Research Methods Exam 2

Probability Sampling

Drawing the sample at random from that a certain population that you are studying. In this EVERYONE in the population has a chance at being chosen.

Categorical Variables

(Nominal Variables): are categories of measurement. Examples: Sex, whose levels are male and female. Species whose levels may be rhesus macaque, chimpanzee, and bonobo.

Ordinal Scale

Permits events or scores to be rank ordered. First, Second, Third Place; Like in a race. Example: A travel website might classify a set of beach resorts as two, three, or four stars. Four-star resorts are better than three- or two-star resorts. List them in order. Example: A professor might use the order in which exams were turned in to operationalize how fast student completed the exam. This is ordinal data because the fastest exams are on the bottom of the pile—ranked 1.

Ratio Scale

Permits rank ordering of events (like ordinal) with the assumption of equal intervals between adjacent events (like interval) and a TRUE ZERO POINT. Examples: Physical Measurements such as intensity of light, volume of water, speed, or action potentials per second. Examples: Income CAN MAKE ratio comparisons such as "Twice as much" or "Half as much".

Interval Scale

Permits rank ordering of events with the assumption of equal intervals between adjacent events. Example: Temperature or GRE Score No True Zero: Even if it is 0 Degrees out a temperature still exists. A 0 on your GRE doesn't mean you have no verbal, math, writing skills, just means you may have very low ones. CANNOT SAY: Twice as hot, Half as intelligent (Does not work)

Social Desirable Responding

(Faking good), respondents may know what they really think, but they are too embarrassed, shy, or worried about giving an unpopular opinion to give that answer on a survey.

Multiple Choice

The respondent must select the most suitable response from among several alternatives. Example: YOU SHOULD KNOW WHAT A MULTIPLE CHOICE QUESTION IS.

Multistage Sampling

A method of sampling in which two random samples are taken from some population: a random sample of clusters and then a random sample of people within those clusters. Like cluster, but you take a random sample from each cluster, not everyone in each cluster.

Open- Ended Questions

A question is asked to which the respondent must construct his or her own answer. Advantage: Information more complete Disadvantage: Participant may not understand what you are looking for; some answers omitted inadvertently; summarizing data is difficult.

Cluster Sampling

A sampling method in which researchers randomly select clusters of participants within the "population" of interest and then collect data from all of the participants in each selected cluster. Interested in all state university student in Florida? Make a list of universities (those are clusters).... Randomly select 5... include ALL students from each of those 5 in your sample.

Internal Reliability

A study participant gives a consistent pattern of answers, no matter how the researcher has phrased the question. Example: So if a participant is asked " Am I satisfied with my life?" and they answer, "Yes". Then when asked, "Am I NOT satisfied with my life?" They should answer "No" in order to maintain internal Reliability. Review scatterplots, focusing on how scatterplots show the direction and strength of a relationship. You can see whether the two ratings agree (if the individual dots are close to a straight line drawn through them) or whether they disagree (if the individual dots scatter widely from a straight line drawn through them).

Moderator

A third variable that, depending on its level, changes the relationship between two other variables. Examples: In asking whether the association between extraversion nd likings for upbeat, popular music would generalize to a group of 70 to 80 year olds, we were asking whether that association would be moderated by age.

Snowball Sampling

Ask participants to recruit others.

Convenience Sampling

Available and willing

Observing People Ethically

Believed that it is ethical to observe any human behavior in public areas, but must not disclose person information. To video tape and or observe people in their private dwellings permission must provided up-front; if researchers did video tape people covertly, must explain the taping at the end of the study and if the participant wants it deleted, the researcher must dispose of it right away.

TO ASSESS RELIABILTY

Both X & Y variables are interval or ratio scale measurements. Data has to be linear (NO CURVILINEAR). Closer the dots are on a POSITIVE Slope line, the better the reliability. Further the dots are on a POSTIVE line, you have less reliability.

Longitudinal Data

Can help establish temporal Precedence. It measures the same variable multiple times across time (days, months, years)

Interrater Reliability

Consistent results are obtained no matter who measures or observes. Examples: When two researchers observing children's behavior at preschool or park are using the same scale to measure happiness, they should come up with the same (or VERY similar) findings. It is important because one observer may miss or overlook a bit of behavior. There also may be some disagreement concerning exactly what was seen and how it should be rated or categorized.

What does it mean to "control for" a variable in a regression?

Control For: holding potential third variable steady while investigating the association between two other variables.

Anchors

the labeled opposite extremes and sometimes middle, especially if it is to represent "neutral".

Can an effect size be large even in cases with small r coefficients?

Effect size cannot be large with small coefficients. A small sample is more easily affected by chance events than a large sample is. A WEAK correlation based on a small sample is more likely to be the result of chance variation and is more likely to be judge "not significant".

Forced Choice Questions

Forced Alternative questions; the respondent must select between two alternative responses. Often is "agree" or "disagree"; "No Opinion" or "Not Sure" are NOT options. Examples: Is the position of women "better" or "worse" than it was ten years ago? Should there be "more" or "fewer" women in elected positions? Example: It could ask you to check one or the other such as: ____ I really like to be the center of attention ____ It makes me uncomfortable to be the center of attention

Concurrent

How well the measure correlates with a current outcome.

Predictive

How well the measure predicts an outcome

Understand why correlations cannot establish causation

In order to have causation. You MUST have... Convariation of the cause variable and the effect variable: There must be correlation, or association, between the cause and the effect. Temporal Precedence: The causal variable HAS to come FIRST in time, before the effect variable. Internal Validity: There must be NO plausible alternative explanation for the relationship. When you take those three things and put them up against most correlations you will not be able to meet those three criteria, why we would call it correlation.

Which are most critical to establishing temporal precedence and causation?

Inferences can be made regarding directionality by looking at criss-lag correlation.

Regression Line

Is a straight line that describes how an outcome variable "y" changes as a predictor variable "x" changes. We often use a regression line to predict the value of "y" for a give value of "x"

How does pattern and parsimony help us understand that the relationship between cigarette smoking and lung cancer IS causal?

Many different studies from many different researchers, all support the theory that smoking cigs causes lung cancer. The diversity of the predictions/ findings makes it hard to argue for third variables. Pattern & Parsimony just play into the picture because once you do something over and over and over again and you get the same results most likely you could have a causal effect.

How is this similar to a psychologist focusing his/her entire research career on one topic?

Many psychologists devote their whole lives to one research item because they seek causation. They do their experiment over and over and over again, trying to prove something. When you do that it takes up a lifetime.

Discriminant Validity

Measure should NOT correlate well with measures of other, DISTINCT constructs. There should NOT be a correlation with measure of other traits Need to demonstrate discriminant validity for other traits the might be reasonably mistake for the trait being measured. The measures do not have to be negatively correlated, just weakly correlated.

Convergent Validity

Measure should correlate strongly with other measures of the SAME construct. Example: If a scale is a good measure of depression, it should correlate well with other measures of depression. BDI (Beck Depression Inventory) & CES-D (Center for Epidemiologic Studies Depression Scale. "r"= 0.68 (Means there is a positive correlation which means it has convergent validity)

Pearson's Correlation Coefficient

Most widely used measure of association Value or "r" can range from (+1) (0) (-1) Magnitude of "r" tells you the strength of LINEAR relationships .10= Small/ weak * .30=Medium/moderate*Large Strong=. 50 The sign of "r" tell you which direction the slope will being going, either positive or negative. Positive linear correlation, line slants UP to the right. Negative linear correlation, line slants DOWN to the right. ZERO correlation: NO LINE

Besides bias, what other concerns must people take into consideration when conducting observational data collection?

Observers might See what the expect to see: So if people are told they are going to see, they have already come expecting to see a certain thing, and they will base their judgments off of the knowledge they obtain. Observers can affect what they see Observer Effects: occur when people changer their behavior (react) when they know another person is watching. For example; when you were in grade school and the teacher said "okay class, the principle is coming in to watch the class and give me an evaluation" and the class was on their best behavior.

What is the affect of an outlier on a correlation?

Outliers can be problematic for an association claim, because even though they are only one or two data points, they may exert disproportionate influence.

Random Assignment Versus Random Sample

Random Sample: researchers draw a sample using some random method—drawing names from a hat or using a random digit. Random assignment: used ONLY IN EXPERIMENTAL DESIGNS. When researchers want to assign participants into different groups, they usually assign them at random to the groups. (keeps internal validity up)

Exact Point Labeling

Reduces ambiguity

How can you reduce social desirable responding?

Reducing Social Desirable Responses: Ask questions in your surveys such as "My table manners at home are as good as when I eat out in a restaurant", if a respondent answers these with obvious things that are not true, then they are considered high on Social Desirability answers and their survey is thrown away. Reducing: You can have filler items. On a 20-question survey they may ask you 5 questions about racial attitudes (the actual focus), but the other questions may be about politics, gender roles, or education. This hides the true meaning of the survey. Reducing: Researchers use special, computerized measures to evaluate people's implicit opinions about sensitive topics.

Operationalization Variable

Represents a researcher's specific decision about how to MEASURE or MANIPULATE the conceptual variable.

Stratified Random Sampling

Researcher selects particular demographic categories on purpose and then randomly selects individuals within each other the categories.

Correlation Coefficient r

Researchers use a single number to indicate how close the dots on a scatterplot are to a line drawn through them . Represents the degree of relation between two variables. The value of a correlation coefficient can range from -1 (Perfect Negative) to +1 (perfect positive). On the scatter plots, whether the dots are close together or further apart matters. The closer together they are on the slope line the stronger the correlation; the further from the slope line the weaker the correlation. When the scatters plot slope is positive, "r" is positive; when the scatter plot slope is negative, "r" is negative.

Biased Sampling

Sampling only those who are easy to contact, Sampling only those who you are able to contact, sampling only those who invite themselves. These all do not get the "representative population" that is needed when creating a sample. We want to make sure that we have a sample that is representative of the population we are trying to study.

Purpose Sampling

Seek out a specific population

Population

Some larger group from which a sample is drawn, which the sample is intended to represent

How can looking at "subgroups" help explain unusual scatterplot results?

Subgroups can cause issues because sometimes, when you have an association between two variables, the apparent over all association is spurious, meaning that the overall relationship is attributable only to systematic mean differences on subgroups within the sample. Example: Imagine you have a large sample of college students and you measure the number of times each student skips class per semester, along with his or her GPA for that semester. When you look at the plot it seems as though more absences are associated with higher grades? But then we notice that that the scatterplot put all students on one graph. Freshman, sophomores, Juniors, Seniors are all subgroups and when we look at them from a sub-group point of view we actually see a negative correlation. So when there is sub-groups we have to be careful that we are not looking at the picture as a whole.

Beta Coefficient:

Tells if a significant correlation exists between 2 variables (predictor variable and dependent variable) when the other predictor variables are controlled for. Can be negative or positive or zero. Can be higher or lower.

Response Set

Tendency for participants to answer all (or most) of the questions the same way whenever questions in a series all have the same choices for responding.

Autocorrelations

Test whether each variable correlates with itself over time.

Cross-Lag Correlations

Test whether early measure of on variable is associated with late measure of other variable. Addresses directionality problem; establishes temporal precedence.

Effect Size

Usually, the stronger a correlation (and the larger its effect size), the more likely the correlation will be STATISTICALLY SIGNIFICANT. That's because the larger an association is, the less likely it could have been sampled, just by chance, a particular correlation is statistically significant by looking at its effect size alone.

Likert Type Scale

The individual answers a question by selecting a response alternative from a designated scale. Example: A typical scale might be following; (5) strong agree, (4) agree, (3) undecided, (2) disagree, or (1) strongly disagree. Likert Scales SHOULD involve a mixture of positive and negative statement including some alternate phrasing of the same item. Such as "Teenagers are rude and disrespectful" & then "Today's teenagers are polite and courteous".

Line of the least squares

The line that makes the sum of the squares of the vertical distances of the data points from the line as small as possible.

Simple Random Sampling

The most basics form; imagine that each member of the population puts a name in a hat, 10 people are chosen from the hat, those 10 people will make up the population.

Systematic Sampling

The researcher starts by selecting two random numbers using a computer or a random number table. Lets say you pick the numbers 2 & 6 and your population was a room full of students. You would go to the 2nd person. Then count off until you hit 6, once you hit 6, that is your next person in for your study.

Outliers

There will sometimes be an extreme score- a sign case (sometimes few) that stands out far away from the pack.

Cross-sectional Correlation

They test to see whether two variables, measured t the same point in time, are correlated

What are ways that you can design a survey to reduce the likelihood of a response set?

Ways to reduce the probability of that happening. Don't Anchor #3, Make it a 4 point scale. Use some statements that require "agree" and others that require "disagree" to address the same construct. Have a mixture of positive and negative statements including some alternate phrasing of the same item. Example: Just have Stronlgy Agree, Agree, Disagree, Strongly Disagree (NO Neutral); The Mixture: "Todays teenagers are rude and disrespectful" & "Todays teenagers are polite and courteous". One has to be agree and the other has to be disagree.

Understand p values as they related to statistical significance as well as what a p value less than .05 means

We are ok with saying there is a relation between 2 variables in our sample when there is not a relation in the population 5% of the time

What is the difference between a moderator and a mediator?

When researchers test for mediating variables, they ask, "Why are these two variables linked?" When they test for moderating variables, they ask, "Are these two variables linked the same way for everyone, or in every situation?"

Would a bar graph or a scatterplot be more appropriate to examine data when one of your variables is measured on a nominal (categorical) scale?

With categories Bar Graphs are the way to go.

Moderator

a THIRD variable that, depending on its level, changes the relationship between two other variables.

Mediator

a variable that helps explain the relationship between two other variables

Quantitative Variables

are coded with meaningful numbers, so when you see numbers they will mean something. Examples: Height and Weight are quantitative because they are measured in numbers, such as 150centimeters or 45 kilograms. Examples: IQ score, level of brain efficiency, and amount of salivary cortisol are all quantitative variables.

Predictive Validity & Concurrent Validity

evaluate whether the measure under consideration is related to a concrete outcome that is should be related to , according to the theory being tested. Example: I want to develop a scale to measure likelihood of college success. I write my items... my scale seems to have good face & content validity. I could give it to entering Freshmen and then measure their success at the end of the school year (PREDICTIVE VALIDITY) I could give it to Sophomores and at the same time correlate it to their current GPA. (CONCURRENT VALIDITY)

Nominal Scale

events are assigned to categories(Any numbers on this scale ONLY serve as labels) Might assign a Species as the category, the levels being Rhesus Macaque, Chimp, and Bonobo. Researcher might then decide to assign numbers to the levels creating a nominal scale. Researcher will use 1 to represent Rhesus Macaque, 2 for Chimp, and 3 for Bonobos, during the data- entry process. THESE NUMBERS HAVE NO NUMERICAL MEANING; JUST FOR CATEGORIZING. ONLY CATEGORICAL SCALE (REST ARE QUANTITATIVE)

Reliability

how CONSISTENT is the measurement?

Known- Groups Paradigm

in which researchers see whether scores on the measure can discriminate among a set of groups whose behavior is already well understood.

Content Validity

involves subjective judgment about a measure; does it capture all parts of a defined construct? Examples: Ability to reason, ability to plan, abstract thinking, comprehension of complex ideas, ability to learn quickly, ect. (A measurement of intelligence with GOOD content validity would HAVE to address ALL of these)

Implicit Measure

is a measure within social psychology designed to detect the strength of a person's automatic association between mental representations of objects (concepts) in memory.

Observer Bias

is a potential threat to construct validity, in which observers record what they want to see or expect to see, rather than what is really happening.

Validity

is it measure what it is SUPPOSED to measure?

Conceptual Variable

is the researchers definition of the variable in question at an **ABSTRACT** level.

Yes or No Questions

iterally questions where you can answer yes or no Examples: Is it possible to believe in both God & Evolution? Do you approve of the job the president is doing?

Physiological Measures:

operationalizes a variable by recording biological data such as brain activity, hormone levels, or heart rate. It usually requires the use of equipment to amplify, record, and analyze biological activity. Examples: Moment-to-moment happiness has been measured using facial electromyography (EMG)- a way of electronically recording tiny movements in the muscles in the face. Examples: Operationalize stress through physiological means would to be measure the hormone cortisol that is released in their saliva, because people under stress show higher rates of cortisol.

Self-Report Measure:

operationalizes a variable by recording people's answers to verbal questions about themselves in a questionnaire or interview. Strength: Popular and easy to use Weakness: You cannot be sure that a participant is telling you the truth (Bias, subjective). ** Examples: Diener's Five-item scale is an example of a self-report measure, in which people answer questions about their own life satisfaction. ** Examples: Self- Report measure of intelligence could ask, "How intelligent are you?" Self-report measure on stress may ask you "In the last month, how often have you felt nervous and stressed?"

Statistically Significant

p<.05= there is a less than 5% chance that we found this relation by chance and that it does not exist in the population

Non-significant

p>.05 = there is a greater than 5% chance that we found this relation in our sample and it does not exist in the population.

How are r and r^2 related?

r & r^2 are related by The relationship between memory score and achievement is statistically significant (RELATIONSHIP= r) How much variance in achievement scores is explained by memory? (VARIANCE = r^2)

Test-Retest Reliability

researchers gets consistent results every time he or she uses the measure. Examples: Test/ inventory is given twice; scores from the two tests are compared. Greater the similarity higher the reliability.

Observational Measure:

sometimes referred to behavior measure. It operationalizes a variable by recording observable behaviors or physical traces of behaviors. Two Types Frequency Count of the number of behaviors that occur in a given period of time (# social interaction s of kids on play ground) Latency: The amount of time it takes for a behavior to occur (reaction time; how long before kids on playground begin to interact) ** Examples: You could operationalize happiness by observing how many times a day a person smiles. ** Examples: An observational measure of stress might involve recording a behavioral signs of stress, such as a person's negative facial expressions (frowning or grimacing) or bodily agitation (tapping toes or tense body language).

Construct Validity

the ability of a measurement tool to actually measure the psychological concept being studied. In other words, does it properly measure what it's supposed to measure?

Sample

the group of people, animals, or cases used in a study

Face Validity

to the extent that it is a plausible measure of the variable in question. Examples: Life satisfaction as a measure of happiness? Head Circumference as a measure of intelligence? Head circumference as a measure of hat size? Speed of problem solving as a measure of intelligence? (Just by looking at these questions, do they make sense or show any validity? If yes, you have face validity, if no then you do not)

Regression Equation

y =bx+a b= Slope; x= the given number a= y-intercept= why the line hits the y-axis