Research Term Test 2 Flashcards
(43 cards)
1
Q
4 levels of data
A
- nominal
- ordinal
- interval
- ratio
2
Q
Nominal
A
- allows to distinguish differences between items qualitatively
- no quantitative ordering or value
assign responses to different categories
- sex
- marital status
- postal code
- university major
- student ID
3
Q
Ordinal
A
- categories have logical order
- starts at lowest and ends at highest
- unknown numerical distance between catergories
- Likert scales - allows for comparison in relative terms
- ex. better or worse, smaller or larger
- letter grade in class: F - A
- Degrees held: BSc, MSc, PhD
- perceived exertion
4
Q
Interval
A
- measurement are numerical values
- intervals of equal length represent equal differences in characteristics
- Zero, does not signify absence of characteristic (think temperature)
- Starting point arbitrary
- ex. behavioural questionnaires, IQ test
5
Q
Ratio
A
- allow for id of absolute differences
- absolute zero
- zero means absence of characteristic
- most measured data first in this category
- ex. BMI, weight, height, VO2, age, time to completion
6
Q
Graphs - Describing data
A
- simplest way for describing data
- self-contained bundle of info
- title indicating variables, clear id of categories and values, units of measurements indicated
7
Q
Pie charts
A
- distribution of cases in form of a circle
- relative size of slice is proportional to proportion of cases within catergory
- can be used for all levels of measurements
- emphasize the relative importance of particular category to the total
- difficult to interpret when there are too many categories (~~5 max)
8
Q
Bar graphs & histograms
A
Horizontal axis: absicca
- categories or values of the scale
- independent variable
Vertical axis: ordinate
- frequencies: raw count or percentage
- calculated data
- dependent vairable
Bar graphs
- used when data is discrete (nominal or ordinal)
- gaps between bars
Histogram:
- sed when data is continuous (internal or ratio)
- no gaps between bars
9
Q
4 aspects of distribution
A
- shape
- centre
- spread
- existence of outliers
10
Q
Shape of graphs
A
- normal distribution
- skewness describes if its shape is off centered to right or left
- positive skewed (long tail to the right)
- negative skewed (long tail to the left)
11
Q
Center of graphs
A
place where equal number of score are on each side
- seen as the average
12
Q
Spread of graphs
A
- how tighly clustered the measurements aroudn the central point
- wide spread = heterogenous scores = platykurtic
- narrow spread = homogenous scores = leptokurtic
13
Q
Outliers of graphs
A
- upper and lower limits
- also ones that are disconnected from rest of the group
14
Q
Descriptive statistics
A
- used to characterize a group based on data taken from the group
Includes measures of:
- central tendency: extent to which data clusters around a point
- variability: extent to which data are spread out
15
Q
Central tendencies
A
- mean: arithmetic average of groups of numbers, calculated as sum of all numbers divided by total numbers of values in set
- (limitation to mean): affected by presence of outliers, sum of values is pulled away from middle when data is skewed, can produce value that is higher or lower than expected
- Median: single data value that resides in the middle of the data distribution
- mode: most frequent score in a distribution (limitation: often does not represent actual middle of values)
16
Q
Measures of variability/dispersion
A
- range: difference between high and low score
- standard deviation: estimate of spread of scores away from the mean
- standard error of the mean: estimate of expected difference between sample mean and population mean
17
Q
Inferential statistics
A
relationship between variables
useful example:
- is there a relationship between a simple field test and a difficult lab test for a variable of interest?
18
Q
Relationship between 2 variables
A
- bivariate statistical analysis
- a change in one variable is associated with a certain change in another
- theoretical model - way to describe relationship
- Ex. smoking history (independent) and health level (dependent)
Two ways:
- measure each variable separately
- measure association between variables
19
Q
Analyzing many variables
A
- regression analysis
- depict relationship between the dependent and 1 or more independent
- asks how multiple known quantities affect dependent (unknown) quantity
- creates one equation showing the relationship and allows to make predictions about dependent of a given individual, if you know all values for independent variables
20
Q
Categories of stats
parametric
A
used for interval and ratio data that meet following assumptions:
- variable of interest must be normally distributed within pop
- must have same variance within samples drawn from pop
- score or measures of variable must be independent
21
Q
Categories of stats
nonparametric
A
used for nominal and ordinal, as well as for interval and ratio that do no meet assumptions stated above
22
Q
Significance in research
A
- reject the null means real difference exists bewteen groups (if a= .05 and p= .049, reject null)
- do not reject null means no group differences exist (if a= .05 and p= .52, do not reject null)
23
Q
PICO
A
P: Population, patient, problem: who are the patients and what is the problem?
I: intervention or exposure: what do we do to them and what are they exposed to?
C: Comparison: what do we compare the intervention with?
O: Outcome: what happens?
24
Q
Purpose of research proposal
A
- provide action plan for study
- serves as contractual agreement between researchers and those who approved the proposal
25
Who Approves the research proposal?
- granting agency
- institutional review board
- faculty advisor and commitee
- ethics boards
26
Rationale for study
- intros topic and makes clear the importance
- explains how the results will fill gap in related literature
- ends with problem statement and specific aims
27
Research Questions or Problem Statement
- focal point for study
- IDs independent and dependent variables
- should be of interest and importance
28
Characteristics of research question
- Novel: has not been answered
- researchable: can be addressed through collection and analysis of data
- practical: necessary equipment, expertise, access to participants, and time are available
- clearly stated: written straightforward , readily comprehensible using appropriate research language
- timely: addresses current problem or issue
- in line with priorities of funding agency or institution
29
Specific aims
- two or more related research questions or problems being proposed for investigation within the same study
- each specific aim has research hypothesis associated
30
Methods for study
- includes comprehensive description of proposed methods
- level of detail should allow competent replication of study
- written in future tense in proposal and past tense in thesis or dissertation or research paper manuscript
31
Info regarding study participants
- recruitment process
- participation inclusion and exclusion criteria
- selection procedures
- informed consent docs
32
In vitro studies (primary research)
Strengths:
- intervention
- can assess mechanism
- strict control of experiment = high interval validity
- can assess cause and effect
Weaknesses:
- low external validity
- not directly comparable to humans
33
Epidemiological studies (primary research)
Strengths:
- large data sets
- high external validity
- can assess relative risk and odds ratios
Weaknesses:
- observational
- hypothesis generating - not causal
- little control = confounding variables = low internal validity
34
Clinical trials (primary research)
Strengths:
- highly controlled = high internal validity
- intervention
- can assess cause and effect
- human participants
Weaknesses:
- can be difficult to conduct on large size
- expensive
- can have selection bias
- placebo effect
35
Reviews (secondary research)
Strengths:
- strict methodology
- combines all available data sets
- can asses effect size
Weaknesses:
- no direct intervention
- relies on data from others
36
Codes in ethics
Nuremberg Code
- 1947 code for biomedical reserach was first to focus on informed consent
- direct response to what Nazis did in WWII
- Not a law but is basis for laws in many countries
Helsinki Declaration
- 1964 declaration provided additional guidance in such areas the use of animals
37
Ethical considerations in research
- informed consent
- ethics boards
- confidentiality
- plagiarism
- data manipulation
- publishing
38
Framework for medical ethics
- respecting autonomy, justice, doing no hard, using resources fairly
- provide basic needs with best available healt care, protect from abuse, neglect, discrimination
- aim to avoid/reduce harm and cost to promote benefits
39
Informed Consent
- well-documented informed consent is hallmark of ethical reserach
- responsibility of the investigator to obtain the informed consent of prospective participant or consent of legal guardian
- respects individual autonomy to participate or not participate
- simple and easily understandable language prior to signing
- provide participants with a copy
- tell participants who's conducting study
- any benefits for the participant to be expected
- any potential risks and have they have been managed
40
Research Ethics Committees (REB or IRB)
- protective barrier between researchers and participants
- review if benefits justify risk
- veto unethical research
- confirm that certain groups have not over-researched
41
Confidential Info
- directly identifying info: name, SIN, health care
- indirectly identifying info: birth date, physical characteristics
- coded info: only people with key can access personal identifiers
- anonymized info: coded and then key is destroyed so data cannot be re-linked
- anonymous info: personal identifiers were never associated with data
42
Plagiarism
- dishonest act of presenting intellectual property as an original idea
- unethical and illegal
- either quote or paraphrase
- cite original source unless its common knowledge
43
Outliers (ethics)
- defined as measurement or observation that is way out of line vs rest of data set
- can results from error in data collection, bug in software, or participant who markedly differs from other participants
- should be evaluated and discarded if bad
- should not be discarded if legitimate data
- may be considered outlier if >2-3 SD from mean
