Research methods chapter 5-9

Random sampling techniques

-Simple random sampling: assigning arbitrary numeric values to each individual and selecting number through random number generation -Cluster sampling: randomly selecting a cluster from the population and sampling every individual from within the cluster -multistage sampling: a random sample of clusters is selected from your population of interest, then a random sample of individuals is chosen from these clusters -stratified random sampling: a multistage technique where the researcher selects specific demographic categories and then randomly selects individuals from each of the categories -oversampling: stratified random sampling where a researcher over represents one or more groups

Data visualization-the standards for good figures are:

-Simplicity -clarity -continuity -information value

Effect size

-The proportion of variability two variables have in common (a.k.a. "explained variance", a.k.a. "coefficient of determination") -calculated simply as R² -can be converted to a percentage by multiplying R² by 100 -larger effect sizes give more accurate predictions

Survey administration Instructions must be...

-clear -concise -easy to follow -thoroughly rehearsed

Five steps to determining mediation:

1. Test relationship between the predictor and criterion variable 2. Test relationship between the predictor and mediator 3. Test relationship between mediator and criterion variable 4. Use regression to test if the effect of the predictor variable on the criterion variable goes away when controlling for the mediator 5. Mediation is only established when the predictor is measured first, followed by mediator, followed by the criterion variable

Categorical measures: #of agreements /#of opportunities X 100 = % reliability

85% agreement is generally considered to be an acceptable minimum level of reliability

Define statistics

A branch of mathematics that involves the collection, analysis, and interpretation of data

Scatterplot

A figure with data displayed as a collection of points each having the value of one variable determining the position on the horizontal axis and the value of the other variable determining the position on the vertical axis. Can help with determining correlation or association claim

Bar graph

A graph in which a summary for each level of a categorical variable is represented as a vertical column. The columns of a bar graph DO NOT touch. Frequently used to depict the results of experiments IV = x axis, DV = y axis

Line graph

A graph in which a summary for each level of a quantitative variable is represented as a dot and each dot is connected by a line

Histogram

A graph in which the frequency for each category of a quantitative variable is represented as a vertical column that TOUCHES the adjacent column

Frequency polygon

A graph that is constructed by placing a dot in the center of each bar of a histogram and then connecting the dots which accomplishes the same as a frequency histogram

Range

A measure of variability that is computed by subtracting the smallest score from the largest score

Census

A set of observations that contains all members of the population of interest

Descriptive statistics summarize...

A set or distribution of numbers (a measure of central tendency and a measure of variability should be included in the summary)

Variance

A single number that represents the average amount of variation in a distribution

Representative sample

All members of the population have an equal chance of being included in the sample

Inferential statistics statistical significance

An inferential statistical test can tell us whether the results of an experiment can occur frequently or rarely by chance -large p-value means the results occur frequently by chance -small p-value means the results occur rarely by chance (Results that occur rarely by chance (p < .05) are significant

"Third" variables:

Any factor that might affect an observed bivariate correlation

Correlations

Any study where two variables are measured (not manipulated) it is correlational -example: one categorical, one quantitative variable

Positive correlation

As one variable increases, scores on the other variable also increases (a perfect positive correlation has a correlation coefficient of 1)

Sometimes, a representative sample is not the top priority in...

Association and causal claims

Bivariate correlations:

Associations that involve exactly 2 quantitative variables

Characteristics of good figures:

-Augment rather than duplicate the text -convey only essential facts -omit visually distracting detail -Are easy to read-it's elements (type, lines, labels, symbols, etc.) are large enough to be read with ease -are easy to understand-readily apparent purpose -are stylistically consistent with one another

Multivariate designs and the four validities:

-Construct validity -external validity -statistical validity -internal validity

Causes of biased samples

-Convenience sampling: sampling only those who are most easy to contact -self selection: sampling only those who invite themselves

Longitudinal designs can provide some evidence for causation by trying to address these three criteria:

-Covariance -temporal dependence -internal validity

Causal claim three requirements:

-Covariation-as one variable changes, the other changes -temporal dependence-the causal variable occurs earlier in time than the dependent variable -Internal validity-other possible explanations have been ruled out

Regression results

-Criterion variables and predictor variables -using B (Beta) to test for third variables

Survey administration-develop protocol and instructions -What steps will be taken in giving the survey

-How will participants be greeted -incentives? -Informed consent?

Correlational study

-Involves the measurement and determination of the relation between MEASURED variables -Useful when one or the other of the variables cannot be manipulated

Tendencies that may threaten construct validity:

-Leading question (the framing of the question affects responses to it) -double-barreled question (more than one question is posed, confounding what participants are actually responding to) -negatively worded question (people who do not drive with a suspended license should never be punished)

Additional strategies for valid observations

-Masking the research design -blending in -waiting it out -measuring the behavior's results -observing people ethically

Pearson product-moment correlation coefficient (r)

-Measure of correlation (ranges from 1 to -1) -Calculated only when both measured variables are quantitative

Multiple-regression analyses

-Measuring more than two variables and assessing relations -Regression results indicate if additional variables affect an observed correlational relationship -Main idea: "control for" one or more variable while examining the correlations between potential "third" variables -still, regression does not establish causation

Potential inaccurate correlation problems:

-Outliers -curvilinear relations -range restrictions -low N

Pattern:

Data from a variety of methods converging on a similar conclusion

Choosing a figure-quantitative information

Data that can be summarized with measures of central tendency and variability (scores on a test, scores on an inventory)

Two main branches of statistics:

Descriptive and inferential statistics

Criterion validity:

Does it correlate with key behaviors?

Face validity and content validity:

Does it look like a good measure?

Validity

Does it measure what it claims to measure (face, criterion, convergent)

Sampling

Does the sample of participants that we actually work with represent the population we intend it to?

Random sampling

Drawing a sample from a population with a random technique (increases external validity) -All members of the population are similar (in a particular respect) -characteristic of interest is normally distributed in the population -each member of the population has equal chance of being included in the sample

Why conduct correlational research?

Ethics, naturally occurring groups (unable to randomly assigned), other questions that cannot be controlled in a laboratory setting

Choosing a figure-categorical data:

Frequencies, yes or no responses

Sometimes, external validity is crucial in...

Frequency claims

Examples of indirect measurement

Hand washing behavior (after coming out of the bathroom we ask people via survey, interview, etc. if they washed their hands

Examples of direct measurement

Hand washing in a public restroom (we actually go into the bathroom, observe, record, and calculate the % of individuals that wash their hands

Negative correlation

Indicates that an increasing in one variable is accompanied by a decrease in the second variable (a perfect negative correlation has a correlation coefficient of -1)

Multivariate designs

Involve more than two measured variables

Convergent validity:

Is it related with other ways of measuring the same thing?

Internal reliability:

Is relevant for measures that use more than one item for a variable. A set of items has eternal reliability if it's items correlate strongly with one another (Cronbach's alpha is used to evaluate internal reliability)

Standard deviation

Is the square root of the variance

To maximize external validity, consider how, not how ______

Many. (As long as it is random, the advantage of more participants diminishes what sample size)

Categorical-- > ordinal-- > interval-- > ratio-- >

Mode median mean, standard deviation geometric mean, coefficient of variation

Measures of central tendency:

Mode, median, & mean

Causal claim:

One of the variables is responsible for changing the other

Measures of variability:

Range, variance, standard deviation

Reverse scoring

Rewrite some of the items so that a negative response represents agreement (control for yea-saying) or a positive response represents disagreement (control for nay-saying)

Biased sampling techniques- Snowball sampling:

Sampling in which participants are asked to recommend other participants for this study (the social network)

Biased sampling techniques- convenience sampling:

Sampling only those who are most easy to contact

Biased sampling techniques- Quota sampling:

Sampling where the researcher identify subsets of the population and then sets a target number (i.e., quota) for each category in the sample

Biased sampling techniques- Purposive sampling:

Sampling where the researcher only recruits a specific type of participants

Common ways to measure variables:

Self-report, observation (naturalistic and structured), sampling behavior with tasks and psychological response measurement

Indirect measurement

That which is measure is different from the focus of the research

Mean

The arithmetic average of a set of numbers

External validity:

To whom, what, or where can we generalize?

Interrater reliability:

Two observers rate the same participants and then an r is computed. If r is positive and strong (.70 or higher) and you have a good interrater reliability. If r is positive but weak, then you do not have good interrater reliability

Conceptualization

What the variable means to the researcher at a theoretical level

Observer bias

When observations are consistent with expectations

Observer effects

When participants behaviors systematically vary as a function of observers behaviors

Reactivity

When participants react to being watched

Cross-lag correlations

Whether an earlier measure of one variable is associated with a later measure of the other variable

Autocorrelations

Whether each variables is correlated with itself across time samples

Cross-sectional correlations

Whether two variables measured at the same point in time are correlated

Regression does not equal ________

causation

Correlational research-Relationships between variables are established, but extremely was variables are not controlled, thus, we are unable to __________ ___________

determine causality

Zero correlation

indicates a lack of relation between the two variables (the correlation coefficient for a "zero correlation" is zero or close to zero)

Reliability

is the measure consistent (test retest, internal, interrater reliability)

Random sampling does not equal

random assignment

biased sample

some members of the population of interest have a disproportionately higher probability of being included in the sample

Population

the complete set of individuals or events that we want to represent

Direct measurement

the phenomenon that is the focus of the research is exactly the same that is being measured

Random assignment:

used in experimental design to assign participants to groups at random (increases internal validity)

Scales of measurement:

Categorical and quantitative (ordinal, interval and ratio scales)

Parsimony:

The degree to which a good scientific theory provides the simplest explanation of some phenomenon

Variability

The extent to which scores distribute around the mean

Sample

The group that we select to represent the population

Likert type scales

The individuals answer a question by selecting a response alternative from the designated scale

Median

The number that divides the distribution in half

Operationalization

The researchers decision about how to manipulate or measure a variable

Yes-no questions

The respondent answers yes or no to the items

Forced-choice questions

The respondent must select between two alternative responses

Multiple-choice questions

The respondents must select the most suitable response from among several alternatives

Mode

The score in a distribution that occurs most often

The relationship between reliability and validity

The validity of a measure is not the same as its reliability