type 1 error
finding an effect when in actuality there is no effect present
type 2 error
the failure to detect an effect that is there (the salk vaccine experiments)
characteristics of observational studies
unlike experiments, the subjects assign themselves to the groups, and the researchers just observe the result
association: one thing is linked to another
Causation cannot be established
majority of studies in HDFS fall under this category
characteristics of controlled experiments
a study where the investigators decide who will be in the treatment group and who will not
randomized controlled (controlled experiments)
when an impartial chance procedure is used to assign the subjects to treatment or control, the experiment is said to be randomized controlled
double blind (controlled experiments)
neither the subjects nor the doctors who measure the response should know who was in the treatment group and who was in the controlled group when nobody knows
Confounder
a third variable which differs between treatment and control groups (others than the treatment itself) which affects the responses being studied
population
the entire group of individuals that we want to learn something about
sample
any subset of the population
parameters
numerical facts about a population that sampling is attempting to learn
statistics
numbers computed from a sample which estimate population parameters
sampling error
discrepancy between sample statistics and their associated population parameters
sampling bias
a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others.
selection bias (sampling bias)
systematic tendency due to the sampling procedure which excludes one kind of person or another from the sample
response bias (sampling bias)
results when responses are distorted due to factors such as how questions are phrased, or the interviewer's tone or attitude
non response bias (sampling bias)
occurs when a large number of those selected for the sample do not respond to the questionnaire or interview and these participants are similar in some way
probability sampling
these method utilize objective chance processes to pick the sample, and do not allow the interviewer any discretion
this approach helps protect against sampling bias
there is a definite procedure for selecting the sample, and it involves the planned use o
simple random sampling
all subjects are chosen at random through a clear procedure without being replaced
cluster sampling (simple random sampling)
the population is broken up into groups (or clusters). then, one or more clusters are chosen at random and all subjects within the chosen cluster are sampled
stratified sampling
the population is broken up into groups (or strata) and a random sample is selected from each of the strata
convenience sampling
subjects are selected who are easy to get to participate or are easily reached
quota sampling
the sample is hand-picked to resemble certain characteristics of the population (who voted for who "president")
descriptive statistics
statistics that describe a sample
they simplify and summarize data
some examples are: mean, graphs, frequency tables, measures of variability
inferential statistics
techniques that allow us to infer information about populations from samples
constructs
theories generally contain hypothetical concepts
help to explain behavior
hypothetical- no way to prove if they actually exist
not directly observable
Measurement
1. we assign numbers to an object according to a rule
example rule: assign the number of lever presses as our measure of the construct addiction
2. properties of the attribute we are mesuring are represented by properties of the numbers
qualitative variables
data that reflect categorization (whether labels or numbers)
nominal (qualitative variables)
the numbers assigned 'name' the groups
property of the attribute is group membership represented by equality
if two objects are in the same group, they are assigned the same number
example book= 1 tv show= 2
quantitative variables
...
ordinal (quantitative variables)
property of the attribute is rank in terms of magnitude
represented by order of the numbers
ordinal means ordered
if two objects have the same number, then they have the same rank and are equal
interval (quantitative variables)
property of attribute is difference in magnitude represented by intervals between numbers
no "absolute" zero
ratio (quantitative variables)
property of the attribute is ratio between magnitudes they are interval scales, but with an "absolute" zero
frequency distribution
...
histogram
a graph for summarizing data
there is no vertical scale
set of blocks where each block represents a
range called a class interval
areas of block represent percentages
mode
the most frequently occuring number found in a set of numbers
mean
the average number, the sum of a list of numbers, divided by how many numbers there are
meidan
middle value when numbers are placed in order
standard deviation
how far away number on a list are from their mean
frequency
the number of observations in a category ( or interval) symbolized as f
modality
distribution with one 'hump' is called unimodal distribution with two humps is bimodal
skewness x2
measures the degree to which a distribution departs from being symmetrical
positively skewed distribution has a larger tail in the positive direction
negatively skewed distribution has a larger tail in the negative direction
kurtosis
measures the degree to which a distribution is pointy or flat and spread out
leptokurtic- distribution is very pointy
platykurtic- distribution is flatter and more spread out
most meaningful for unimodal distribution
normal distribution
a theoretical distribution which is unimodal, symmetrical (i.e., 0 skew), and bell shaped ( not leptokurtic or platykurtic
central tendency statistics
mode is the most frequent or "typical value" median is the middle value where 50% of the data is above and 50% of the data is below mean is the average value or the balance point they do not always agree
if distribution is symmetric then median and mean w
central tendency statistics when they don't agree
if bimodal and symmetric there are multiple modes and they are not equal to mean and median if distribution is unimodal but skewed all three will disagree and remember skewed distribution has a long trail in one direction positively skewed distribution is
outlier
are extreme scores scores much higher or lower than the rest of the sample mean strongly influenced by the presence of extreme scores
mean is strongly influenced, median is not sensitive, standard deviation is strongly influenced
adding to data
adding the same number to every observation on a variable: will create a new mean exactly that number larger than the old mean the standard deviation does not change (everything gets affect besides the standard deviation)
multiplying data entries by the same number
both the mean and standard deviation change the average and standard deviation each get multiplied by that same number if multiplier is negative, standard deviation is multiplied by absolute value
chance error
no matter how careful you are, repeated measurements can turn out a bit differently
bias/systematic error
pushes measurements in the same direction
individual measurement = exact value + bias+ chance error
measurement bias
measurement are either systematically too high or too low
sampling bias
is a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others.
slope
rise/run
run = increase in x coordinate
rise= increase in y-coordinate
intercept
height of a line where x=0
validity
the degree to which a test or examination measures what it purports to measure
content (validity)
expert judgment/development
criterion
prediction, both concurrent and predictive
construct
the theory
reliability
the overall consistency of a measure
the extent of unsystematic variation (random error) in the quantitative description of some characteristic of an individual when that individual is measured a number of times
alternate forms
construct two forms as parallel as possible
administer both forms
take the correlation of observed scores ( coefficient of equivalence)
split half
split test in half and separately score
correlation is an estimate of reliability for a test half as long
test-retest
administer the same test twice take the correlation of observed scores ( coefficient of stability)
coefficient alpha ( Cronbach's alpha)
measure of internal consistency to what extent the components all measure the same attribute
interrater reliability
degree to which raters agree in their ratings
not necessarily in value, but at least in rank ordering
ethics
are the values by which human behavior is morally evaluated
guidelines the APA have developed to ensure research is ethical
respect for persons and their autonomy
prospective participants must know what they are getting into voluntarily agree if unable to give consent (i.g. kids) must have appropriate representative
beneficence and nonmaleficence
research must have some concei
type 1 error
finding an effect when in actuality there is no effect present
type 2 error
the failure to detect an effect that is there (the salk vaccine experiments)
characteristics of observational studies
unlike experiments, the subjects assign themselves to the groups, and the researchers just observe the result
association: one thing is linked to another
Causation cannot be established
majority of studies in HDFS fall under this category
characteristics of controlled experiments
a study where the investigators decide who will be in the treatment group and who will not
randomized controlled (controlled experiments)
when an impartial chance procedure is used to assign the subjects to treatment or control, the experiment is said to be randomized controlled
double blind (controlled experiments)
neither the subjects nor the doctors who measure the response should know who was in the treatment group and who was in the controlled group when nobody knows
Confounder
a third variable which differs between treatment and control groups (others than the treatment itself) which affects the responses being studied
population
the entire group of individuals that we want to learn something about
sample
any subset of the population
parameters
numerical facts about a population that sampling is attempting to learn
statistics
numbers computed from a sample which estimate population parameters
sampling error
discrepancy between sample statistics and their associated population parameters
sampling bias
a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others.
selection bias (sampling bias)
systematic tendency due to the sampling procedure which excludes one kind of person or another from the sample
response bias (sampling bias)
results when responses are distorted due to factors such as how questions are phrased, or the interviewer's tone or attitude
non response bias (sampling bias)
occurs when a large number of those selected for the sample do not respond to the questionnaire or interview and these participants are similar in some way
probability sampling
these method utilize objective chance processes to pick the sample, and do not allow the interviewer any discretion
this approach helps protect against sampling bias
there is a definite procedure for selecting the sample, and it involves the planned use o
simple random sampling
all subjects are chosen at random through a clear procedure without being replaced
cluster sampling (simple random sampling)
the population is broken up into groups (or clusters). then, one or more clusters are chosen at random and all subjects within the chosen cluster are sampled
stratified sampling
the population is broken up into groups (or strata) and a random sample is selected from each of the strata
convenience sampling
subjects are selected who are easy to get to participate or are easily reached
quota sampling
the sample is hand-picked to resemble certain characteristics of the population (who voted for who "president")
descriptive statistics
statistics that describe a sample
they simplify and summarize data
some examples are: mean, graphs, frequency tables, measures of variability
inferential statistics
techniques that allow us to infer information about populations from samples
constructs
theories generally contain hypothetical concepts
help to explain behavior
hypothetical- no way to prove if they actually exist
not directly observable
Measurement
1. we assign numbers to an object according to a rule
example rule: assign the number of lever presses as our measure of the construct addiction
2. properties of the attribute we are mesuring are represented by properties of the numbers
qualitative variables
data that reflect categorization (whether labels or numbers)
nominal (qualitative variables)
the numbers assigned 'name' the groups
property of the attribute is group membership represented by equality
if two objects are in the same group, they are assigned the same number
example book= 1 tv show= 2
quantitative variables
...
ordinal (quantitative variables)
property of the attribute is rank in terms of magnitude
represented by order of the numbers
ordinal means ordered
if two objects have the same number, then they have the same rank and are equal
interval (quantitative variables)
property of attribute is difference in magnitude represented by intervals between numbers
no "absolute" zero
ratio (quantitative variables)
property of the attribute is ratio between magnitudes they are interval scales, but with an "absolute" zero
frequency distribution
...
histogram
a graph for summarizing data
there is no vertical scale
set of blocks where each block represents a
range called a class interval
areas of block represent percentages
mode
the most frequently occuring number found in a set of numbers
mean
the average number, the sum of a list of numbers, divided by how many numbers there are
meidan
middle value when numbers are placed in order
standard deviation
how far away number on a list are from their mean
frequency
the number of observations in a category ( or interval) symbolized as f
modality
distribution with one 'hump' is called unimodal distribution with two humps is bimodal
skewness x2
measures the degree to which a distribution departs from being symmetrical
positively skewed distribution has a larger tail in the positive direction
negatively skewed distribution has a larger tail in the negative direction
kurtosis
measures the degree to which a distribution is pointy or flat and spread out
leptokurtic- distribution is very pointy
platykurtic- distribution is flatter and more spread out
most meaningful for unimodal distribution
normal distribution
a theoretical distribution which is unimodal, symmetrical (i.e., 0 skew), and bell shaped ( not leptokurtic or platykurtic
central tendency statistics
mode is the most frequent or "typical value" median is the middle value where 50% of the data is above and 50% of the data is below mean is the average value or the balance point they do not always agree
if distribution is symmetric then median and mean w
central tendency statistics when they don't agree
if bimodal and symmetric there are multiple modes and they are not equal to mean and median if distribution is unimodal but skewed all three will disagree and remember skewed distribution has a long trail in one direction positively skewed distribution is
outlier
are extreme scores scores much higher or lower than the rest of the sample mean strongly influenced by the presence of extreme scores
mean is strongly influenced, median is not sensitive, standard deviation is strongly influenced
adding to data
adding the same number to every observation on a variable: will create a new mean exactly that number larger than the old mean the standard deviation does not change (everything gets affect besides the standard deviation)
multiplying data entries by the same number
both the mean and standard deviation change the average and standard deviation each get multiplied by that same number if multiplier is negative, standard deviation is multiplied by absolute value
chance error
no matter how careful you are, repeated measurements can turn out a bit differently
bias/systematic error
pushes measurements in the same direction
individual measurement = exact value + bias+ chance error
measurement bias
measurement are either systematically too high or too low
sampling bias
is a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others.
slope
rise/run
run = increase in x coordinate
rise= increase in y-coordinate
intercept
height of a line where x=0
validity
the degree to which a test or examination measures what it purports to measure
content (validity)
expert judgment/development
criterion
prediction, both concurrent and predictive
construct
the theory
reliability
the overall consistency of a measure
the extent of unsystematic variation (random error) in the quantitative description of some characteristic of an individual when that individual is measured a number of times
alternate forms
construct two forms as parallel as possible
administer both forms
take the correlation of observed scores ( coefficient of equivalence)
split half
split test in half and separately score
correlation is an estimate of reliability for a test half as long
test-retest
administer the same test twice take the correlation of observed scores ( coefficient of stability)
coefficient alpha ( Cronbach's alpha)
measure of internal consistency to what extent the components all measure the same attribute
interrater reliability
degree to which raters agree in their ratings
not necessarily in value, but at least in rank ordering
ethics
are the values by which human behavior is morally evaluated
guidelines the APA have developed to ensure research is ethical
respect for persons and their autonomy
prospective participants must know what they are getting into voluntarily agree if unable to give consent (i.g. kids) must have appropriate representative
beneficence and nonmaleficence
research must have some concei