APA2180 Final Exam
What are raw scores
Data typically consist of a set of values or raw scores taken from measurements
What is descriptive epidemiology?
Describes relationship between basic characteristics and disease states
What are descriptive statistics?
Describing a known data set
What is internal validity?
Did the treatments (IV) cause the change in the outcome (DV)
What is simple randomization?
Each individual is randomized to one treatment group
Explain maturation
Effect on experimental results caused by experimental subjects maturing or changing over time
Explain reactive effects of testing
Eliminate pretest/familiarization session
What is standard deviation?
Estimate of variability of scores around mean
Explain history
Events occurring at the same time as study not part of the treatment
What is systemic sampling
Every nth member selected
Explain external validity
Generalizability of results
What is stratified random sampling?
Group divided on a characteristic before random sampling
What is block randomization?
Groups of individuals are randomized to a treatment group
What are strata?
Homogeneous subgroups of members in a population
What are post-hoc explanations
Justifying sample representativeness when no random selection
What differentiates quasi-experiments from true experiments?
Lack high degree of control that is characteristic of true experiments
What are the extraneous variables that jeopardize internal validity?
MRS. SMITHE Maturation Regression to the Mean Selection bias Selection-maturation interaction Mortality Instrumentation Testing History Expectancy
Explain normal distribution
Most used statistical distribution Often arises naturally Basis of statistical tests Mean, median and mode at the same point
What are the possible options for the control group?
Nothing, placebo, usual care
Explain the basic difference between odds ratio and relative risk
OR is a ratio of two odds whereas the RR is a ratio of two probabilities
Explain multiple-treatment interference
One treatment may influence the next treatment
What is a factorial ANOVA
An ANOVA in which there is more than one independent variable
What is statistics?
An objective way of interpreting a collection of observations
What is a repeated measures ANOVA?
Analysis of scores for the same individual on successive occasions
List and explain the two threats to validity in analytical study designs
Bias Systematic deviation of a calculated (estimated) value from the true value Information from • Recall • Observer/Interviewer • Misclassification Selection Confounding factors Extraneous factors
Explain the two types of randomized trials
Clinical trials - Individual level Community trials - Community level
What are the two types of analytical designs?
Cohort studies and case-control studies
Explain ecological designs
Compares groups using pre-existing data for both exposure and outcome to compare and contrast rates of disease by characteristics of an entire population
What are the types of raw scores
Continuous and discrete
How do you formulate the research question?
PICOT+: Population Interventions/exposures Comparison/control groups Outcomes Time period/timing +study design
What are the categories of statistical tests?
Parametric and nonparametric
Election selection-maturation interaction
Passage of time influencing groups differently
List the two types of external validity
Population and ecological
List and explain the statistical assumptions for any type of T-test or ANOVA used
Population assumptions: Data are drawn from normally distributed populations, Variance in each group is similar Sampling assumptions: Data represent random samples from the population, Samples are independent of each other
Explain interaction effects of testing
Pretest may make participants sensitive to treatment
What is the key to controlling threats to internal validity
Random assignment
What is the key to controlling most threats to external validity?
Random selection
Explain selection bias
Sampling bias that results from differential selection of respondents for the comparison groups - Non-random participant selection
Explain how to control threats to external validity
Selecting from a larger population Participants Treatments Situation
What is convenience sampling
Selection based on availibility, volunteers
Explain reactive effects of experimental setting
Setting constraints may influence generalizability Hawthorn effect: performance changes when aware of being studied
Explain t-test and power
T is the true variance (or real difference between the means) divided by the error variance (or variation about the mean)
What is a main effect in a factorial ANOVA?
Test of each independent variable when all other independent variables are held constant
Explain a t-test between a sample and a population mean
Test of null hypothesis - states that there is no difference between the sample mean (M) and the population mean (μ)
What are degrees of freedoms?
The number of independent pieces of information that went into calculating the estimate
Explain relative risk
The extent to which is it more/less likely an outcome will occur in an exposed vs. unexposed group Calculation: (incidence in exposed group) / (incidence rate in unexposed group) Meaning: RR = 1 means no relationship RR > 1 positive relationship - Those exposed are X time more likely to have the outcome RR < 1 negative relationship - Those exposed are X times less likely to have the outcome
What is population validity
The extent to which the results can be generalized from the experimental sample to a defined population
What is ecological validity
The extent to which the results can be generalized from the set of environmental conditions in the experiment to the other environmental conditions
What is random assignment
Unbiased group selection
What is random sampling?
Unbiased sample
What are systematic reviews?
Use a clear and specific question to identify, appraise, and synthesize all the empirical evidence that meets pre-specified eligibility criteria to answer a given research question
What does a post-hoc test show us?
Where the differences that cause a significant F lie
Explain the ordinal measurement scale
o 2nd level, magnitude only o Characterized by ability to rank order on the basis of an underlying continuum o No common unit of measurement o Eg: class ranks, place of finish in a race
Explain the interval measurement scale
o 3rd level, magnitude + equal intervals o Data having known and equal distances between score units o Arbitrary zero point o Eg: temperature on Fahrenheit scale
Explain the ratio measurement scale
o 4th level, combines all 3 qualities o Possesses same properties as interval data o Does have a true zero point o Eg: percentage
What is a parameter and its mean?
o A characteristic of a population o μ (mu) is the mean of the population parameter
What is a statistic and its mean?
o A characteristic of a sample o M (mean) of the sample statistic
What is a meta-analysis ?
o A comparison of studies using similar methodology and looking at similar relationships o Uses results of individual studies to produce an overall statistic called a point estimate or summary statistic
Explain alpha
o A level of probability set at the beginning of the study o AKA level of significance o Used to avoid type I error o If P-value is < alpha then you can reject null (difference did not occur by chance)
Explain odds ratio
o A measure of association between an exposure and an outcome o Odd of an outcome in an exposed group (cases) vs odd of an outcome in an unexposed group (control)
What is central tendency + give some examples?
o A single score that best represents all the scores o E.g. mean, median, mode
Explain review of systematic reviews or meta-analyses
o AMSTAR o Assessment of Multiple Systematic Reviews is a tool consisting of 11 items and has good face and content validity for measuring the methodological quality of systematic reviews
Explain type II error
o Accept the null hypothesis, when in fact it should have been rejected o Probability is equal to beta (β)
How do we choose a sample?
o Best to take a random sample - each person has an equal chance of being selected o Balance between sample size and feasibility Use this data to estimate the population parameter (characteristic of the entire population) - This information is called a statistic
Explain instrumentation
o Calibration and test reliability o Halo effected (previous knowledge can influence an observer's evaluation)
List the methods of randomization
o Coin toss o Pulling numbers out of a hat o Random number list o By telephone o Online random allocation (numbers) o Sealed envelopes containing allocation numbers (carbonized systems)
Explain how correlation is used for prediction
o Compare predictor to criterion; The stronger the relationship, the more accurate the prediction o A prediction equation is developed, and a line of best fit is formed
What are the 6 criteria for evaluating scientific literature
o Consistency - Do studies show the same trend? o Strength - How strong is the relationship? o Appropriately sequenced - Does the exposure predate the condition? o Biologic gradient - Is there a dose response? o Plausibility - Is there a biological/physiological/biochemical rationale to explain the relationship? o Experimental evidence - Have randomized-controlled trials been done to measure the effect?
What are the three ways to measure the strength of an association?
o Contingency table o Relative risk (RR) o Odds ratio
What are two types of descriptive epidemiology
o Cross-sectional designs o Ecological designs
What are the 3 types of epidemiological study designs?
o Descriptive o Analytical o Experimental
What are the classifications of statistics?
o Descriptive o Inferential
Explain an independent t-test
o Determines whether the means from two separate groups are significantly different o Effect size determines the strength or magnitude of the difference in the mean scores
Explain the distribution, determinants, and application of epidemiology
o Distribution Frequency •Prevalence: # of existing cases •Incidence: # of new cases •Mortality rate: Death rate Patterns: person, place, time o Determinants: defined characteristics associated with change in health o Application: translation of knowledge to practice
Explain nonparametric tests
o Distribution is not normal o Nominal or ordinal or non-normally distributed sample - Small sample size
What are confidence intervals?
o Effective technique to interpret a variety of statistics, such as means, medians, and correlations o Also used in hypothesis testing o A C.I. provides an expected upper and lower limit for a statistic at a specified probability level (usually 95% or 99%) o Based on the fact that any statistic possesses sampling error o A C.I. provides a band within which the estimate of the population mean is likely to fall
Explain the purpose of randomizing
o Equipoise o Internal validity o Tends to produce comparable groups, measured and unmeasured, known and unknown prognostic factors and other characteristics of the participants at the time of randomization will be, on average, evenly balanced
What is a one-way ANOVA?
o Extension of the independent t-test o Evaluates the null hypothesis among two or more group means on one dependent variable o F-ratio is calculated
What is the F-ratio?
o F = true variance / error variance o A ratio of the true variance (variance between groups) over the error variance (variance within groups)
Explain kurtosis distribution
o Flat/peaked point o Leptokurtic (higher values) → positive → more homogenous (similar) o Platykurtic (lower values) → negative → more heterogenous (different) o Mesokurtic (normal distribution) → provide a useful indication of a departure of normality, can be useful for detecting outliers
How can we evaluate whether heterogeneity exists?
o Forest plots - do CI intervals overlap o Statistical test o Subgroup analysis
What are the stages of a systematic review?
o Formulate the research question o Register the protocol o Identify relevant literature o Collect the data o Assess the quality of studies o Analyze and summarize the evidence o Interpret the findings and publish o Update the review
Explain what you look at when organizing and graphing scores?
o Frequency distribution o Graphing techniques ie. Histogram, Scatterplot o Normal curve ie. Bell-shaped curve, Skewed distribution
What is meant by grading the evidence
o Grading of Recommendations Assessment, Development and Evaluation (GRADE) o Summarizing evidence and rating its quality o Used to develop conclusions and recommendation
Explain cohort studies
o Group of individuals with a common feature identified and followed over time to document outcome of interest - Usually disease-free at baseline o Compare those who developed outcome of interest vs. those who did not o Can establish causality, but are costly
What does post-hoc mean
o Groups are not randomly selected, but you still want to show the sample is similar (homogeneous) to the population o Just because the groups do not differ in one parameter, does not mean they are the same
What is the focus of systematic reviews?
o Health interventions o Test validity and reliability o Public health interventions o Social interventions o Etiology/disease causes oAdverse effects o Economic evaluations
What threats to internal validity does the non-equivalent control group potentially eliminate, if the groups are comparable to begin with?
o History o Maturation o Testing o Instrumentation o Regression
Explain meaningfulness and how it is calculated
o Importance or practical significance of an effect or relationship o Can be determined by effect size (ES) - Represents the standardized difference between two means, tells us how big the difference is o Formula: ES = (M1-M2) / sd o ES of 0 is no difference, 0.2 is small, 0.5 medium, and 0.8 large
Explain testing
o In before-and-after studies, pretesting may sensitize subjects when taking a test or when taking it a 2nd time (familiarity) o May cause subjects to act differently than they would have if no pretest measures were taken
Explain the uses for inferential statistics
o In descriptive research when making comparisons between groups o For determining differences between groups in experimental research o To evaluate effects of an independent variable on a dependent variable
List the threats to external validity
o Interaction effects of testing o Interaction of selection bias & treatment o Reactive effects of experimental setting o Multiple-treatment interference
Explain the values for alpha
o Liberal - .1 (less critical) o Common - 0.05 o Conservative - 0.01 (critical issue
Explain the nominal measurement scare
o Lowest level, none of the 3 qualities o Classifies objects in accordance with similarities and differences with respect to some property o No hierarchy of scores o Eg: names, lists of words
Explain type I error
o Made when the researcher rejects the null hypothesis when in fact the null hypothesis is true o Probability equal to the significance (alpha) level set by the researcher o Thus, the smaller alpha = lower the chance of committing a Type I error (but greater chance of a type II error)
List and explain the three qualities of measurement scales?
o Magnitude - greater than or less than o Equal intervals - possible scores are an equal distance from each other o Absolute zero - a point where none of the scale exists or where a score of zero can be assigned.
What is multiple regression?
o Making a prediction from two or more predictors (IVs) o More accurate than linear regression because more information is used in making the prediction
Explain cross-sectional designs
o Measure outcome at a single point in time o Compares individual (rather than groups) o Can 'control' for confounding variables using statistics or matching groups based on variables o Outcome and exposure measured at same time - Temporal sequence is not known o Cannot establish longitudinal relationship
What are the scales of measurement?
o Nominal o Ordinal o Interval o Ratio
What are the two kinds of quasi-experiments?
o Non-equivalent control group o Interrupted time-series design
Explain parametric tests
o Normal distribution o Equal variances o Independent observations o Ratio or interval scale
What is a one group pre-test post-test ?
o Observation₁ → Treatment → Observation₂ o Can observe if change in performance occurred but cannot say why
What are the three types of pre-experimental designs?
o One shot study o One group pre-test post-test o Static group comparison
Explain protocol development for a systematic review?
o Outline question being addressed o Inclusion criteria o Management of review process o Health problem and intervention under investigation o How outcomes will be measured, and appropriate study designs
What is a non-equivalent control group design?
o O₁ → T → O₂ o O₃ → O₄ o Can assess amount of change due to treatment o Control group is "like" the treatment group o Chosen from same population o Pre- and post-test measures obtained for both groups so similarity can be assessed
What is a time series design?
o O₁, O₂, O₃ → T → O₄, O₅, O₆ o In T-S designs, performance is measured both before and after treatment o If there is an abrupt change in performance at time of treatment, we conclude that treatment worked o With reversal: shows the effect of adding and removing quasi-independent variables (rules out maturation and history)
What is pre-test post-test randomized group comparison?
o O₁→T→O₂ o O₃→P→O₄ o Can assess amount of change produced by treatment
Explain reporting guidelines
o PRISMA - Preferred Reporting Items for Systematic Reviews and Meta-Analyses o An evidence-based minimum set of items for reporting in systematic reviews and meta-analyses
What are the sources of weakness of the solomon four group design
o Practicability - difficult to conduct two simultaneous experiments o Difficulty location more subjects of the same time
What are the three types of experimental designs?
o Pre-experimental o Quasi-experimental o True-experimental
What are continuous raw scores?
o Progression from smallest to largest o Measurement possible at any point o May be expressed as a fraction o Eg. Height, weight, temperature, strength
Explain the purpose of epidemiological methods
o Quantify the magnitude of health problems o Identify the factors that cause disease o Provide quantitative guidance for the allocation of public health resources o Monitoring the effectiveness of prevention strategies using population-wide surveillance programs
List the ways to control threats to internal validity
o Random assignment or double-blind setups o Reactive effects of testing o Instrumentation o Limit experimental mortality o Placebos
What are the ways to sample?
o Random selection o Stratified random sampling o Systematic sampling o Random assignment o Convenience sampling o Post-Hoc explanations
What are the types of true experimental designs?
o Randomized group comparison o Pre-test Post-test Randomized Group Comparison o Solomon Four Group Design
Explain experimental designs
o Randomized trials o Experimental designs allow researchers to identify the effects of a specific intervention on a health outcome in a group of people (experimental group) while simultaneously monitoring changes in the same health outcome in people not receiving the intervention (control or comparison group)
Explain random assignment/double blind setups
o Real randomization o Matched pairs (not matched groups) o Randomizing treatments of counterbalancing
Explain correlation
o Relationship between two or more variables o Whether two characteristics vary in the same way o Positive correlation: As the value for one variable increases, the value for the other variable also tends to increase o A negative (or inverse) correlation: As the value for one variable increases, the value for the other variable tends to decrease
What do statistics tell us?
o Reliability of effect o Strength of the relationship o Relationships among variables o Differences among groups
Explain case control studies
o Select participants from a group with a disease (cases) and compare to those without the disease (control) - controls matched on important characteristics o Inquiry about retrospective exposure o Often used for rare diseases
What is the meaning of the coefficient of correlation?
o Significance Reliability: Likelihood of similar relationship if study was repeated o Meaningfulness Coefficient of determination (r2) Portion of total variance in one measure explained by variance in the other measure The more variance explained, the more meaningful the relationship
What are the types of ANOVAs?
o Simple (one-way) o Factorial (2x2) o Repeated Measures o Analysis of Covariance (ANCOVA) o MANOVA
What are the types of regression
o Simple linear regression o Multiple regression o Hierarchical multiple regression
Explain hypothesis testing procedures
o State the null (H0) and alternative hypotheses (H1) o Select the level of significance/probability level (alpha α) o Determine the value needed for significance o Calculate the test statistic and get p-value o Reject or fail to reject H0 (i.e. p < α?) o Draw statistical and practical conclusions
What are the benefits of the solomon four group design?
o Strong because involves conducting the experiment twice, one with pretest and one without pretest o If the results of these two experiments are in agreement, the experimenter can have much greater confidence in his findings
Explain skewness distributon
o Symmetry o Direction of the skew refers to the direction of the tail
What is randomized group comparison?
o T → O₁ o P → O₂ o Designs allows for conclusion that difference between observations are due to treatment
List the types of T-tests
o T-test between sample and population means o Independent t-test o Dependent or 'paired' t-test
What is the difference between when T-tests vs. ANOVAs are used?
o T-test: compares two mean scores o ANOVA: compares two or more mean scores
Explain analytical designs
o Testing hypotheses about causal links between exposure and outcome using observational methods
What are the three criteria for cause and effect?
o The cause must precede the effect in time o The cause and effect must be correlated with each other o The correlation between cause and effect cannot be explained by another variable - If the condition is necessary and sufficient to produce the effect, then it is the cause
What is variability + give some examples?
o The degree of difference between each individual score and the central tendency score o E.g. standard deviation, variance, range, minimum and maximum
Explain the mathematical measure for correlation (r)
o The person product moment coefficient of correlation (r) o 2 scores needed: Criterion (dependent) variable Predictor (independent) variable
What is the P-value
o The probability of obtaining a test statistic at least as extreme as the one that was actually observed, assuming that the null hypothesis is true o Small p-value (typically ≤ 0.05) = strong evidence against the null hypothesis = reject the null hypothesis. o Large p-value (> 0.05) = weak evidence against the null hypothesis = fail to reject the null hypothesis. o Report actual p-value so readers can draw their own conclusions
Explain a dependent or 'paired' t-test
o The purpose is to compare two mean scores that are related Pre/post data of one group Comparison of two groups matched on a key trait o Meaningfulness (magnitude of change) Looking at percent change can provide important information in addition to a significance value
Explain statistical significance
o The value of the calculated statistic warrants rejection of the null hypothesis o Suggests a real difference, and not one due to sampling error
What are the problems with non-equivalent control group design?
o Threats to validity due to interactions with selection may not be eliminated o Selection and maturation
What are the components of the protocol?
o Title and background o Review question(s) and objectives o Inclusion/exclusion criteria (PICOt+) o Search strategy o Data screening methodology o Data abstraction/collection o Critical appraisal/quality assessment o Data synthesis o Dissemination plan o Review team
What is a one-shot study?
o Treatment → Observation₁ o Can only say participant performed at a certain level
What is a static group comparison?
o Treatment → Observation₁ o Placebo → Observation₂ o Can observe differences but cannot say why
Explain heterogeneity
o Variability o Sources: Clinical diversity Methodological diversity
Explain regression to the mean / statistical regression
o When groups are selected based on extreme scores Ex. "you can only go up from here" o If these threats are uncontrolled, the change in the DV may be difficult to attribute to the manipulation of the IV
Explain one and two tailed tests?
o Where H₁ is directional: one tailed test o When H₁ is non-directional: two tailed test
What are discrete raw scores?
o Whole units, no fractional units o Eg. Size of family, # of schools in country Dichotomous: 2 category variables
To find a significant difference, when do you need more subjects?
o Your chosen power is high o You choose a conservative alpha o The sample statistic effect size is low
Explain statistical power
oThe probability of rejecting the null hypothesis when the null hypothesis is false - Detecting a real difference o Power = 1 - β o By tradition - 80%
How do you calculate confidence interval
observed statistic ± standard error x specified confidence level value
What are the types of randomization?
simple, blocked, stratified
What is a randomized control trial?
A planned intervention study in which each member of a study population has the same chance of receiving one or more experimental or control treatments
Explain beta
Increasing sample size will reduce beta, and increase power
What is stratified randomization?
Individuals are grouped into strata and then randomized into one treatment group
Explain expectancy
Influence of experimenters on participants
What is variance and how is it calculated?
Square of the standard deviation
Explain how to determine if T is significant
Step 1: Calculate your degrees of freedom - Df = n - 1 Step 2: Choose your alpha Step 3: Choose whether you would like a one-tailed or two-tailed test Step 4: Is your t greater than the critical t? - If yes, it is significant
What are inferential statistics?
Studying samples, and applying the results to make generalizations about the population
Define probability
The relative likelihood of an event occurring (or not occurring) relative to some other event
What is randomization?
The solution to getting two group as similar as possible is to allocate subjects using some random system, by tossing a coin maybe: heads = treatment, tails = control
Define Epidemiology
The study of the distribution and determinants of health-related events or disease in specific populations, and the application of the study to the control of health problems
Why is power important?
To decide before initiation of a clinical study whether it is worth doing, given the needed effort, cost, and in the case of clinical experiments, patient involvement
Explain interaction of selection bias and treatment
Treatment may work only on participants selected on a specific characteristic