Quiz #1 Flashcards
levels of measurement:
Nominal
Ordinal
Interval
Ratio
Extent to which information from a sample can be used to inform us about persons, places, or events that were not studied in the entire population from which the sample was taken
Generalizability
Summary statistics that capture the “typical,” “average,” or “most likely” score or value in a variable distribution.
Measures of Central Tendency
Most Frequently occurring data
mode
middle-point of the data
median
average of the data
mean
the width of the data
range
There is a concept called ______-____ ________ which, simply put, means reanalyzing data that already exist. These data usually come from one of two places: Either they are official data collected by local, state, and federal agencies (e.g., rates of crime reported to police, information on incarcerated offenders from state correctional authorities, or adjudication data from the courts), or they are data collected from surveys sponsored by government agencies or conducted by other researcher
Secondary data analysis
________ predictions are based on logic and probability.
theoretical
_________ outcomes are a numerical result from 1 or more samples I conduct
empirical outcomes (or observed outcomes
___________ __ _________ (_________): Capture how different the values of a variable are. The more dispersion there is in a variable, the more different the values are from each other or from some central tendency and the more heterogeneity there is in the distribution.
Measures of dispersion (variability):
average amount of variability in a set of scores.
In other words, how much do scores tend to vary from the average?
What is this term?
Standard deviation
Standard deviation squared
We don’t interpret this statistic, but it is one that is used in other
advanced statistics.
What is this term?
variance
If you have the Variance and need
to find the Standard Deviation, all
you need to do is find the _______ ______:
square root
If you have the Standard Deviation
and need to find the Variance, all
you need is to find _____ ________
the square
_____ is essentially the collection, organization, and analysis of numerical data
statistics
In an __________, the researcher has control to manipulate certain variables on the research subjects and record data. Maybe the researcher separates subjects into two groups, where one group of offenders is given some type of therapeutic treatment while the other group is not (the control group) in order to measure likelihood of committing more crime.
experiment
__________ variables do not change based on something else.
They can include characteristics that vary from person to person, but do not change based on another factor, such as your demographic questions: Age, sex, education, ethnicity, employment status, etc
Independent
______________ variables are things you measure based on something else. For example, how much you studied for an exam (independent variable) affects your score (dependent variable).
Dependent variables
a _________ is an “educated guess.”
hypothesis
________ are small or partial quantities of a larger ___________, or the entire group.
samples; population
Despite their usefulness and the seemingly indisputable “hard” evidence that statistics produce, the ________ (factual soundness) of statistics is not without problems and can be misleading or even dangerous
validity
“factual hardness” means: V
means validity
r________ = (consistency)
reliability
As such, properly constructing questions—be they for a survey or qualitative interview—is a careful, thoughtful process guided by logic. We call this process __________________.
operationalization
absolute zero
ratio
distance is meaningful
interval
attributes can be ordered
ordinal
attributes are only named; weakest
nominal
________ measures provide answer choices that are simply categorical. They have only two characteristics: they are exhaustive and mutually exclusive
nominal
________ measures have the same characteristics of mutual exclusivity and exhaustiveness, but the answer choices for these measures can also be ranked in some hierarchical fashion. Military ranks, grade level, and satisfaction scales are common ________ measures.
ordinal
Before moving on to ratio measures, we will take a brief moment to discuss a special type of measure: ________ measures. _________ level measures are the first instance of measures that can determine mathematical differences between subjects on a given concept. The most common example of an ___________ level measure is temperature
interval
Interval measures have a meaningful distance between points that is consistent across the whole range of the variable. In other words, measuring temperature in Fahrenheit means that a one-degree increase at any point on the temperature scale is the same change. E.g., 40 to 41 degrees is the same amount of change as 95-96 degrees. Interval measures are the first instance where we can place a meaningful numerical value that is consistent across the entire range of values. These measures are not often used in social science research, but our highest level of measurement—ratio measures—is quite common.
________ measures are similar to interval measures in that they have meaningful and consistent numerical values, but the added feature that sets them apart is that they have an absolute zero point.
ratio
However, in a _______ level measure zero means “absence”—it means that the concept has not occurred. In our truancy example, we could ask simply “How many times have you been truant in the past 30 days?” Making this question open-ended (where the student will write in the number of days) gives us a ______ level measure.
ratio
Measurement ______ evaluates the relationship between a concept and an empirical indicator of that concept. The question raised in assessing _______ is whether an indicator accurately measures what it is intended to measure. Importantly, the _________ of a measure must be contextualized with regard to its anticipated use
validity
Easily the weakest form of validity assessment, _______ validity involves the visual inspection of an indicator to ascertain whether it simply appears to accurately measure the concept at hand. This criterion amounts to little more than an individual’s subjective judgment that the measure looks correct and takes place after a measure has been created. For example, a measure of drug use that includes only marijuana would not pass the _____validity test. It clearly does not come close to accurately reflecting the concept “drug use” in its entirety.
face
_______ validity concerns itself with whether or not a given indicator captures the full range of the concept it is intended to measure. Many of the concepts of interest in the social sciences, such as crime, family behavior, or depression are multidimensional in nature. A valid measure of any of these concepts will be one that taps into multiple, if not all, of its dimensions.
content
_______ validity refers to the theoretical relationship that is expected between concepts. A valid measure in this regard will be related to other theoretically relevant (and valid) measures as suggested by theory and empirical research. A measure of poverty that has construct validity will be positively associated with both an area’s homicide rate and infant mortality rate.
construct
________-related validity concerns how well a measure relates to some external criterion that either measures the same concept or which the measure is intended to predict. This form of validity is generally useful in selecting individuals based upon an expected behavior or outcome given their value on another variable.
criterion
__________ validityis assessed when the item in question and its criterion are measured at the same point in time. Akers and Lee used concurrent validity to gauge the accuracy of responses by teenagers to their survey question on tobacco use. Teenagers, due to legal and social implications, may have strong motivation to lie when asked if they smoke cigarettes. In light of this, Akers and Lee tested the validity of this question by also taking saliva swabs of their respondents. If their question was accurate, the respondents who reported using tobacco should test positive for thiocyanate (a chemical found in nicotine) and those who report not using tobacco should test negative for this substance. Concurrent validity can be assessed in a single sample when two distinct measures of the same concept have been collected.
concurrent validity
_________ validity. Predictive validity is attained if a given measure predicts a future outcome that it is designed to forecast. The Graduate Record Exam (GRE), a widely required standardized test, is intended to measure one’s aptitude for graduate or business school. As such, if it is a valid measure it should accurately predict who succeeds in their graduate training and who does not. The GRE exhibits predictive validity in a variety of measures of graduate school success such as grade point average, comprehensive exam scores and program completion.
predictive
Statistics used to describe the distribution of a sample or population
Descriptive statistics
Statistical tools for estimating how likely it is that a statistical result based on data from a random sample is representative of the population from which the sample has been selected.
Inferential statistics:
Distribution of all possible outcomes of a trial and the associated probability of each outcome.
Probability distribution:
a _________ is simply a scientific “hunch” or assumption about the relationship between two variables that is tested empirically. In explanatory research, we start with a hypothesis, collect data or information that pertains to that hypothesis from a sample, and then, with the help of a known probability distribution, come to some decision about the hypothesis.
hypotheses
Consider a study that is observing aggression in children and using researcher assistants to count and record the number of aggressive acts engaged in by each child. How could we be certain that the number of aggressive acts reported is not based upon which research assistant was observing that particular child? If one research assistant is recording more subtle forms of aggression (such as shoving or tripping) while the second research assistant is only recording more serious aggressive acts (such as hitting and biting) we will not have reliable data. To ensure that all research assistants are providing the same information, they may both be asked to rate a series of the same children. Ideally, their recorded aggressive acts per child will be identical, but slight disagreement is acceptable. Using correlational analysis, the researcher can assess the inter-rater agreement.
which type of reliabilty is this?
inter-related reliability
When collecting data via a survey, a researcher may be concerned with whether or not an individual respondent would answer the same way every time they were asked such questions. Obviously, a survey that reports different responses from the same individual would be said to provide unreliable data. One might suspect that a question gauging impulsivity might suffer from such a potential concern. While many people might consistently report their impulsivity in a certain way, others might provide wildly different answers depending upon the time of day or year the question is asked or after a recent negative life experience. If a researcher suspects this type of problem with his measures, he should administer the survey to the same respondents twice within a short time frame (two weeks apart would generally be ideal). Reliable measures would have a strong correlation between answers provided at time 1 and time 2.
which reliability test is this?
In other words, if the measure of impulsivity is reliable, respondents will tend to provide the same answers both times, which is test-retest reliability.
in reality it is less than ideal to measure a concept like impulsivity with a single question. Impulsivity is multidimensional and can depend upon the context and the task at hand. For such concepts, researchers rely upon scales containing multiples measures of the same concept. For example, a researcher might ask 40 questions concerning impulsivity in speech, eating habits, employment, etc., to gather a more complete depiction of impulsivity. How can we be certain that all of these measures are capturing impulsivity and can thus be combined into a scale? Two methods exist to analyze scales using multiple measures to gauge ___-_____ reliability
inter-item
split half method and cronbach’s alpha. The split half method involves randomly splitting these measures into two groups and assessing the correlation across these two subsets. If the entire 40 questions are reliable as a measure of impulsivity, then any two subsests of these measures should be highly correlated. The second method, Cronbach’s alpha is a more advanced method and entails statistically comparing each individual item to all other items across individuals. Higher Cronbach’s alpha values (on a scale of 0 to 1) indicate that the measures are internally consistent—meaning that respondents who report being impulsive on one measure are doing so across all the other measures. Alternatively, respondents who report lacking impulsivity are doing so across all 40 measures included in the scale.
Tentative statement about empirical reality, involving the relationship between two or more variables.
Hypothesis
