |
|
Validity: Content Examples
|
• To develop an instrument which is to measure affective characteristics, a researcher would
– Have knowledge of the subject – Do a thorough investigation of the literature – Conduct some type of investigational study |
|
Validity: Content Examples
• Cognitive scale |
– Must ask the question: “how representative are the questions on this test of the universe of all questions that might be asked on this topic?”
– If a researcher were interested in developing a scale that tested people’s knowledge of the seven danger signals of cancer, the instrument would have to include the 7 danger signals identified by the ACS – CAUTION |
|
Validity: Content
Disadvantages, Methods |
• Based on judgment
• Not completely objective methods of ensuring the adequate content coverage of an instrument • May use a panel of experts in the content area to evaluate & document the content validity – Has a least 3 members – Ask experts if items are relevant and appropriate – Whether individual items measure all dimensions of the construct • May use a Content Validity Index |
|
Validity: Criterion-Related
|
• Emphasis is on establishing the relationship between the instrument and some other criterion
• Whatever attribute is being measured, the instrument is said to be valid if its scores correlate highly with some other criterion • Pick a criterion to compare an instrument to |
|
Validity: Criterion-Related
Example |
• instrument to measure birth control use among sexually active teenage girls- criterion to compare it to may be subsequent premarital pregnancy
• Measure professionalism among nurses (attribute to be measured) to number of articles published (criterion) • Measure effectiveness of nursing care to supervisory ratings of nurse • Criterion must be clear cut & objective |
|
Validity: Criterion-Related
Predictive validity |
• Criterion related validity is a relationship but also is a predictive relationship
• Predictive validity – Degree to which an instrument can accurately forecast the future |
|
Validity: Concurrent
|
– Judgment is to the degree to which an instrument can accurately identify a difference in the present
– Use 2 instruments to measure the same concept |
|
Validity: Construct
Types of questions |
• What is the instrument really measuring?
• If instrument measures pain- how does researcher know scale is not measuring anxiety? |
|
Validity: Construct
Methods and Example |
known groups technique
– Groups that are expected to differ – For example: • Fear of labor : Use primipara & multipara women • Limitations in functional ability: use one group of people with emphysema and one group with no emphysema |
|
Validity: Construct
• Factor analysis |
– Method for identifying clusters of related variables
– Each cluster is called a factor • Represents a group of items that identify the same characteristic – Procedure is used to identify and group together different measures of some underlying attribute – Is a complex equation |
|
Validity
|
• Validity is never proven, verified
• It is supported by evidence • An instrument may be useful in one situation but not another – An anxiety scale may be useful in a group of presurgical patients but not in a group of nursing students getting ready to take an exam • Each new use requires new supporting data |
|
Reliability & Validity
|
• Are not totally independent qualities
• A measuring device that is valid must be reliable • An instrument can be reliable without being valid |
|
Reliability & Validity
Low reliability/low validity |
Draw
|
|
Reliability & Validity
High reliability/ low validity |
Draw
|
|
Reliability & Validity
High reliability, high validity |
Draw
|
|
Levels of Measurement
Variables Types:Continuous |
-Values can be represented on a continuum
-infinite number of values between two points -Example: weight |
|
Levels of Measurement
Variable Types:Discrete |
-has a finite number of values between any two points
-representing discrete quantities -Example: number of children, 1, 2, 3 |
|
Levels of Measurement
Variables: Discrete Categorical |
-have only a few discrete values
-Ex: marital status, blood type -Dichotomous- have only two values -married/not married, alive/dead, pregnant/not pregnant |
|
Types of Measurement
|
-4 levels
- Type of measurement will determine the statistics that can be used - Each level is classified in relation to certain characteristics |
|
Types of Measurement
Nominal |
- First level of measurement
- Variables that are discrete & noncontinuous - Categories: gender, marital status, - Can be dichotomous: has two categories - is the most primitive method of classifying information - Is the demographic data - Only simple statistical tests can be used |
|
Types of Measurement
Nominal Examples |
-Gender
-Marital Status -Blood Type -Types of meds |
|
Types of Measurement
Nominal |
- Classifications must be mutually exclusive
- Collectively exhaustive - Numbers used in nominal measurement cannot be treated mathematically - cannot calculate the “average” gender - calculate in percentages |
|
Types of Measurement
Ordinal Level |
- Second level
-Variables are assessed incrementally - pain - ability to perform ADLs - Goes beyond categorization - Variables are ordered according to some criterion |
|
Types of Measurement
Ordinal Level Differences |
- Difference between nominal & ordinal measurement
- there is an ordering (relative standing) -a value assigned show the relationship to the other subjects - does not reveal anything about how MUCH greater one level of an attribute is than another - cannot say one person has twice the functional capacity of another |
|
Types of Measurement
Ordinal Level Restrictions |
- Types of statistical tests and mathematics are restricted
|
|
Ordinal Level
|
Pain Slight Moderate Intense
Anxiety experienced Frequently Occasionally Rarely Bonding Absent Minimal Moderate Strong |
|
Interval Level
|
- Third level
- scale that is quantitative in nature - there is both rank-ordering of objects on an attribute & distance between numeric values - Increments on the scale can be measured & are equidistant are continuous variables |
|
Interval Level
- Examples |
-SAT tests: score of 550 is higher than 500
-which is higher than 450 -all are equidistant apart |
|
Interval Level
|
- Gives more information than first two categories
- Major disadvantage: cannot give absolute magnitude of the attribute - no real or rational zero |
|
Interval Level
Example |
- Thermometer: is interval level
- no “0” in the Fahrenheit scale - 600 is not twice as hot as 300 |
|
Interval Level
|
- Does expand statistical tests that can be used
- can add subtract data- so there can be an average |
|
Ratio Level
|
- 4th & highest level of data
- Characterized by variables that are assessed incrementally with equal distances between increments - A scale that has an absolute (meaningful) zero |
|
Ratio Level
|
- All arithmetic operations can be used
- All statistical procedures can be used - Is ideal for researchers but not attainable for many variables - Example: weight, height |
|
Converting Data to Lower Level
of Measurement |
- Data can always be converted to a lower level but not higher
- loss of accuracy & information - Example: how could you change age from ratio to ordinal level? - Use categories: 10-15 15-20;20-25etc - What is lost by doing this? |
|
Levels of Measurement
Examples |
- What are adolescents’ views and practices relating to tattooing?
- Data collected on: - if the subject had a tattoo - grades in school - purpose of tattooing (scale) - age at first tattoo |
|
Give the highest possible level of measurement that a researcher could obtain for each of the following:
|
-Attitudes towards the mentally handicapped
-Birth order -Length of labor -White blood cell count -Blood type -Tidal volume -Unit assignment for nursing staff -Motivation for achievement -Amount of sputum - Pulse Rate |
|
Descriptive Statistics:
Frequency Distributions |
-A systematic arrangement of numeric values from lowest to highest
-With a count of the number of times each value was obtained |
|
Descriptive Statistics:
Frequency Distributions 2 Parts |
Two parts:
-Observed values or measurement -Frequency or count of the observations falling into each class -Classes of observation must be mutually exclusive & collectively exhaustive |
|
Descriptive Statistics:
Frequency Distributions |
Table
|
|
Descriptive Statistics:
Frequency Distributions |
- Can take raw data & organize into a Histogram
|
|
Descriptive Statistics:
Frequency Distributions |
- Histogram of scores
|
|
Descriptive Statistics:
Shape of Distributions |
-Set of numbers can have an infinite number of shapes
- Symmetrical: two halves that are mirror images of each other |
|
Descriptive Statistics:
Frequency Distributions |
Bar Graph
|
|
Descriptive Statistics:
Frequency Distributions |
- Asymmetrical distribution
- Usually described as skewed - One tail is longer than the other - When longer tail is pointing toward right -the distribution is positively skewed |
|
Descriptive Statistics:
Frequency Distributions Positive Skew |
Positive Skew
|
|
Descriptive Statistics:
Frequency Distributions |
Negative Skew
|
|
Descriptive Statistics:
Frequency Distributions |
- Frequent shape is called normal distribution or bell-shaped curve
- Is unimodal - Symmetrical - Not too peaked |
|
Descriptive Statistics:
Frequency Distributions Bell Curve |
Bell Curve
|
|
Descriptive Statistics: Central Tendency
|
- Yields more important information -Are indexes of typicalness
- more representative if values come from center - “average” is term used to designate central tendency |
|
Descriptive Statistics:
Central Tendency Mode |
- numeric value in a distribution that occurs most frequently
- 50 51 51 5253535353545556 |
|
Descriptive Statistics:
Central Tendency What is mode? |
-Is quick and easy
-Rather unstable - tend to fluctuate widely within a population -Can be used to give the most typical subject.... Such as....”the typical subjects was an unmarried white female living in an urban area with no prior history of STDs” |
|
Descriptive Statistics:
Central Tendency Median |
-Point at which & below which 50% of cases fall
-2 2 33 334 5 6 789 (12scores) |
|
Descriptive Statistics:
Central Tendency Mean |
-Point on the score that is equal to the sum of the scores divided by the number of scores
-The mean is affected by the value of every score -Most widely used measure of central tendency (interval & ratio level data) |
|
Descriptive Statistics: Variability
|
- How spread out of dispersed the data are
-Two distributions of scores can be totally different but means can be identical |
|
Descriptive Statistics: Variability
Range |
-Highest score minus the lowest score in a distribution
-Easy to compute -Highly unstable index- only based on 2 scores -Ignores the variations in scores between the two extremes -Used as a gross descriptive index |
|
Descriptive Statistics: Variability
Semi-quartile Range |
- A point below which any percentage of the scores fall
- Calculated on that basis of quartiles within a distribution -Half the range of scores within which the middle 50% of scores lie -Upper quartile Q3= point below which 75% of the cases fall Q1= point below which 25% of the scores lie |
|
Descriptive Statistics: Variability
Standard Deviation |
- Most widely used measure of variability
- SD summarizes the average amount of deviation of values from the mean - Based on deviation scores: - Calculate mean -Subtract mean from each individual score - Squaring each deviation score & then adding them together -Divide by number of cases -Take square root |
|
Descriptive Statistics: Variability: SD Calculation
|
Image
|
|
Descriptive Statistics: Variability: SD
|
- SD is an index of the variability of scores in a data set
- Tells us how much the scores deviate from that mean - So scores deviate 1.76 from mean - That is: scores are more clustered around mean |
|
Descriptive Statistics: Variability: SD
|
- Can be used to describe an important characteristic of a distribution
- Used to interpret the score of performance of an individual in relation to other in the sample - Is a stable estimate of a population parameter (ratio or interval data) |
|
Descriptive Statistics: Variability: SD
|
Image
|
|
Assumptions
|
• Beliefs that are held to be true but have not necessarily been proven
• OR • A principle that is accepted as being true based on logic or reason, without proof |
|
Assumptions
• Type 1 |
1) universal: beliefs assumed to be true by a large percentage of society
- all humans need love |
|
Assumptions
• Type 2 |
2) Derived from a theory of previous
research -stress causes disease |
|
Assumptions
• Type 3 |
3) specific to a certain research study
- evidence of a fit between what the researcher believes can happen & the data produced... assumptions that prediction is possible, facts can be verified, testing of theoretical relationships -In other studies (qualitative: assume patient is an active participant in some social environment ) |
|
Assumptions
|
• Assumptions ARE NOT THE research question or hypothesis
|
|
Limitations
|
• Def. Uncontrolled variables that may affect study results and limit the generalizability of the findings
•Limitations mentioned could include: – Sample deficiencies – Design flaws – Weaknesses in data collection |
|
Correlation
|
• Most common method of describing a relationship between two measures
• Correlational research examines relationships among variables interest without any intervention on the part of the investigator • Variables are interval or ratio |
|
Correlation
Pearson’s r |
• Calculated when data are interval/ratio
• -1to+1 |
|
Inferential Statistics
|
• To estimate the probability that statistic found in the sample accurately reflects the population parameter
• To test hypotheses about a population |
|
Probability
|
• Defined: likelihood that something will occur
• Researchers analyze data to determine the likelihood that differences in study groups are the result of chance as opposed to the manipulation of variables |
|
Probability
|
• Errors can always occur in a study
• Always a chance that the differences occurring in the study due to chance rather than the treatment • SO.....Research results cannot claim to prove anything but that there is a low probability that the results are due to chance |
|
Probability
|
Always a sampling error
– Discrepancy between the characteristics of the sample & the population • Researcher must decide if sample being used is good estimate of population parameters |
|
Probability Levels
|
• Calculating statistics provide probability that result is caused by sampling fluctuations
• Usually set no higher than .05 • p<.05 means there is less than 5% chance that results are due to chance |
|
Null Hypothesis
|
• Statement that there is no actual relationship between variables & that any such observed relationship is only a function of chance or sampling fluctuations
• Statistical hypothesis testing is a process of rejection |
|
Type I & Type II Errors
|
• Errors in decisions about rejecting or accepting the Null
|
|
Type I
|
• Rejecting the null when it is true
• EG if a researcher concludes that experimental treatment was more effective than the control in alleviating anxiety, when the study outcomes were actually a result of sample differences in anxiety scores • The lower the p level, the less likely for a Type I error |
|
Type II
|
• Not rejecting null when there was a significant difference between groups
• Researcher concludes that differences in group anxiety levels were result of chance – when the experimental treatment did have an effect on anxiety |
|
Type II-Significance
|
• Level of significance set at .05 because if there is more than 1 chance in 20 that the outcome of interest has occurred by chance rather than manipulation, the results are not considered to be of value
|
|
Type II: ways to decrease chance of Type II errors
|
• Larger sample size
• Decrease sources of wide variation (control) • Increase level of significance |
|
Scenario
|
Test for genetic defect- if defect exists & is diagnosed early- it can be successfully treated
•If not diagnosed, & treated, child will become severely debilitated. •If child is diagnosed as having defect & treated, no damage occurs •Null hypothesis: The test for the genetic defect will reveal no difference between control & experimental groups. •Type I error: diagnosing defect when it does not exist- child not harmed because treatment won’t hurt him •Type II error: declaring child normal when child is not so child is severely damaged. |
|
Type I & Type II errors
|
• Would like to eliminate them
• Can’t without using entire population |
|
Statistical Significance
|
Image
|
|
Statistical vs Clinical Significance
|
• Statistical significance- the differences observed are probably true differences & not result of chance fluctuations in sampling
• Clinical significance- findings must have meaning for patient care – Nurses must determine if results have meaning to patient care & practice |
|
Two tailed tests
|
– Most often used
– Both ends (or tails) of sampling distribution are used to determine the range of improbable values |
|
Two tailed tests
|
There is a difference between males & females in their approval of physical conflict
Images |
|
One Tailed
|
Females are less approving of physical conflict than males
Image |
|
Hypothesis Testing Procedure
|
• Determine the test statistic to be used
• Establish the level of significance |
|
Parametric/Nonparametric
|
• Two classes of statistical tests
• Each have own characteristics |
|
Parametric Tests
|
• Involve the estimation of at least one parameter
• Require measurement(s) on at least the interval scale • Assume the variables are normally distributed in the population • Powerful tests and offer flexibility • Preferred for data analysis |
|
T-test (Student’s T)
|
• Test for differences between groups
• Use mean scores between groups and test for differences • A t-value is generated • Probable values are computed • p level is determined • Null rejected/accepted • Used for independent groups |
|
Paired T-test
|
• Two measures from the same subjects over time
|
|
Analysis of Variance (ANOVA)- one-way
|
• Tests for differences between means for 3 or more groups
• One independent variable • ANOVA yields an F-ratio statistic |
|
Multifactor ANOVA
|
• Two or more independent variables on a dependent variable
|
|
Non parametric tests
|
• Tests for nominal or ordinal level data
• Shape of distribution is not a concern • Sample size can be small |
|
Chi-square X2
|
• Used with categories of data
• Hypotheses concerning the proportions of cases that fall into the various categories • Chi-square is computed from contingency tables • Differences are calculated based on what occurs from a study compared to what the expected frequencies are • Must have at least 5 in each group |
|
Chi-square X2
|
Image
|
|
Power
|
• Ability of a study to identify relationship or detect real differences among variables
|
|
Power Analysis
|
• Method for reducing risk of Type II errors (Not rejecting null when there was a significant difference between groups)
• Estimating occurrence of Type II error |
|
Power Analysis
|
• Used to estimate the sample size needed to obtain a significant result and allow the researcher to conclude that the research hypothesis is supported
|
|
Power Analysis
Four elements |
– Significance level or alpha
– Sample size – Effect size – power |
|
Power Analysis
Alpha |
• probability of making a type I error
• Usually set at .05 |
|
Power Analysis
• Beta |
– Probability of making a type II error
– Should be no more than 4 times the value of alpha – .20 (based on alpha of .05) |
|
Power Analysis
Power |
–1-beta =1-.2=.8
– Conventional standard accepted for power is .80 |
|
Power Analysis
• Effect size |
– Strength of the relationship among study variables
– Measures how false the null hypothesis is – that is how strong the effect of the independent variable is on the dependent variable – When relationships are strong- large samples are not needed |
|
Power Analysis
• Effect size determined from |
– ROL
– Researchers’ own pilot data – Estimates based on clinical experience |
|
Power Analysis
|
• With all four of these pieces of information, a sample size can be determined using the Power Analysis formula
|
|
Qualitative Traditions: General Characteristics
|
• “Multi-method focus that involves an interpretive, naturalistic approach t its subject matter
• A holistic approach to questions that recognize that human realities are complex • Research question is very broad •Individual’s perspective is very important |
|
Qualitative Traditions: General Characteristics
|
• Concerned with in‐depth description of people or events
• Data collected through interviews or observations • Uses an inductive approach • An emergent design |
|
Qualitative Traditions: General Characteristics
|
• Flexible‐ capable of adjusting to what is being learned during the course of data collection
• Merges together various methodologies • Holistic‐ strives for understanding of the whole |
|
Qualitative Traditions: General Characteristics
|
• Focused on understanding a phenomenon or social setting‐ not on making predictions about the setting or phenomenon
|
|
Qualitative Traditions: General Characteristics
|
• Requires that the researcher become intensely involved for lengthy periods of time
• Requires that the researcher becomes the research instrument • Ongoing analysis of the data in order to formulate subsequent strategies & to determine when field work is done |
|
Qualitative Traditions: General Characteristics
|
• Forces researcher to define their role & identify their own biases
|
|
Designs & Planning
|
• Plans for data collection setting
– Gaining entry – Obtaining consent – Identifying the site’s “major players” – Determining maximum amount of time available for the study – Equipment needs: tape recorder, lap‐top etc • Training of assistants |
|
Designs & Planning
|
• Data collection format
– Plans for “unforeseen” events – Setting for interviews • Researcher & team must analyze their own biases & ideology |
|
Phases
|
• Orientation &overview
– Must decide what they don’t know and how to handle the phenomenon • Focused exploration – Focused scrutiny and in‐depth exploration of those aspects of the phenomenon that are judged to be salient •Confirmation &amp; closure – Try to establish that their findings are trustworthy |
|
Design Features
|
• No IV or DV identified before data collection begins
– May look back to find antecedent factors leading up to the occurrence of that phenomenon • Flexible • No manipulation o rcontrol • Group comparisons not planned but may occur • Can be cross‐sectional or longitudinal • Setting‐real‐world |
|
Qualitative Research –Types: Ethnographic
|
– Social scientific descriptor of a people & the cultural basis of their identify
– Study of cultures/behaviors – Only focus is to describe.... Is a theoretical • data about cultural groups • Is oldest qualitative approach – Cultural behavior‐ what members of the culture do – Cultural artifacts: what members of the culture make &amp; use – Cultural Speech‐ what people say |
|
Qualitative Research –Types: Critical Social Theory
|
– Also called feminism
– Goal is to raise consciousness, politicize or activism – Research focus is on oppressed groups |
|
Qualitative Research –Types: Content Analysis
|
– Used to describe the basic content of data
• Speeches • Books • Interviews • Media (films, photography) – Counts the number of times something occurs‐ eg the number of times the same words are used |
|
Qualitative Research –Types: Content Analysis
|
• Data are coded to identify recurrent themes
• Used to describe attitudes, expectations & perceptions • Can be used alone or in conjunction with other methods |
|
Qualitative Research Types
Narrative Analysis |
• Method applied to stories of meaningful account of events over time
• Structure of a story includes what form it takes such as the history of a disease or an account of personal experiences & how events are related to each other |
|
Qualitative Research –Types: Grounded theory
|
• Data are collected & analyzed
• Study of social processes & structures • Uses open‐ended interviewing, sensitization to concepts • Main focus: theory is developed that is grounded in the data |
|
Qualitative Research –Types: Historical
|
– Identification, location & evaluation & synthesis of data from past
– Purpose is to answer questions concerning causes, effects or trends relating to past events that may shed light on present behaviors or practices. |
|
Qualitative Research –Types:
Case studies |
– In‐depth investigations of a single entry or a small series of entities
• May be an individual but also can be families, groups, etc. • Focuses on why the individual thinks, behaves or develops in a certain way rather than on what the status, actions or thoughts are. |
|
Qualitative Research ‐Analysis
|
• Perform a content analysis
•Analyzes the narrative data to determine themes or patterns • Very tedious‐no systematic rules for analyzing & presenting qualitative data • Must interpret data‐can be subjective so must develop a method to classify & index materials • Coding of data‐once categories have been established ‐ then data is re‐reviewed for coding |
|
Evaluation of Qualitative Research
|
• Must be rigorous
• Credibility: the “truth” of the findings from the subjects that is: “is this what you were saying?” • Transferability‐ study’s ability to preserve meanings, interpretations & inferences when applied to another similar context |
|
Evaluation of Qualitative Research
•Confirmability |
obtaining direct & repeated affirmation of what the research has heard, seen or experienced
|
|
When is Qualitative Research a Good Choice?
|
• When a topic has not been studied
• When quantitative approach does not or would not give a full picture |
|
Qualitative Research
Examples |
-Grounded Theory- what is the social process underlying a nurses' experience with implementing developmental care in a neonatal ICU?
-What are the critical health-related, social & economic issues of the Afgan refugee community? |
|
Application to Practice
|
-How do the results of this study help me care for patients?
-How does this study help me understand my relationship with my patients & their families" |