Exam 1 Flashcards
100%
Research question
Research has selected a topic and formulated a research question.
Very specific and clear to what extent it will be studied.
What are Variables?
Variable is a measurable property that differs among entities or across time.
Variables need to be specific.
Levels of variable
Conceptual and Operational (measurement)
Conceptual Definition
Describe the theoretical meaning of a variable
Operational Definition
Provides a tool for quantification and measurement of a variable
Identifying variables
Identify the variables of interest. Informed by the research question and guides hypothesis. Need clearly defined IV and DVs
Independent variable
What you are manipulating. Independent of other variables
Dependent variables
What you are measuring (outcome) Affected by changes in an IV
Forming the Hypothesis
The hypothesis is the expected results of the study. It is based on theory and/or previous research. Hypothesis must be testable. Two types
Null hypothesis
The prediction that there are no differences among treatments or no relationship among variables. Denoted by H0
Alternative Hypothesis
Prediction that there are differences among treatments or there is a relationship among variables. Denoted by H1. Directional or Non-Directional
Directional hypothesis
Predicts specific relationship/outcome and the direction. Example: the people that come to class will do better on the final exam
Non-directional hypothesis
There will be a difference but not sure where. Example: there will be a difference in the Mid-Term from people who come to class compared to those who don’t
Understanding the Null and Alternative Hypothesis
In scientific research, we always either: Reject the null hypothesis or fail to reject the null hypothesis or accept the alternative. We do not ‘accept’ the null hypothesis and we do not ‘prove’ or ‘disprove’ a hypothesis
data collection
need to decide on the specific procedures to gather the data to test hypotheses. consider: design, validity, reliability, sampling techniques
study designs
experimental, quasi-experiment, qualitative
Study validity
internal and external
internal validity
extent to which the results of a study can be attributed to the treatments used in the study
external validity
the generalizability of the results of a study. we want to infer our sample findings to the population
test validity
degree to which a test/instrument measures what it is supposed to measure
reliability
refers to consistency/repeatability of a measure. a test cannot be valid if it is not reliable. however, a test can be reliable but not valid
population
a large group of people from which a sample is taken. estimate population characteristics from a sample. larger samples more representative or generalizable. sample type, too specific: lose generality
probability sampling
every person has an equal probability of being selected. includes: random selection, systematic sampling, and stratified sampling
non-probability sampling
no assurance is given that each item has a chance of being selected. includes: convenience sampling and purposive sampling
random sampling
each member of the population has an equal chance of being selected
random sampling steps
assign a number to each member of the population. use a random number table or software to select numbers
random sampling benefits
every case in the populate has an equal chance of selection
systematic sampling steps
assign a number to each member of the population. choose a random starting point. from that point, choose every Kth person.
systematic sampling benefits
faster than random sampling
systematic sampling drawbacks
possible systematic error
stratified sampling
the population or sampling frame is divided and grouped on a characteristic before random selection takes place
Stratified sampling steps
the population is divided on some characteristics. sample is then randomly selected proportionally from the different strata. this approach can be particularly important if there is a certain characteristic that needs to be represented in the sample
convenience sampling
a process of drawing a sample from groups of people that are familiar or convenient. clinicians might ask patients to participate in their studies. kin profs might ask students, coaches, teams
convenience sampling benefits
quick and easy
convenience sampling disadvantages
not random, not always representative
purposive sampling
involves identifying units that represent a characteristic of interest. sample is identified with that purpose in mind. serval types of purposive sampling; snowball sampling, quota sampling- if you have to split in groups 50/50
replication
study should be replicable. results should be able to be reproduced
analyzing and interpreting results
data analysis phase, interpretation phase, and communication phase
data analysis phase
analyze the data using appropriate statistical techniques
interpretation phases
compare results with the hypotheses on the basis of your theory. do your results support the hypotheses, theory/ previous research?
communication phase
prepare written and/ or oral report for publication/ presentation
experimental research
intervention/ treatment introduced. attempts to provides explanations. allows causal inferences
non-experimental research
no intervention/ treatment introduced. often trying to describe. hearing their story. not manipulating the variable
qualitative research
based on the generation and interpretation of non-numerical data. three main sources of qualitative data: open-minded interviews, direct observation, Witten documents
defining features of qualitative research
well-suited for understanding peoples meaning of experience. data collected in participants natural setting. the researcher is an integral part of the research process. researchers play an integral role in generating data. use of the term data generation rather than data collection. emphasizes the manner in which researchers and participants work together to generate data
qualitative research types
five common types of qualitative research.
1.Narrative 2. Ethnography 3.Phenomenology 4. Case study 5. Grounded theory
narrative
stories are used to bring understanding or meaning to the lived experience of individuals. various specific forms of narrative inquiry; life history, oral history. stories typically generated via in-depth and unstructured interviews
ethnography
seek to understand cultures or a cultural group. specifically, the behaviours, values, and beliefs. data generation primarily through participant observation. interviews and documents may also be used
Phenomenology
purpose is describe a lived experience of a phenomenon from participants perspectives. use multiple methods to collect data. uses bracketing. a method used in qualitative research to mitigate the potentially deleterious effects of preconceptions that may taint the research process. research goes into the field with no preconceived attitudes, beliefs or opinions themselves.
case study
study of the complexity and distinctiveness of a case with important circumstance. cases of interest: people, team, event, organization, or community. data generation through interviews, observation, visual methods
grounded theory
purpose in theory development. relies on constant comparative method to develop theory. simultaneously collect and analyze data. examine the data against each other in an effort to identify similarities and differences. continue this until saturation is met, until no new themes are emerging. data generated via interviews, in dept interviews
sampling
purposeful sampling. researchers may choose to identify a specific form of purposeful sampling. extreme case sampling. maximum variation sampling. snowball sampling
data generation
interviews are the most common method for generating data in qualitative research. qualitative studies often use more than one method of data generation
interviews
one on one. groups interviews need group rules. important to blood and maintain rapport, making sure they feel comfortable. typically comprised of three main phases- introduction, questioning and closing. interviews are often recorded and then you transcribe them
structured interviews
same questions, same order, same wording
semi-structured interviews
list of questions but flexibility to ask additional questions
unstructured interviews
concepts and ideas you want to touch upon buts its more like a conversation
observation
going into the natural setting to try to better understand the topic of the study. spending a prolong amount of time in a setting. field notes are taken throughout the observation and are focused on what is seen
written documents
various types of written documents can be used to generate data in qualitative research . public documents, written documents from participants, medical records, memos
trustworthiness, four aspects
one method to evaluate qualitative research. four aspects of trustworthiness; truth value, applicability, consistency, neutrality
truth value
credibility of the study, confidence in the “truth” of the study findings for participants
applicability
transferability of study, forming understanding that may be relevant to other contexts or participants
neutrality
findings are based on participants meaning and experience, findings are not a mere function of researchers’ biases, interests and perspectives
consistency
dependability of a study, seek to understand variability of study findings, understand unique experiences
evidence of trustworthiness
audit trail, member checking, peer debrief, present negative or discrepant information, prolonged engagement, purposeful sampling, research flexility, rich thick descriptions, triangulation
audit trail
researchers maintain detailed description of entire research process. someone external to study examines various components of study
member checking
study participants review data or study interpretations. opportunity to add, alter, delete
peer debrief
researchers pushed by professional “peer” to critically reflect on study
present negative or discrepant information
presenting information that counters main study findings. highlights opposing views and unique experiences
prolonged engagement
sustained time spent with participants “in the field”
purposeful sampling
recruiting information-rich participants who can best inform research question
researcher flexility
researchers position themselves. reflect on biases, experiences, and background to consider how these shape research
rich, thick descriptions
collecting through descriptive data. presenting findings in a rich manner, quotes
triangulation
crosscheck study findings and interpretations. use variety of data sources, perspectives and methods
usefulness of qualitative research
understanding the individual experience. understanding the subjective experience. problem: generalizability
what are statistics?
a set of mathematical processes that deal with collecting, organizing and interpreting quantitive data. descriptive techs, correlation techs, differences among groups
inferential statistics
generalization of results to some larger population
types of variables
continuous, and discrete (categorical)
continuous data
attributes of characteristics that can theoretically have infinitely fine gradations. can be expressed as fractions
discrete data
variables in which there are no intermediate values possible. cannot be expressed as fractions
levels of measurement
nominal, ordinal, interval, ratio
nominal variables
classify object or events into categories. assuming numbers to classify characteristics into categories. the assigned values are simple labels. when there are only two levels of nominal variable this is referred to as dichotomous or binary variable and the number associated does not matter
ordinal variable
variable that as categories that are ordered. differences between categories is meaningless.
interval variables
equal intervals on the scale. distance between any two points are equal. cants say one is twice the other. no absolute 0, the zero point is arbitrary
ratio variable
equal intervals on the scale. distance between any two points are equal. ratios. absolute zero point
central tendency
values that describe the middle characteristics of a set of data. mean, median and mode
mean
the arithmetic average of a variable in a group or sample. mean= sum of all sources/ # of sources
median
middle value in a set on ordered numbers
mode
most frequently occurring number
variability
an index of how the score vary or disperse. measures of variation; range, variance, standard deviation
standard deviation
spread of score above and below the mean. the larger the SD, the more variability you have in you sample. ever time you report a mean you MUST report a SD
normal distribution
in normally distributed data, the standard deviation is approximately 3SD above and below the mean
kurtosis
leptokurtic curves and platykurtic curves
leptokurtic curves
limited to variability, positive kurtosis, more peak
platykurtic curves
large variable, negative kurtosis, more spread
skewness
the direction of the hump of the curve and the nature of the tails of the curve. positive or negative
positive skewness
the hump of the curve is shifted to the left with the long tail to the right
negative skewness
the hump of the curve is shifted too the right with the long tail to the left
categories of statistical tests
parametric or nonparametric
parametric
normal distribution of population, equal variances, independent observations, large sample size, DV has to be measured on at least interval or ratio scale
nonparametric
distribution is not normal, DV cane categorized or ordinal, above assumptions of parametric stats are not met, sample size requirements are less stringent than the parametric tests
Alpha level
level of significance, probability value (P value), power. alpha is the predetermined crucial value for significance. usually 0.05. if the study was done 100 times, the decision to reject the null hypotheses would be wrong 5% of the time
p-value
the actual critical value obtained from data analysis. it indicates the probability that the result could be a function of chance. compared to your predetermined aphelia value. p-value is considered significant if it is less than or equal to alpha. the p-value for a statistical test should be reported as the exact number
power
the probability of rejecting the null hypothesis when the null hypothesis is false.
type 1 error
every time something is compared we increase our chances of making a Type 1 error
type II error
accepting the null hypothesis when the null hypothesis is false. false negative FN or beta error. it is the result of. a lack of power. caused from under sampling, exaggerating the estimate of an effect size which leads to under sampling
significance vs. effect sizes
statistical significance just tells us if the differences are unlikely to have occurred by chance. the ES refers to the practice significance or how meaningful results are. the magnitude/ change that happens to your DV that an be attributed to you intervention/ treatment. an ES of 0 is no difference
Relationship between power, ES, alpha, and sample size
sample size is influenced by alpha level, power, and the ES. if power and the ES stays the same but a more stringent alpha is used a greater number of particles are required to detect a significant difference
Planning research
alpha level is predetermined. effect size: should be based on prior research. power generally sat at a value of 0.8. sample size: you want a balance between statistical and practical significance
experimental research
A treatment is intentionally introduced to see the effect on the dependent variable
true experimental design
most powerful means of generating new knowledge. the only quantitative design used to identify a cause and effect relationship. typically conducted in a lab within a controlled environment. this level of control helps to ensure the study’s internal validity. external validity is not as easy to claim
true experimental design includes
an experimental group, a control group
group membership
using a method of random assignment
Randomized control trail (RCT)
is a type of true experimental research design. pre and post test design. post test only design
criteria to establish cause and effect
- the cause must precede the effect. 2. the cause and effect must be correlated 3. the correlation between cause and effect cannot be explained by another variable
derivates of true experimental designs
between subjects designs and within subjects designs
between subjects designs
separate groups of participants, each being tested by a different treatment/conditions. each group is given a separate treatment
within subjects designs
all participants are exposed every treatment to every treatment or condition. multiple testing sessions.
common characteristics of an experiment
random selection from the population of interest. random assignment into groups. blinding. treatment-control comparisons. pre-test assessment (sometimes). post-test assessment
random sampling
any member of the population of interest has an equal chance of being selected
random assignment
every participant must have an equal probability of being assigned to either the control or experimental group. helps to ensure groups do not differ at the beginning
blinding
open label. single blind. double blind. triple blind
open label
no blind, everyone is aware of treatment allocation
single blind
either participants or tester unaware of the treatment allocation
double blind
neither participants nor tester knows which treatment the participants are receiving
triple blind
participants, experimenters and investigators are all unaware, reduces placebo effect
treatment-control comparisons
comparing treatment to control provides information about how treatment improves over time. make sure the treatment is better than no treatment
baseline evaluations
pre-test to establish baseline. can then compare baseline to post-test. complications: practice effects
internal validity
extent to which the results of a study can be attributed to the treatment used in the study. ability to say that any difference in the DV is a result of the IV. true experiment designs strives for high internal validity
external validity
the generalizability of the results of a study. external validity is difficult to claim in true experimental designs. the real world does not allow control.
validity tradeoff
researchers often attempt to maximize internal and external validity but… nearly impossible to have both. impossible to have both high internal and external validity. impossible to have both high internal and external validity
quasi- experimental designs
studies that are “sort of” experimental in design. no randomization to groups. low on internal validity but higher in external validity. basic formula for a quasi- experiment study; people are studied in real-world settings, an independent variable in introduced or manipulated, there is a dependant variable (the effect) that is measured
treats to internal validity
testing, instrument accuracy, experiment drop-out/ mortality, selection bias, placebo effects, statistical regression,
treats: testing
the effects of one test on subsequent administration of the same test. pre-testing can provide a practice effect
instrument accuracy
changes in calibration, inappropriate use of entrustment(s), different techniques used between first and follow-up measure, between researcher differences
experimental drop-out/mortality
participants dropping out or leaving the study
selection bias
forming groups without random assignment. may cause groups to be biased on some or many characteristics. groups could be different due to treatment
placebo effect
participants reacting in a way they expect they would react
statistical regression
groups selected on the basis of extreme score are not as extreme on subsequent testing. scores change in the direction towards the mean
controlling treats to internal validity
randomization, blind setups, calibration of instruments
treats to external validity
setting and treatment interaction, selection and treatment interaction
setting and treatment interaction
studies conducted in highly controlled environments. difficult to know if the same outcome would be found in real world settings
selection and treatment interactional
unique characteristics of participants makes the treatment only effective for them. can’t generalize results to people who do not have the same characteristics
controlling treats to external validity
conduct experiments on groups that have different characteristics. conduct experiments in new settings. select participants that represent a larger population
Epidemiology Definition
is the study of the frequency, distribution and determinants of health nd disease in populations, and the application of this study to control health problems
what do epidemiologists do?
search for the cause of disease. identify people who are at risk. determine how to control or stop the spread. identify new diseases that have never been seen before and causes of them. examine how and where disease outbreaks start. make recommendations to control spread or prevent future occurrences
John Snow
1813-1858
father of modern epidemiology. 1854 cholera epidemic in London England. miasma theory, “cholera caused by bad air”. skeptical of this theory so began researching, collecting data, and speaking with local residents
Snow’s work
discovered nearly all deaths occurred around broad street. used dot map to illustrate the cluster of cholera around broad street pump. advised officials to close pump that supplied water to the neighbourhood. once he pump was shut down cases of cholera diminished. later discovered that this public well had been dug 3 feet from an old cesspit that was leaking fecal bacteria
measurements
proportions and rates
proportions definition
number of health events divided by total populations. in percentages
rates
a measure of the frequency with which an event occurs in a defined population in a defined time. in rates
prevalence
the proportional of population that are affected by a disease at specific time. total number of cases (old and new) of the disease in a given population/ total number of people in that population. is a proportion, usually a percentage
Incidence
the number of new cases of a disease that occur during a specific period of time in a population at risk for developing the disease. number of new cases of a disease in a population during a specific period of time/ total number of propel at risk of developing the disease in that population during the same period of time. is a rate. expressed as n case per N population
three points apply when calculating incidence
- pre-existing case of the disease cannot be included in the numerator 2. people who already have the disease or who are incapable of having the disease are excluded from the denominator 3. the calculation is based on the population that all at-risk individuals are observed for an period of time
increase in prevalence
due to:
increased incidence, decreased mortality, increased duration of disease (chronic disease), better or increased screening of disease
incidence vs. prevalence
harder to measure new cases (incidence) than total cases (prevalence). Canada has a good estimate of cancer incidences because of cancer registries. poor estimates of diabetes incidence because new cases are not often reported to central depository
observation study
gather dat by observing events
experimental study
researcher imposes an intervention
why observational studies?
expose is too dangerous (unethical), exposure is too expensive, experimental would take to long to get result
cross sectional studies
data collected from a sample at a specific point in time. disease and exposure determined simultaneously. participants are then classified based in their exposure and disease status at that point in time
cohort studies
a cohort is assembled with no of these individuals having the outcome/disease. once in the study, people in the cohort are classified according to their exposure. all members are then observed over time to see which of them develop the outcome/disease
cohort definition
a group of people who have something in common when they are first assembled and then who are observed for a period of time
famous cohort study
framingham heart study
framingham heart study
well-recognized influence of hypertension and hypercholesterolemia on the development of coronary heart disease is confirmed. a “risk score” = using standardized variables to predict. 10 year cardiovascular disease risk. a gold standard in medicine for assessing need for treatment
smoking causing lung cancer
follow forward 187,766 men. link proven “beyond a reasonable doubt”
case control studies
participants are grouped on disease status: case= someone affected by the disease/ outcome. control= someone not affected by the disease/ outcome. retrospectively investigate whether participants were exposed to the factor of interest. case and control participants are then classified as either exposed or not exposed
famous case control
disease carcinoma of the lung. 709 lung carcinoma patients. 709 non cancer control patients. patients interviews about smoking. 688 lung carcinoma patients were smokers
cross selectional advantages
can target high risk populations, relativity fast and inexpensive
cross sectional disadvantages
temporal issues cannot determine whether exposure preceded disease
cohort advantages
know temporal relationship, good for rare exposures
cohort disadvantages
not useful for rare outcome, often ling follow-up times, expensive
case-control advantages
good for rare diseases, relatively inexpensive and fast
case control disadvantages
not good for rare exposures, most susceptible to bias
Independent sample
samples chosen are independent, no relationship between those chosen. Subjects chosen in one sample have no bearing on who is chosen in the other sample. Example; right handed people vs. left handed people
Dependent sample
the subjects chosen in one sample depend directly on who was chosen in the other sample. Examples: test-retest experiments, brothers and sisters, mothers and daughters
Choose which test: suppose we are interested in determining whether men watch more TV than women in one week.
independent samples T-test
Independent samples T-test
compares the means between two independent groups on the same continuous dependent variable.
which test answers the question: is there a significant difference between the two sample means?
Independent samples T-Test
independent samples t-test assumptions
- DV is continuous (interval or ratio scale). 2. IV is categorical consisting of two independent groups (dichotomous). 3. independent of observations (participants can only be in one group) 4. no signifiant outliers. 5. homogeneity of Variance (variances pf two populations are equal
Homogeneity of Variance
Assumes the samples were drawn from two populations that have approximately equal variance. tested is SPSS via levees test of homogeneity of variance.
when do you not record an ES?
if your result is not significant
Mann Whitney U statistic
nonparametric equivalent to an independent t-test. this test is used to: 1. compare two groups on a DV that is ordinal. 2. Compare two groups on a DV that is continuous but does not meet the assumption of normality
which test:Suppose 10 dancers are given a jump-and-reach test. Then they take part in 10 weeks of dance activity that involves leaps and jumps 3 days/week. The dancers are then given the jump-and-reach test after the 10 weeks.
dependent samples t-test
Dependent samples t-test
compares the means of two related groups to determine whether there is a statistically significant difference between these means. often the same participants are tested more than once. Therefore, the same participants are in both groups.
Possible relationships
one group of participants is test twice on the same variable. Ex; comparing baseline and post-test score of a single group after undergoing an intervention. A single group compared on two continuous outcomes. ex. comparing whether students miss more class because of oversleeping or illness.
Dependent samples t-test assumptions
- DV is continuous (measured on a interval or ratio scale). 2. IV is categorical consisting of two ‘related groups’, 3. No significant outliers in the difference between the two related groups. 4. The distribution of the differences in DV between the two related groups should be normally distributed.
Wilcoxon Signed Rank Test
Non-parametric test equivalent to dependent samples t-test. this test is used when the two groups of scores are related and: 1. the DV is measured on an ordinal scale. 2. the DV is measured on an continuous scale but does not meet the assumption of normality
which test for this scenario: Suppose we are interested in knowing if Canada, France, or Australia won more gold medals in the 2016 Rio Olympics.
Analysis of Variance (ANOVA)
ANOVA is used…
to determine whether there are any statistically significant differences between the means of two or more independent groups. extension of an independent samples t-test.
Why not use several independent samples t-test?
increase the risk of Type 1 error, too time consuming
ANOVA is…
an omnibus test statistic. tells you whether groups means are different but not which pairs of means are different
post-hoc test (ANOVA)
post-hoc tests are used to determine which pairs of groups means are significantly different. many types of post-hoc tests to control for type 1 error. ex. bonferonni, tukey
ANOVA assumptions
- DV is continuous (interval or ratio) and is normally distributed. 2. IV is categorical consisting of two or more independent groups. 3. independence of observations. 4. no significant outliers. 5. homogeneity of variance, tested using Levene’s test in SPSS
Reporting ANOVA results
include statement about Levens test and P-value. P-value. F-value. The degrees of freedom between groups. The degrees of freedom with groups. and effect size when ANOVA is significant
Kruskal-Wallis test
non-parametric equivalent to an ANOVA. this test is used to: 1. compare 3 or more independent groups on a DV that is ordinal. 2. Compare 3 or more independent groups on a DV that is continuous but does not meet the assumption of normality.
Which test: Suppose we are interested in knowing whether the distance that children throw a ball using an overhand pattern increases over time. We decided to measure the children’s distance thrown every 4 months during a year. Thus, each child’s throwing distance was assessed 3 times.
repeated measures analysis of variance (RMANOVA)
RMANOVA used..
to determine whether there are any statistically significant differences between the means of two or more related groups. extension of dependent t-test. RMANOVA is an omnibus test statistic.
omnibus
shows if there is a difference but you have to look at the table to see where the difference is
post-hoc test (RMANOVA)
RMAVOA only tells you whether there is a difference, but not specifically what pairs of means scores are significantly different. post-hoc tests are used to determine which pairs of mean score are significantly different. many types of post-hoc tests to control for Type 1 error
RMANOVA assumptions
- DV is continuous (interval or ratio). 2. IV is categorical consiting of two or more related groups. 3. no significant outliers in the related groups. 4. the distribution of the DV in the two or more related groups should be normally distributed. 5. Sphericity, variances of the differences between all combinations of related groups must be equal. Measured using Mauchly’s test is SPSS. Mauchly’s test should be insignificant to meet the assumption of sphericity
what test do we look at mauchly’s test for sphericity was significant
you would then look at and report the Greenhouse Geisser row
Friedman
non-parametric test equivalent to RMANOVA. This test is used when the three (or more) groups of scores are related (dependent) and: 1. the DV is ordinal. 2. the DV is continuous but does not meet the assumption of normality
correlation
a statistical test to determine the relationship (association) between two or more variables. no cause and effect assumed. one variable is not a direct cause of another
Correlation coefficient (r)
quantitative value of the relationship between two or more variable
Correlation coefficient ranges
between -1 and 1. -1 indicates a perfect negative relationship (inverse relationship) +1 indicates a perfect positive relationship. 0 indicates no relationship
correlation coefficient tells us…
direction of the relationship (positive, negative, no relationship). strength of relationship between two variables, it does not indicate the cause of that relationship
person moment correlation
used to determine correlations between two continuous variables. Has one outcome (DV) and one predictor (IV). every participant has two scores
assumptions of Pearson moment correlation
- two variables are continuous. 2. correlated variables must have normal distribution. 3. no significant outliers. 4. linear relationship between X and Y. 5. Homoscedasticity, correlated variables must have equal variability
Coefficient of Determination: Effect size
(r2), the squared correlation coefficient. represents proportion of shared variance.
Using correlation for prediction
one purpose of correlation is prediction, the stronger the relationship between two variables, the more accurately you can predict one from the other. we do not encounter perfect relationships in the real world, there is always some error
simple linear prediction (regression)
statistical method used to predict a dependent (or outcome) variable (Y), from one independent (or predictor) variable, (X). if two variables have a relationship between them, we can use that information to make a prediction.
fitting a regression line
to be precise, you would need to calculate the slope and y-intercept and calculate Y-predict using the equation Y-predict = a+bx
Spearman Rank-order correlation
non-parametric test. this test is used to examine a relationship:
issues with research
publication bias, replication bias, fraus and misconduct, pseudoscience, media influence
Publication bias
tendency for statistically significant findings to be published over nonsignificant findings
publication bias happened due to
researchers not submitting their studies for publication, particularly if they have null results. journals being more likely to publish significant results than studies without statistical significance
problems causes by publication bias
can lead to the overestimation of how effective an intervention is. leads to skewed perception of a field of research. limits the replicability assumption of science. due to pressures to publish, this can led researchers to p-hack
replicability crisis
concerned with the large number of scientific studies that cannot be reproduced
replication
using the same methodology as a prior study but using different participants in order to confirm their results
why can’t studies be reproduced
inappropriate or incorrect SS analysis. insufficient sample size. publication bias
fraud and misconduct
violations of the standard codes of scholars conduct and ethical behaviour research. fraud is intentional and misconduct in not
Dr Wakefield
dumbfuck with the autism study
Brain Wansink
professor at Cornell, many problems with data reporting, committed p-hacking. he refused to let failure be an option. 15 studies retracted
science
is the discovery of knowledge
pseudoscience
the pursuit of knowledge or collection of related beliefs about the world mistakenly regarded as being based on scientific methods
dangers of pseudoscience
opportunity cost, blocks scientific thinking, direct harm