Exam 2 Flashcards
Quantitative data analysis, community-based research, survey research
What is descriptive research?
characterizing the sample or data
- average age of students in the class, highest/lowest marks
What is inferential research?
trying to reach conclusions that extend beyond the immediate data alone
- make judgements of the probability that an observed difference between grips is a dependable one
- used when you can’t measure everyone or everything
What are the 3 scales of measurement?
Nominal, ordinal, continuous
What is a nominal scale?
unordered data, no ranking
- can’t calculate a mean
- ex. sex, favourite colour, breed of dog
What is an ordinal scale?
predetermined order among response classification
- can’t take an average
- ex. education level, placing in race
what is a continuous scale of measurement? (interval/ratio)
most precise, can get a mean
What are statistics?
an objective way of understanding our data
What 2 things do statistics inform us of?
- reliability
-significance (is this due to chance?) and meaningfulness (does this matter in the real world
What is central tendency?
single score that best represents all scores for a group of individuals
- Mean - most common measure of CT
- Median - number occurring at the midpoint of the series
- Mode - most frequently occurring number
What is a normal curve or normal distribution?
Characterized by symmetrical distributing of data about the centre of the curve
- Mean, median and mode all located at middle of curve
- frequency of scores decline in a predictable manner as scores deviate further from the centre of the curve
What is variability?
Measure of spread or dispersion of data
- how spread out the data is
- deviations from the mean
- used to know whether data is closely collected around the mean or not
What is standard deviation?
measure of spread
- how spread out the data is from the mean
- How much variability there is in the data
What is the difference between low standard deviation and high?
Low - data was closely clustered around the mean
High - data is dispersed around a larger range of values
How does standard deviation impact large data sets?
- 68% of data is within 1 SD
- 95% of data is within 2 SD
- 99% of data is within 3 SD
What is skewness?
Describes direction of hump and nature of the tails
- positively skewed - hump to left, tail to right (mean is higher than median and mode)
- negatively skewed - hump to right, tail to left (mean is lower than median and mode)
What is kurtosis?
describes vertical aspect of curve
- used to describe the degree to which scores cluster in the tails or peak of a frequency
- Leptokurtic curve - high kurtosis have distinct peak near mean and decline rapidly, have heavy tails
Platykurtic curve - low kurtosis tend to have flat top near mean
Mesokurtic curve - normal distribution
What is the basis of community engaged research? (5)
- not usually done in a lab
- Contributes to academic and community success
- community based organizations have credible, legitimate and intimate understandings of the assets, concerns, values and activities in their community
- Partnership based on both parties having valuable knowledge
- Process of planning a change, acting and observing, reflecting on processes an consequences and re-planning
What is a community?
Group of people linked by social ties who share common perspectives or interests, and may share a common geographical location
- Not homogenous and seldom speak with a single voice
Why do we use community- engaged research? (5)
- Increasing interest in health disparities in diverse communities
- Challenging health and social problems can’t be answered by experts alone
- Disappointing results with introducing evidence based practices in real world
- Increasing community desire for collaboration, not just research on a group
- Increasing funder interests in community-driven versus community-placed research
What are barriers with community-engaged research? (6)
- researchers aren’t connected to community organizations or diverse communities
- community organizations don’t realize they can collaborate with researchers (no cost)
- takes time and money
- researchers need peer reviewed papers for their career
- researchers have to give up some control to community partners
- poor communication and unrealistic expectations can lead to conflicts between academic and community partners
How to combat survey fatigue? (5)
- exhaustive vs. exhausting
- expectations - “Survey will take approx. 10 mins”
- know your audience - invested in topic makes people more likely to finish survey
- open (written answers) vs. closed (on a scale, multiple choice)
- relationship of the respondent or topic
What is important about the wording of surveys?
using grade 10 literacy levels, not using highly technical language, clear, short and easy to understand
How many questions can you ask at once on a survey?
1
How do you order questions on a survey?
Group similar topics assuming it doesn’t cause other issues
- leave questions that may cause offence to be very personal to the end (reduce bias in earlier responses or leaving survey)
- Logic flow (design rules that dictate what questions will come after one another)
What is prompting or priming in a survey?
exposure to a topic, stimulus or idea which then biases how they answer future questions
What are the 4 types of bias?
referral, recall, confirmation, healthy worker effect
What is referral bias?
referred from a certain group of people
ex. diabetes survey - get all participants from same clinic, won’t necessarily reflect entire population
What is the healthy worker effect?
People who work are healthier than those who don’t
How can you ask sensitive questions? (4)
- normalize - “in talking to people we’ve found people often have too much work to exercise regularly. how often do you exercise?”
- re-affirm it is an opinion - “In your opinion…”
- Other people approach - “do you know of anyone who…”
- recognize measurement error will be an issue
What are inferences?
Can be made from sample to population only is you use a confidence interval or p value
What is a confidence interval?
How accurate the measurement is likely to be
- population with lots of variability leads to large CI
- larger samples tend to give us more info and we can be more sure of our estimate - lead to smaller CI
- estimated value of population parameter, indicates precision, narrow CI indicates more confidence
What is a P-value?
determines statistical significance in hypothesis test
- how likely it is to get a result like this is null hypothesis is true
- attempts to show that no variation exists between variables or that a single variable is no different than 0
- low p-value suggests data is inconsistent with the null, favouring an alternative hypothesis
What is a null hypothesis?
nothing happens, no difference
What is an alternative hypothesis?
change is assumed
What is a type I error?
falsely assume you can reject the null hypothesis and that the alternative hypothesis is true
- also called false positive result
- incorrect rejection of null hypothesis
- usually leads one to conclude that a supposed effect or relationship exists when it doesn’t
- can’t be completely avoided but investigators should decide on an acceptable level of risk when designing research
What is an example of a type 1 error?
Test shows patient to have a disease when they don’t
What is a type II error?
falsely assume you can reject alternative hypothesis and null hypothesis is true
- false negative result
- failure to reject a false null hypothesis
- leads to conclusion that an effect or relationship doesn’t exist when it does
What is significance in quantitative data?
odds observed result is die to chance
- don’t want this - set it low
What is power in quantitative data?
- probability of rejecting the null hypothesis while it is false (avoiding type II error)
- ranges from 0 to 1
- greater the power the more likely you are to detect a real difference or relationship
- odds you will observe a treatment effect when it occurs
- you want this - set it high
What is the most commonly used p-value?
0.05
What does a low p-value mean?
data is inconsistent with the null, favouring an alternative hypothesis
What is effect size in quantitative data?
shows that an effect was large enough to rule out null hypothesis
- effect size is independent of sample size
- tells you meaningfulness of the results
- large effect size means there is a strong relationship
What is Cohen’s d?
measures size of difference between 2 groups
What is Pearson’s r?
strength of relationship between 2 variables
What is a between subjects design?
different people test each condition, each person is only exposed to a single user intervention
-prevents carryover effect (lingering effects of being on 1 experimental condition on a subsequent condition
- shorter duration
- requires more participants and resources
- individual differences may threaten validity
What is a within subject design? (repeated measures)
same people test all the conditions
- requires fewer participants
- no variations in individual differences
- carryover effects - first test influences the other