Research Methods

Using Correlations

Can't control variables, can't manipulate variables for ethical reasons. -use in personality/abnormal psychology. Twin Studies- nature/nurture debate

Variations on Interrupted Time Series

Combining nonequivalent groups design and interrupted time series- do many data points, but have one experimental group that experiences the interruption, one control group that does not. Switching Replications- a program (the same treatment) can be introduced in different locations at different times to different groups. There is no control group, but the design provides the benefit of built-in replication. If outcome pattern in location 1 matches that of location 2, can be more confident bout the generality of the phenomenon being studied. Measure multiple DV- absence of control group again, but measure several DVs, some expected to be influenced by the interruption, some not expected to change.

Threats to Time Series Design

No control group- changes could have been influenced by other factors History- did other events happen over the time period that could have changed behavior besides the known interruption. Instrumentation- techniques for scoring and recording data can change Subject selection- sometimes people quit and sometimes people join group being looked at over time.

Survey

a structured set of questions or statements given to a group of people to measure their attitudes, beliefs, values, or tendencies to act. Requires careful attention to sampling procedures -assume generalizability but we should replicate & extend findings to multiple samples

Manipulation Control

control the researcher exercises when selecting the IV -used only in experimental or quasi-experimental designs -decisions on control based on experience & expectations -different researchers control variables differently -consistency on implementing control can be a problem

Randomization Control

leaving some aspects to 'chance' Random sample- assumed to be representative Random assignment- some methods aren't actually random. Long term solution. "Sufficient" experience is necessary to ensure randomization

Use appropriate control based on

logical considerations theoretical frame of reference technical issues of design

Waitlist Control Group Designs

used in research designed to assess the effectiveness of a program or in studies on the effects of psychotherapy. -participants in the experimental group are in a program because they are experiencing a problem the program is designed to alleviate. wait list controls are also experiencing the problem. ex: evaluating types of therapy (relaxation and desensitization) to nightmares. Three groups, relax, desensitize & control. Control group had to be nightmare sufferers, too, but didn't get immediate treatment. At end of experiment, the control usually gets treatment for ethical reasons tho.

Regression Analysis

strong correlations allow us to make predictions about behavior. Regression line = "Line of best fit" y=ax+b. a=slope, b= y intercept used to predict a value for Y based on a given value of X. Y is criterion variable, X is predictor variable

Written Survey

survey through mail. Disadvantages- return rate is very low. not necessarily representative of population. Non response bias, social desirability bias Advantage- anonymity possible,

Phone Survey

survey via phone -most houses had phone -disadvantages- telemarketing and cell phones. Do not call lists arise -advantages- cheap, efficient, personal, more participants can be contacted over a shorter time

Interrupted Time Series

taking measures for an extended period before and after the event expected to influence behavior. -the treatment is the 'interruption' -number of measures taken before and after T will vary from study to study, also not necessary that the number of pre-interruption and post-interruption points be the same. The more data points, the better, recommended at 50. -Allows researcher to evaluate data trends- relatively consistent patterns of events that occur with the passing of time. -They eliminate need for control group -Serve to rule out alternative explanations of apparent change from pre-to posttest. (ex: because of event or because of general trends already occurring)

Two Disciplines in Psychology

correlational & experimental psychology correlational- concerned with investigating the relationships between naturally occurring variables and with studying individual differences. Observes variables & relates them. Asks the question how are people different? Experimental- minimizing or controlling individual differences to show some stimulus influences every individual 's behavior in a predictable way. Manipulates variables & observes outcomes. Creates general laws that apply to everyone One is not better than the other. Correlational studies only variance among organisms, experimental studies only variance among treatments. It is important that you study both, also need to be concerned with interactions between organismic variables and treatment variables

Case Study

detailed description and analysis of a single individual. Contains both qualitative and quantitative data (results of psychological testing). Useful in clinical work, in which the case of someone with a disorder is used to illustrate factions leading to disorder and treatment for it. -provide a level of detailed analysis not found in other research strategies. Can provide prototypical descriptions of certain types of individuals and disorders. Can provide inductive support for a theory. Serve the purpose of falsification.

Cross-Sectional Research

different age groups at the same point in time. Uses a between-subjects approach. -Advantages in that it takes a shorter amount of time. However a problem is cohort effects.

Effective Wording

don't be ambiguous don't use double barreled questions Don't use leading questions Do use simple questions Do use complete sentences Don't use too many negatives Do use balanced, non biased items Don't use abbreviations, slang, colloquialisms or jargon

Electronic Survey

e-mail, internet, e-survey, incentive surveying -advantage- large amount of data in short time. Internet open 24 hours for responses. disadvantages- survey appears as spam, gets ignored, self-selected, people will reply for a reason- not necessarily random opinions

Graph v Table

graphs striking if non-linear effects present. tables preferred when lots of data points. Also when author informing readers of precise values of means and standard deviations, as those would have to be guessed from graphs.

Yerkes-Dodson Law

great example of non-linear effect. Relationship between arousal and performance. Moderate levels of arousal leads to the best performance.

Participant/Subject Bias

having some sort of expectation from the subject that impacts the results of the experiment. "Good subject"- participants in the spirit of trying to help the experimenter and contribute meaningful results. This can bias data if they correctly guess the hypothesis and behave in a way to confirm it. "Demand Characteristics"- those aspects of the study that reveal the hypothesis being tested. Can reduce internal validity. "Evaluation apprehension"- subjects wants to be evaluated positively and so they may behave as they think the ideal person should behave.

Dependent/Paired Samples T-Test

if the independent variable is a within-subjects factor or if the two groups are formed in a way such that some relationship exists between them -matched groups design -repeated measures design

Archival Data

information gathered for a reason aside from the current research project. Could be public information like census data or previously collected research data. -often undergoes "content analysis"- any systematic examination of qualitative information in terms of predefined categories. Typically includes multiple coders and interrater reliability estimates -Benefits in that there is soooo much archival information available. There is no "reactivity" (the participants knowledge that their behavior is being observed can influence their behavior in ways that yield a distorted result). -Disadvantages in that some info may be missing to researcher, & may not be representative of some population. Also has experimenter bias- attending more closely to record that support one's hypothesis or interpreting content in a way biased to expectations. Does not allow for random assignment, so quasi experimental.

Factorial Designs

'factorial' is the same thing as "independent variable". These studies include more than one independent variable -when reported, each digit represents an IV, the numeric value of the digit represents the levels of the IV, ex: a 2x3 has 2 IVs, the first has 2 levels, the second has 3 - the total number of conditions to be tested in a factorial study can be identified by looking at all possible combinations of the levels of each independent variable. Produces a "factorial matrix" -can be completely between subjects, completely within subjects, or a mixture of the two

Multivariate Analysis

(Bivariate- relationship between two variables) examines relationships between two or more variables. Multiple regression- one criterion variable and a minimum of two predictor variables. Allows you to determine not just that they combine to predict some criterion, but the relative strength of the predictors. Factor analysis- a large number of variables are measured and correlated with each other. It is then determine whether groups cluster to form factors. Factor loadings- correlations between each of the measures and each of the identified factors.

Mixed Factorial Design

- a mixture of between-subject and within-subject variables exist. Simple mixed factorial means that the between subject IVS are manipulated variables. ex: maids in a group. Between subjects control- half told their work is exercise and provided nutritional information other half just think of work. Within-subjects factor- time. measures were taken at start of study and again 4 weeks later.1

Experimenter Bias

-Experimenter expectations on the hypothesis altering the results. -Experimenters testing hypotheses sometimes may inadvertently do something that leads participants to behave in ways confirming the hypothesis. -ex. treating research participants in the various conditions differently, such as frowning when they don't answer as the experimenter wanted

PXE Factorial Design & Mixed PXEct

-Subject and manipulated variables in the same study -Studies individual differences. - interaction between the type of person in the study and the situation in the environment. Environment broadly defined to include any manipulated dependent variable. Between subjects factorial, just some subject, some manipulated in IV -mixed PXE factorial- includes both a between-subjects and within-subjects factor. Between subject is the subject variable. When using subject variables- no causal conclusions. Can conclude that different types of people perform differently. With manipulated variables- causal conclusions with appropriate controls. Situation causes behavior differences. -If significant interaction- shows that for one type of individual, changes in the environment have one effect, while for another type of individual, the environmental changes have a different effect

Interactions

-remember threats to external validity- should generalize to other populations, environments, and times. -treatments-attributes interaction- which treatment is best for whom -treatments-setting interaction- effects of treatment setting on research -multiple-treatment interference -pretest sensitization -posttest sensitization -Treatment (IVs) interact with subject or manipulated variables which results in a decrease of generalizability

T-Test- Analyzing single factor designs

-used for interval or ratio data -difference between mean scores of each group and determines whether the difference is larger than would be expected from chance alone. -Assume normal distribution -Homogeneity of variance- means the variability of each set of scores ought to be similar. --corrections or non-parametric

Cohort Sequential Design

A group of subjects is selected and retested every few years, and additional cohorts are selected every few years and also retested. Enables comparison of overall differences among cohorts -add cohorts to research over time -creates cross-sectional & longitudinal data

Small N Designs

A small N is a small amount of participants. You could study the behavior of one person or yourself. -The strategy was to show the phenomenon occurring reliably for each person, not for the average person, in order to replicate the finding. -necessary because potential subjects are rare or difficult to find -Large N designs may occcasionally fail to reflect the behaviors of individuals. -Behavior must change due to treatment, not any other factor. Operationally defined-baseline (A), Treat & Monitor (B) -individual subject validity- extent to which research outcome applies to any one individual subject in a study.

Longitudinal Research

Same people, multiple points in time. uses a within-subjects or repeated measures approach. Have a problem with attrition- people quitting the study over time. Either from withdrawing, dying, or moving away.

Types of Single-Subject Designs

A-B design- A is baseline, B is treatment. ideal outcome is for the behavior to change when A changes to B Withdrawal Designs- ABA or ABAB, introduction and removal of the treatment to confirm the behavior is happening because of B. ABAB is considered more ethical because it ends with treatment in place. Multiple baseline designs- several baseline measures are established and then treatment is introduced at different time. 3 types- same treatment in different individuals, multiple behaviors in one individuals, and one behavior in multiple environments Changing criterion designs- inspired by shaping. The target behavior is too difficult for the person to reach all at once, so it must be shaped in increments. "steps" Alternating Treatment designs- evaluate more than a single treatment approach within the same study. After usual baseline is established, different treatment strategies are alternated numerous times (counterbalancing). Different treatments at different times.

Challenges of Observational Research

Absence of control- have to use operational definitions of behaviors to be observed, but much of the behavior is relatively uncontrolled and they must take what circumstances can provide Observer Bias- having preconceived ideas about what will be observed. Ambiguous behavior could be misinterpreted by observer who believes the study will go one way or another. Can use behavior checklists(ethogram)- predefined behaviors that observers are trained to spot. Time sampling- behavior is sampled at predetermined times. Event sampling- selects specific events for observation. Participant reactivity- behavior influenced by knowledge that you were being watched. Can be reduced by unobtrusive measures- when subject is unaware of measurements. Ethics- could be invasion of privacy. Public behavior, can't interfere with behavior, confidentiality & anonymity, IRB, informed consent

Nonprobability Sampling

Choose participants Convenience sampling- a group of individuals who meet the general requirements of the study and are recruited in nonrandom ways. Often from a "subject pool" Purposive Sample- a specific type of person is recruited for the study Quota Sample- attempts to accomplish same goal as stratified sampling (representing groups proportionally) but in non random fashion. Snowball Sample- referral sample- once a member of a particular group has been surveyed, the researcher asks that person to help recruit additional subjects through a network of friends. Usually biased

Controlling Order Effects

Counterbalancing- use more than one sequence in a within-subjects design to control order effects. Testing once per condition- only allowing the subjects to evaluate conditions one time -compete counterbalancing- every possible sequence of conditions will be used at least once. Can be calculated from X!. Problem is that as the number of levels of independent variables increase, number of possible sequences increase dramatically -partial (incomplete) counterbalancing- sampling from the complete set of all possible orders. Pretty much just randomizing the order conditions for each subject. Testing more than once per condition- participants experience more than one test condition -Reverse counterbalancing-the experimenter presents the conditions in one order, and then presents them again in the reverse order. Problem is that the subjects can memorize the order -block randomization-every condition must occur once before any condition can be repeated. Within each block, the order of conditions is randomized. Eliminates the possibility that participants can predict what is coming next.

Graphs of Factorial Designs

DV goes in the y-axis one IV goes on the x-axis, other IVS are presented through a legend identifying the line type -if the lines on the graph are parallel, no interaction. If lines are nonparallel, an interaction probably exists.

Single Factor Multilevel Designs (within subjects)

Depending on how many times each condition is presented, all kinds of counterbalancing are available- if just once per subject can use both full and partial counterbalancing. if more than once, can use both reverse and blocked randomization.

Problems with Interpreting Correlations

Directionality- which variable causes the change? The existence of the correlation by its self does not allow one to decide the direction of causality. Can be fixed by a cross-lagged panel correlation. Observes correlations across time. Third-variable problem- confounds to the experiment are being offered as alternative explanations to the problem. There could be a third variable that causes the change in both variables you measured. Partial correlation attempts to control for third variables statistically.

Between-Subjects Design

Each group is exposed to 1 level of the IV -each subject is experimentally naïve -differences might be due to differences in group composition, not the IV -necessary when subject variables are being studied -each subject enters the study fresh and naïve with respect to the hypotheses to be tested ex: IV with levels A & B, one group receives A, one group receives B, neither will get both

Within-Subjects Design

Each participant is exposed to all levels of the IV, so fewer people needed to conduct experiment -sometimes called 'repeated measures' -eliminated variability between participants -order effects can skew data- the experience or altered circumstances could influence performance in later parts of the study. progressive effects- performances changes steadily from trial to trial. Carryover effects- the order in which the conditions are presented, independently of practice of fatigue effects, might influence the study's outcome ex: an IV with conditions A &B, both groups will get conditions A & B

Design Problems in Applied Research

Ethical dilemmas- relating to informed consent & privacy. Proper debriefing is not always possible. Validity- increased extraneous or confounding variables. Harder to control a field environment. Between Subjects- often impossible to us random assignment to form equivalent groups. Reduces internal validity by subject selection problems or interactions between selections and threats such as maturation and history Within Subjects- not always possible to counterbalance properly in applied studies. Hence, may have uncontrolled order effects. Attrition can be a problem for studies extending over a long period.

Three Ways of Presenting Data

Sentence/paragraph form- find for reporting results of experimental studies with two or three levels, but makes tedious reading as amount of data increases. Table- show results in table graph- DV on Y, IV on X, becomes a bit more complicated when more than one independent variable is used

Varieties of Survey Methods

Interviews phone electronic written challenge (social desirability)

How to choose right graph

Line graph- has a continuous variable. One for which a number of immediate values exist. Such as dosage level. Reasonable to interpolate between the points of the graph and what the effects of intermediate values might be. Best for multilevel designs. Cannot be used for discrete data. Bar Graph- best when there is a discrete variable. No interpolation can be done, and to connect the points with a line would imply an intermediate that doesn't exist. Can also be used for continuous data. -add error bars-indicate the amount of variability that occurred within each condition -Mark significant differences on graphs -Don't be misleading with weird scales

Outcomes of Factorial Designs

Main effects- overall influences of each IV. The difference between the means of the levels of any one independent variable. Determining main effect involves combining all of the data for each of the levels of that factor and comparing it to the data of the other factor. Interactions- examine whether the variables combine to form a more complex result. Effect of one IV depends on the level of another IV. Provide more interesting results to study. Even if no main effects occur, an interaction can occur and produce an interesting outcome.

Correlation Coefficients

Pearson's r- calculated for data measured on interval or ratio scale Spearman's rho- used for ordinal data Chi-square- used for nominal data Descriptive statistics- just tell us the size & direction of the relationship, not if the relationship is significantly different from zero.

Cohort Effect

a cohort is a group of people born about the same time. Differences in not only chronological effects. but in terms of the environment in which they were raised in. For example, intelligence in 40,60,80? 80 year olds went to school in depression, 60 after WWII boom, and 40s on TV. These factors could bias results.

Factorial Matrix

a display of all possible combinations of the levels of each independent variable. Conditions- number of cells in the matrix levels- number of levels of an IV

Ex post facto design

a subject variable (such as gender, race, etc.) is being investigated. The subjects in the study are placed into groups "after the fact" of their already existing subject characteristics. Attempts to make the groups as similar as possible with reference to other variables. Different from random assignment because after matched pairs have been formed, they can still be randomly assigned to groups- in ex post facto designs- subjects are already in one group or another by virtue of the subject variable

Program Evaluation

applied research specifically designed to assess and influence public policy or special programs. -most concerned with the question, "Did X work?" -includes procedures for -determining if a need exists for a particular program, and who would benefit if the program is implemented, -assessments of whether a program runs according to plans, and if not what changes can be made, -methods for evaluating program outcomes -cost analysis to determine if program benefits justify funds spent.

Probability Sampling

each member of the population has a definable probability of being selected for the sample. Representative samples should represent the whole population. If not, considered biased- only surveying one group, SEC, etc. Self-selection bias- People select themselves to return the survey, creating abnormal or undesirable conditions in the group. survey will appear, along with appeal to reply, the results then try to establish validity. Impress you with total number of returns rather than the representativeness of the sample. Reason behind responses

Random Sampling

each member of the population has an equal chance of being selected to participate. Often effective in creating representative samples & fairest way to choose. Problems: you might want a specific feature of the population in your sample.(stratified sampling) Procedure may not be practical if your sample is very large.(cluster sampling)

Yoked Control Group Designs

each subject in the experimental group, for one reason or another, participates for varying amounts of time or is subjected to different types of events in the study. Each member of the control group is matched, or "yoked", to a member of the experimental group so that, for the group as a whole, the time spent participating of the types of events encountered is kept constant. ex: Giving experimental group who experienced trauma electroshock therapy until they reached stress level of zero. The control subjects were given matching amounts of therapy according to times from experimental group and measure results.

Elimination or Inclusion Control

elimination- variables converted to "constraints"- based on the specifics of the study- makes less generalizable Inclusion- include the extraneous variable in design to study the potential impact-creates a factorial design- makes more generalizable

How to Create Equivalent Groups

equal to each other in every important way except for in the levels of the independent variable -random assignment- every person volunteering for the study has an equal chance of being placed in any of the groups being formed. Helps equally spread out unaccounted differences between participants -blocked random assignment- a procedure ensuring that each condition of the study has a participant randomly assigned to it before any condition is repeated a second time. -matching- used when random assignment doesn't create equivalent groups. Participants are grouped together on some subject variable, called a 'matching variable' Sometimes used when the number of subjects is small and random assignment alone is therefore risky and might not yield equivalent groups. Must be confident that the matching variable is correlated with the dependent variable. Also must be a reasonable way of measuring or identifying participants on the matching variable. One difficulty is thte number of matching variables to use

Controlling Bias

experimenter bias- mechanize procedures as much as possible. Objectivity of scoring. Protocols- highly detailed descriptions of the sequence of steps that experimenters should follow in every research session. Use a double blind procedure- the experimenters and the participants are kept in the dark about what to expect. Subject Bias- reduce demand characteristics to a minimum. Can be done through deception. Use a placebo group. Manipulation check- asking the participant to indicate what they believe the true hypothesis to be. Conduct field research- if participants are unaware they are in a study, they are unlikely to spend anytime reacting to the experiment.

Making Research Meaningful

explanatory research tries to explain variability of DV by attributing it to the IV while ruling out other explanations Types of variables- Explanatory- IV & DV that are research is focused on Extraneous- controlled, confounded, randomized

Problems with Single-Subject Designs

external validity- the extent to which results generalize beyond the specific conditions of the study and replicate consistently. Statistics- don't use statistical analyses but rely on visual inspection of the data Can't test for interactions- most of the times one cannot occur without the other Selecting the DV- reliance on rate of response- usually doesn't consider other important DVs like RT, correction, amount of time spent looking

Interview Survey

face to face surveying -advantages in being comprehensive, yielding detailed information, interviewer can clarify question on spot -disadvantages in sampling, many portions of population refuse to be interviewed, or live in area interviewer would prefer to avoid. Cost, logistics, and interviewer bias.

Monitoring Programs

formative evaluations- monitoring the progress of a program while it is in progress Program audit- providing data on how the program is being used, looking at whether the program as described in the agency's literature is the same as the program that is actually being implemented.

Applied Research

goal is to solve a specified problem. Usually takes place in clinics, SS agencies, jails, gov't, some sort of field research. Address problem & provide psychological explanation as to why it happened.

Placebo Control Group Designs

members of a test group are led to believe they are receiving treatment when they really aren't. Could be a fake drug or deception of procedure. -Shows if participants held a belief in the experiment. Ex: give fake drink and tell them its alcohol. People expect alcohol to slow down reaction time and so would behave slower when given the placebo

Analyzing Multilevel Designs (ANOVA)

multiples t- tests would increase type-1 error- the more t tests you calculate, the greater the chances are of having one accidentally yield significant differences between conditions. ANOVA (ANalysis Of VAriance)- tests for the presence of an overall significant effect that would exist somewhere between the levels of the independent variable. To determine where exactly the significance lies requires use of post hoc analysis. EX: In a study with three levels, the null hypothesis is "level 1= level 2= level 3". Rejecting the null hypothesis does not identify exact location of significance. subsequent testing would analyze each of the three pairs of comparisons, but only after the overall ANOVA has indicated some significance exists. Yields an F ratio- examines the extent to which the obtained mean differences could be due to chance. or are the result of some other factor (presumably the independent variable) Inferential Statistic=Variability between conditions (systematic+error)/Variability within each condition (error) One way ANOVA for independent groups- multilevel independent groups design & multilevel ex post facto design One way ANOVA for repeated measures- multilevel matched groups design & multilevel repeated-measures design Factorial ANOVA- F ratios for main effects and interactions. Simple effects analysis- compares each of the levels of one factor with each level of the other factor.

Effective Survey

must answer empirical question and its terms. Open-ended questions- response beyond yes or no Closed question- answered with yes or no or simple options Likert Scale- type of interval scale for measuring closed question responses. odd number gives neutral options, even number forces choice. Allow don't know sparingly as an answer- forces response Add demographic data to include age, gender, socioeconomic status, material status, etc. Good, but more questions = higher chance people will get tired and tune out survey questions later on.

Quasi-Experimental design

no random assignment possible. Serve when ethical or practical problems make random assignment impossible. Often produce results with clear benefits to people's lives. -single-factor ex post facto designs with two or more levels -ex post facto factorial designs -PXE factorial designs (the P variable) -All of the correlational research -nonequivalent control group designs and interrupted time series designs

nonresponse bias

people who return surveys differ in some important way from those who don't return them.

Psychometrics for reliability & validity

split-half reliability- divide in half the items that make up a particular subtest and correlating the two halves. The correlation should be high if the test is reliable. Test-retest reliability- the relationship between two separate administrations of the test. These reliabilities should have a high correlation. -used most in personality/abnormal psychology- differences in traits and disorders are investigated. -twin studies- can study twins who will be genetically similar or different in home environments (either different or similar). Interclass correlations- calculated whenever pairs of scores are said to be 'unordered. Calculates 'interobserver reliability'

Observational Research

produce descriptive information on behavior. Global v Specific- can observe a variety of behaviors vs focusing on a specific behavior. Control of environment- may be in laboratory or outside of the lab. Naturalistic observation- goal is to study the behaviors of people or animals as they act in their everyday environments. Sometimes semi-artificial environments are use (zoos). Important environment not be impacted by the experimenter's presence. Hide or hope participants habituate Participant Observation- researchers will join a group being observed or at least make their presence known to the group. Strategy is its power to get the investigator as close to the action as possible. Usually involve narrative analysis. However, participants know they are being studied, so do they show true behavior?

Non-equivalent Control Groups Designs

purpose it to evaluate the effectiveness of some treatment program. those in the program are compared with those in a control group who aren't treated. Random assignment isn't possible and so in addition to levels of independent variable, the members of the control group differ in some way(s) from those of the treatment group. -groups not equal at start of study. In addition, they experience different events in the study its self. -best treatment graphs have treatment group below the control group, yet surpasses the controls by the end of the study. Regression to the mean can be rules out as causing the improvement for treatment group because one would expect it to raise them only to level of control group, not beyond it. Crossover effect is considered good evidence of program effectiveness.

Coefficient of Determination

r^2 -offers an explanation of variability- portion of the variability in one of the variables that can be accounted for by variability in the other variable.

Cluster Sampling

researcher randomly selected a cluster of people all having some feature in common.

Need Analysis

set of procedures for predicting whether a population of sufficient size exits that would benefit from the proposed program, whether the program could solve a clearly defined problem, and whether members of the population would actually use the program -census, surveys, key informants, focus groups, and community forums.

Single Factor Multilevel Designs (Between subjects)

single factor study using three or more levels. Include both between and within subjects designs. Have same four types. Advantage: nonlinear effects- adding more levels can give different results, ruining a linear relationship. Possible one doesn't exist until after level three, when you saw simple linear between one and two earlier. More informative and provide more complex outomes- eliminate explanations that compete with researchers preferred explanation.

Single-Factor Designs

single independent variables with two or more levels.

Social desirability bias

sometimes people respond to a survey question in a way that reflects not how they truly feel but how they think they should respond. Can be eliminated by anonymity.

Nominal or Ordinal Data

sometimes t-test, sometimes non-parametric

Stratified Sampling

the proportions of important subgroups in the population is represented precisely.

Threats to Validity of Nonequivalent Control Group Design

threats to validity of nonequivalent control group design -Matching->Match from two different populations, creating non-equivalent groups. when two groups are sampled from populations that differ on the factor being used as the matching variables. If this occurs, matching can enhance the influence of the regression to mean problem and make it appear that a successful program failed. - regression to mean-> the experimental group is typically formed from those with the greatest need for the program because their skills were so week. It is assumed though that there were some who had very high scores, and those simply regressed towards the average score.

Basic Research

to increase our core knowledge about human behavior and mental processes- knowledge may eventually have practical application. Knowledge valued as an end itself. Often takes place in laboratory.

Basics of Correlation

two variables are associated or related -correlation coefficient ranges from -1.00 to +1.00. Absolute value indicates strength f relationship. small=.0, medium=.30, Large =.50+. Sign indicates direction of relationships. Positive- direct, negative= inverse.

Response acquiescence

type of response bias in which you have a tendency to agree with statements. To avoid, balance favorable and unfavorable statements.

Statistical Control

use statistical procedures to control for extraneous variables- ANCOVA looks at covariance, decreases error. Partial correlation- partials out a variable

Matched Group Design

used when the independent variable is manipulated and a matching procedure, followed by random assignment was used.

Independent groups Design

used when the independent variable is manipulated and simple random assignment is used to create equivalent groups

Repeated measures design

used when the independent variable is tested within subjects. Each participant in the study experiences each level of the independent variable.

Independent Samples T-Test

used when the two groups of participants are completely independent of each other. Occurs when we use random assignment or if the variable is a subject variable involving two different groups. -independent groups design -ex post facto design

Scatterplots

visual representation of correlation data allows you to look for potential outliers and see general trends in data -weaker relationships- data is more random -stronger relationships- data approaches a line -Focus on linear relationships, but know that non-linear relationships do exist Restricting range can imply a relationship that doesn't exist or isn't representative. Weakens your correlation. Outliers- scores that are dramatically different. Can result in a type 1 error.