Quantitative research Flashcards
What’s quantitative research
it is used to quantify the problem by way of generating numerical data to explain observable phenomena
what does quantitative research do/purpose
uses measure able data to formulate facts and uncover patterns in research
what’s a research design
a structured plan or blueprint which outlines how a study will be conducted. - details methods & procedures
why is a research design important
- Consider the purpose of the study
- allows hypothesis to be tested
- ethical considerations
- reduces the chance of error
- understand the conclusion which can be drawn from the study
purpose of the hierarchy of scientific evidence
shows how strong or weak evidence is
common research designs
observational = participants are observed
experimental = effect of an intervention is assessed
observational study - flow chart
no intervention -> group comparison ->
Yes (cohort study, case control) or No (case series, case study)
what is an observational study
- observational (non-experimental) studies
- find a naturally occurring experiment
- comparison of 2 or more populations that yields information about the relationship between 2 or more variables
why do we do observational studies
- gain real world insights
- ethical considerations
- can provide valuable insights in chronic health conditions
- large sample size
Experimental study - flow diagram
intervention -> experimental -> random allocation ->
Yes (randomised control trial) or No ( controlled study)
what is experimental research design
- most common type of study
- intervene by providing an intervention
- manipulate IV to see what effect it has on DV
- primary purpose is draw a conclusion about a particular procedure treatment
involved pre and post intervention measurements
independent variable IV
dependent variable DV
IV = change, control group
DV = measure, outcome
Experimental designs
- Parallel (stay in groups)
- crossover (participants receive all conditions)
randomised control trial
all participants should have similar characteristics (e.g. age, sex)
What’s blinding
method used to prevent bias by keeping certain information hidden from participants, researchers, or both
single blind
participants don’t know what treatment they are receiving (active treatment, or placebo) but researchers do.
- reduces bias in participant response and behaviour
double blind
Both participants and researches do not know who is receiving the active treatment and placebo
control group
- don’t receive any treatment
- compare effects of a given intervention with baseline measures
placebo group
- equivalent or inert treatment
- shows any observed effects are caused by treatment and not the procedure of administering the treatment
what’s bias
a systematic error or tendency that distorts findings, interpretations, or conclusions
Types of bias
Cognitive, confirmation, design, selection, data collection/messurement, analysis, survivorship, publication
cognitive bias
- ways of thinking that predispose one to favour of a certain viewpoint
what can bias effect
- can occur at each stage of the research process
- can impact validity and reliability of study findings, and misinterpretation of data
confirmation bias
interpret information in a way that confirms one’s preconceptions, while ignoring information that doesn’t support preconceptions
selection bias
both the process of recruiting participants and study inclusion
survivorship bias
focus on the individuals that have survived a certain process, intervention, while ignoring those who didn’t
publication bias
“scientific studies are more likely to be published if reporting statistically significant findings”
- positive results are more interesting
scale of measurement
- Nominal scale
- Ordinal scale
- Interval scale
- Ratio scale
Nominal scale
Simple, variables have no numerical value, have categories
e.g. gender, race, type of sport
Ordinal scale
Variables are in categories with an underlying order to their value, rank-order from high to low, intervals may not be equal
e.g. pain ratings, RPE
interval scale
ordered categories and the difference bbetween two values is meaningful, no absolute 0
e.g. temperature, time
ration scale
ordered categories, equal intervals and a true 0
e.g. age, body weight, blood pressure
parametric distribution of data
normal distribution
non parametric of data distribution
non-normal distribution
assessment of normality
- need to establish what we consider as normal
- achieved by assessing difference between mean and median
- statistical tests: Shapiro-Wilk & Kolmogorov-Smirnov
Measures of Central Tendency and when to use them
Mean - average (normally)
Mode - most common (not often used)
Median - middle (non normally)
Variance
- how scattered around the average value is
- small v = values on average are closer to the mean
- large v = measured values vary widely from the mean
measures of data spread
- standard deviation
- range
- interquartile range
standard deviation
Small SD = numbers close to average
Large SD = numbers are more spread out
choosing a measure of spread
standard deviation- normally distributed
interquartile range - not normally
Null hypothesis
statement of no difference/ no relationship, tested using statistics
Null hypothesis
statement of no difference/ no relationship, tested using statistics
inferential statistics definition
used to analyse data that involved using different statistical tests, which allows researchers to make conclusions or inferences about a given population, based on data from a sample
- descriptive statistics
- inferential statistics
- correlational statistics
- provides some valuable insight into our data
- allows u to make predictions (inferences) from that data
- allows us to tell whether a relationship exists between two variables
Type l and Type ll error
Type 1 = false positive
Type 2 = false negative
P values
P = probability of error
- low probability (better), can be more confident in finding
- cut off is 0.05, if P value is less then <0.05 the result is significant (reject null hypothesis)
Testing differences: T test
allows u to compare the means of 2 groups to determine whether there is a genuine difference between group or a product of chance
paired sample T-test
used to test the whether two samples means, collected from the same group on two separate occasions, are significantly different from each other
unpaired sample T-test
allows us to test whether sample means from different populations are significantly different from each other
What test?
same group
separate groups
same group = paired sample T-test or Wilcoxon test
separate groups = unpaired sample T-test or Mann-Whitney test
Intention to treat ITT vs Per protocol analysis PPA
ITT - all participants that where included in the study (no matter if they didn’t adhere to the protocol)
PPA - only included participants who fully complied with the study protocol
R value
expressed as a correlation coefficient
- a number between +1 (positive relationship) and -1 (negative relationship)
Pearsons vs Spearman Rank
Parametric = Pearsons correlation coefficient
Non-parametric = Spearman Rank
Both provides R value and a P value
correlation analysis
shows us how 2 variables may be related but correlation does not tell us causality
