Compiling data Flashcards
ways to classify data
- type
quantitative vs qualitative - source
internal vs external sources
primary vs secondary sources
experimental vs non-experimental aquisition
primary vs secondary
primary:
collected from the institution that originally obtained the info
secondary:
obtained from a source other than the primary data source
experimental acquisition vs non-experimental acquisition
experimental acquisition
- collecting data by manipulation or selective sampling to isolate their effects
ie. how sediment responds to waterflow in a stream
go to a flume, control flow rate of water going through a system
non-experimental acquisition
- no control is exercised over the collection of the data
ie. collected info from a natural stream (cant control dimensions of stream, flow rates, etc.
factors to evaluate data
validity
- does this measure what we think it is measuring
ie. IQ test doesn’t actually measure intelligence, it measures responses to questions. that is used as a proxy to estimate intelligence
accuracy
-agreement between measured value and true value
- difficult because we may not know the true value
eg. scale reads one pound but has nothing on it - not accurate
reliability
- deals with the freedom from substantial bias
eg. the scale measures one pound today, but two pounds tomorrow -> not reliable.
it would be reliable if every day it measured one pound over and over
what drives water decisions
environmental conditions
- maintaining flow
- ecosystem health
economics
- cost of irrigator
politics
- interests those in power have
cultural identities and values
who is involved in decisions?
regulators (govern how decisions are made, instigating decisions)
stakeholders
-irrigators
-industry
-public interest groups
need to be inclusive, include anyone who feels they are involved
basic steps for decision makers
identify stakeholders and how they are involved
how to address the decision
-> time frame?
->future projections?
-> how its been done historically
identify methods, metrics, data requirements
-> do we have the means to do it
look at results and recommend decisions
for all steps, always check in with stakeholders as decisions are made (ie. methods chosen)
-continuous involvement of stakeholders
statistics vs mathematics
math is deductive
- cannot go beyond the information given
stats is inductive
-beyond info given, inducing something from data provided
eg. data says that 20% of adults in a sample of a suburb commute to the city centre,
could then induce that 20% of all adults in the suburb commute to the city centre
inferential statistics
procedures for
- calculating limits and probabilities
- testing statistical hypotheses
- drawing statistical inferences
overall goal of statistical analysis
come up with a better understanding of the thing we are interested in/studying
start with our understanding before the analysis
go get data and collect info
organize and manipulate the data
interpret the data
finally, should have a different understanding than from before the analysis
Basic objective of statistical analyses
- data reduction
- summarizing data and reducing them into something simpler - inference
- assessing degree of accuracy of our measurements
- are there real differences in sets of data or is it error - identify associations of relationships
- correlational
-> no conclusion about cause can be made
-> theres a correlation but cant say for certain that one thing causes the other
-causal
-> mechanistic conclusions can be drawn
eg. smoking causes cancer
objects
sources of data
- observational data
eg. people, places, time periods, etc
variables
properties of the objects
eg. height of a person
variables and objects are dependant on each other
(height means nothing without knowing what it is the height of)
4 types of scales
nominal
- variable of object has values of a KIND only (eg. color)
- numbers have no meaning other than a label
-cannot be ranked in a meaningful way
eg. different brands of toothpaste
ordinal
- values of variables can be arranged in a meaningful order
- degree of difference cannot be assessed on a scale
- can be ranked
eg. bond grades (AAA is better than AA is better than an A)
interval
- equal distances between scale values have equal meanings
- variables are quantitative
-can be ranked
-ratios of values have no meaning because there is no true zero value
eg. thermometer (40 is hotter than 20, but is not twice as hot as 20)
ratio
-equal distance between scale value with equal meaning
-variables are quantitative
-ratios do have meaning ( has a genuine zero)
-gives most amount of info
-can be rankked
eg. a ruler
classes of variables
qualitative: nominal or ordinal scales
quantitative: interval or ratio scales
discrete: can only take on a finite number of values (cant have 2.45 people)
continuous: can take on any numeric value
dichotomous variable: can only assume 2 values (aka binary)
-> dummy variables: useful for turning qualitative into quantitative
