Neur 2001 Midterm 2

Front 1

"Validity of a Study"

Back 1

Degree to which the studyaccurately answers the question it was intended to answer

Front 2

"A Threat to Validity"

Back 2

Anything that limits the study’s ability to answer theintended research question

Front 3

External Validity

Back 3

The extent to which the results obtained in astudy will hold true outside that specific study

ie. Will the results stand if I use:

a. Different Sample (results generalize to

population?)

b. Different Setting (different researchers at

different university)

c. Different Measurement (use physiological

measure vs self report)

d. Real world/ natural behavoir

Front 4

Internal Validity

Back 4

Whether there are factors about the study that raise doubts about the interpretation of results

-Any factor that allows for an alternate

explanation of results (ie. time of day vs

effect of music) is a threat to internal validity

**in order to assert 1 explanation for study researchers must eliminate compounding variables

Front 5

Example of Threats of External Validity

Back 5

1) Participants

2)

Front 6

Threat to External Validity Via Participants

Back 6

Participants may not be representative of their intended population due to:

1) Selection Bias (ie. only collecting in English)

2) Convience Sample (ie. Psych majors)

3) Participant characteristics (ie. Conservative)

4) Volunteers (??)

5) Cross-species generalizations (ie. Some species you can generalize from, others you cant)

Front 7

Threat to External Validity Via Featuresof the Study

Back 7

1) Novelty effect

-ie. novel environment affecting behavoir

2) Experimenter characteristics

-ie Experimenter’s behaviour affecting study variables

3) Multiple treatment interference ("testing effects")

-ie. fatique, practice or when participants experience more than one condition

Front 8

Threat to External Validity Via Measuring

Back 8

1) Timing of measurement

-ie. immediately or 6 months after

2) Sensitization

-ie. assessment sensitization- when the assessment of behaviours/thoughts/beliefs can have an effect

Front 9

Threats to Internal Validity via Extraneous variables

Back 9

Additional (extra) variables that change in theexperiment

- counfouding variables- 3rd party variable that interfere with the ability to interpret results

– These variables are generally not directly investigated

Any variable that can vary with the independent variable, & therefore, might effect the dependent variable.

Front 10

Threats to Internal Validity via Assignment Bias

Back 10

When participants one experimental condition arenoticeably different from participants in otherexperimental conditions

-ie placing more women in the worrycondition than the neutral condition

Front 11

Threats to Internal Validity via One Group Over Time

Back 11

1) Pt History- Stuff happens in participant’s lives that may affect your data

-ie. arriving day 2 with cold

2) Maturation– Systematic change in participant’s psychology or physiology

*Especially the case with children and the elderly

3) Instrumentation – Changes in a measuring instrument that occurs over time

-ie. Researcher becoming slightly different with the manner with which they observe

Front 12

Threats to Internal Validity via Testing Effects

Back 12

"Multiple Treatment Interference"- Any possible change in performance caused byparticipation in a previous treatment:

1) Fatigue Effect- when memory performance worsens because people are getting tired

2) Carry-over effect= when scores in the second condition are affected by the experience of the first condition

- know "what happened" in first condition

Front 13

Threats to Internal Validity via Regression Towards the Mean

Back 13

-Statistical phenomena whereby an extreme score is likely to be followed by a less extreme score

-Occurs because an individual’s score is a function of stable factors AND chance

ie. Someone with depression score of 60/63 at the start oftreatment, that depression score will come down even ifthe treatment is ineffective.

Front 14

Balance Between Internal and External Validity

Back 14

As you increase 1, you’ll decrease the other

∴ Researchers need to strike a balancebetween internal and external validity

**If study has highexperimental control and therefore high levelsof internal validity. Replicate with less experimental control to support external validation

Front 15

What affects both internal AND externalvalidity?

Back 15

1) Experimenter bias – Findings are influenced by experimenter’s expectations

2) Demand characteristics and participant reactivity – Participant acts according to the expectation of the study

Front 16

Use of Exaggerated Differences

Back 16

Experimenters often exaggerate the differencebetween groups in order to see an effect

ie. Compare and anger induction to a happy induction(rather than neutral condition)

Front 17

Designing a Research Project

Back 17

1) Chose research strategy

2) Set out research design

3) Define Procedure

Front 18

Research Stratagies

Back 18

1) Descriptive

2) Correlational

3) Experimental

4) Quasi-experimental

5) Non-experimental

Front 19

Research Design

Back 19

More specific parameters for study:

1) Will you have a group or an individual?

2) Will participants experience all conditions or will different participantsexperience difference conditions?

3) Which variables will you observe/manipulate?

Front 20

Research Procedure

Back 20

Detailed outline of exactly what you will do and exactly who willparticipate

-ie. 100 participants (aged 18 to 25) will be invited to participate....

Front 21

Descriptive Research

Back 21

Before you can start explaining why somethingoccurs, you must first be able to show that it evenoccurs

-ie. show that people "do" before ask "why"

Goal= to describe a phenomenon

Front 22

Types of Descriptive Research

Back 22

1)Observational:

– Have to decide how to observe behaviours, how to quantify behaviours you observe and in what settings to observe behaviour

2) Survey: Questionnaires & Interviews

3) Case study

Front 23

Observational Research Design

Back 23

Observe & systematically record behaviours forthe purpose of describing that behaviour

**can be used for correlational/experimental research if goal isn't to strictly describe behavoiur

Problem= observations can be subjective

Front 24

Limiting Observational Disruption in Observational Research Design

Back 24

1) Conceal yourself

2) Habituation= Allow participants to get used toyou & record behaviour after a certain amount of timebegin

Front 25

Limit the Subjectivity of Observations in Observational Research Design

Back 25

1) Have specific guidelines beforehand of whatyou’re watching for

-ie. Pre-existing list of aggressive behaviours that will count as “aggressive”

2) Have multiple observers

**need to establish the reliability of measurement via inter-rater agreement

Front 26

Methods for Quantifying Observations

Back 26

1)Frequency method– count the instances of a behaviour during a fixed time interval

-i.e.# of aggressive acts during first 30 mins

2) Duration method – record the amount of time an individual engages in a specificbehaviour during a fixed-interval period

-ie. Time spent playing in solitude / 30 mins

3) Interval method – divide the observation period into a series of smaller interval& record whether a specific behaviour occurs during eachinterval

-ie. divide 30 mins observation period could be split into 15 2-minuteintervals

Front 27

Sampling Observations

Back 27

Take a sample ofobservations instead of recording everything during complex situations

*(lots going on-becomes difficult to see everything)

Front 28

Methods of Sampling Observations

Back 28

1) Time sampling: – Observe during the 1st interval, thenrecord during the 2nd interval

2) Event sampling: – Identify 1 specific event to observe inthe 1st interval, then a different eventto observe in the 2nd interval

3) Individual Sampling:– Choosing 1 participant to observe during the 1st interval, then shift to anddifferent participant for the 2 interval

Front 29

Examples of Indirect Observtion

Back 29

1) Content Analysis- examine behaviour/events inliterature, movies, television, or othermedia

-ie. violence in cartoons

2) Archival Research- examine documents/ records to measurebehaviours or events that occurred in the past

-ie. count # of publications devoted toeach anxiety disorder from 1998 to 2008

Front 30

Types of Observations

Back 30

1) Naturalistic

2) Participant

3) Contrived

Front 31

Naturalistic Observation

Back 31

Observe behaviours in their natural setting

(+)=High external validity

=Useful for behaviours that areunethical or

impossible tomanipulate

(-)= may need to waituntil desired behaviour/ eventoccurs

Front 32

Participant Observation

Back 32

Interacting with participant & becoming one of them

-ie. Joining a fraternity for the purpose ofobserving behaviour

(-) = Interacting with participants might lead to changes in participant behaviour

= potentially dangerous

Front 33

Contrived Observations

Back 33

Setting up situation where the desired behaviour is likely to occur instead of waiting for the desired behaviour to occur

** usually conducted in laboratories (but not always)

(-)= Environment is less natural and may have unnatural consequences on behaviour

Front 34

Survey Research

Back 34

Asking directly about behavoiurs

– Better w/ information that people are more willing to disclose

**can be used for correlational/experimental research if goal isn't to strictly describe behavoir

Front 35

Types of Survey Research

Back 35

1)Questionnaires- can be done in person, online, by mail or by phone

2) Interviews- can be done in person or on phone

Front 36

Organizing Survey Research

Back 36

1)Develop questions

– What do you want to know about?

– How will you ask it?

– Are there good measures already developed that assess yourconstruct?

2) Organize the order of questions

3) Who will complete the survey?

4) How will they complete it?

Front 37

Response Format for Survey Research

Back 37

1) Restricted– respondent chooses from a list of possible answers

2) Open ended- respondent offers own answer

**more common w/ interviews

Front 38

Types of Restricted Response Questions

Back 38

1) Dichotomous

2) Ordinal Rating Scale

3) Rating-scale Questionnaires

4) Checklist

Front 39

Dichotomous

Back 39

Given two possible choices

-ie. yes/no, male/female, agree/disagree

Front 40

Ordinal Rating Scale

Back 40

Given a list & rank order their preferences

-ie. first choice, second choice...

Front 41

Rating-scale Questionnaires

Back 41

-Likert scales: 1- strongly agree...

**extremes= anchor

**Problem= participants use the same answer for most questions (may not accurately reflect thoughts if rushed)... use reversed scored items

Front 42

Checklist

Back 42

Given multiple response options and told to select allthat apply

ie. Indicate all countries you've travelled to

Front 43

Considerations While Constructing a Survey

Back 43

1) Demographic Info

ie. age, sex, ethnic group, marital status

2) Sensitive/embarrassing question

– ie. depression, alcohol, sex, drugs...

3) Appropriate language/vocabulary

-ie. How to ask a 12 year old v. 20 year old

Front 44

Considerations for Survey Questions

Back 44

1) Do your respondents have the necessaryinfo/background knowledge to respond to yourquestion?

2) Do you questions have the right specificity?

** too general can cause the information obtained to be difficult to interpret

-ie. start with “what are your thoughts about theHunger Games" and follow with rating, would they recommend etc.

3) Is the wording of questions appropriate?

a) Question clear (ie. medication vs treatment for headache)

b) Un-presumptuous (ie. what is your favourite alcoholic drink vs beer)

c) Appropriately personal (ie. do YOU think that parking is satisfactory)

d) Free of leading- (ie. Quebec separate vs sovereign)

**similar questions= different answers because people want to help your cause

Front 45

Recruiting Participants for a Survey

Back 45

Increase likelihood of response by introducing self/topic, make importance of topic clear and emphasizing the importance of their participation

-Mailing lists are available if you want to target a certain group

ie. single females between 30 and 35 who subscribe to the New York Times –

Front 46

Features of Interview Survey Research

Back 46

a) Questions asked in an open-ended, interview format

b) Conducted one on one

c) Time consuming

d) Small N

e) A lot of detail

f) Allow for opportunity for follow-up questions g) Necessary for certain populations (kids/ illiterate)

Front 47

Case Study

Back 47

The study of one individual for the purpose ofobtaining a description of that individual usuallyduring diagnosis & treatment

-Most used in clinical psychology

*useful for rare phenomena or situations in which experimental manipulation would be unethical

ie. dissociative identity disorder, traumatic accident

- allow for development of theories to explain phenomena/ behavoiur

Front 48

Case History

Back 48

Similar to case study but carried out when no treatment has been offered (yet)

-provide detailed description of a disorder & circumstancesthat surrounded it

Front 49

Validity Concerns in Case Studies

Back 49

1) Low Internal Validity– whether an abundance of causal explanations for observed behaviour

ie. speech difficulty being same regardless of neglect

2) Low External Validity– Would this same pattern of behaviour be observed with others?

ie. similar neglect of another child causing same deficits

Front 50

Correlation

Back 50

The relationship between 2 or more variables

Can be:

1) Bivariate or multivariate

2) Positive or negative

Offer a description of the direction & degree of arelationship NOT an explanation for it

Visualized in ascatterplot

-closer a point is to line of best fit "regression line", the stronger the association

Front 51

Scatterplots

Back 51

Use to visualized a correlation

-shows the degree & direction of associationbetween 2 continuous variables

1) X-Axis– Variable that the research believes does the predicting (similar to IV)

2) Y-Axis – Variable that the research believes gets predicted(Similar to DV)

Closer a point is to line of best fit "regression line", the stronger the association

Front 52

Information Gained from a Scatterplot

Back 52

1) Association- is there one and how strong?

2) Direction- if association is positive or negative

3) Linearity– is association linear or non-linear?

4) Presence of Outliers- data point that stands apart from the pack

– Outliers affect the magnitude of the relationship

**can find outlier and remove

Front 53

Correlation Coefficient

Back 53

1) Pearson’s correlation (r)

2) Spearman’s correlation (ρ)

Front 54

Pearson’s Correlation (r)

Back 54

A bivariate statistic that measures the degree of linear association between 2 quantitative variables.

Front 55

Spearman’s correlation (ρ)

Back 55

A bivariate statistic that measures the degree ofnonlinear association between 2 variables whereat least one is ordinal

Front 56

Cohen's Suggestion

Back 56

Used to determine strength of correlation if research does not determine otherwise:

Weak= 0 to .29

Moderate= .30 to .49

Strong= .50 to 1.00

Front 57

Point-biserial Correlation

Back 57

Where one variable is interval or ratio, & the othervariable is nominal with only 2 categories

-ie. Is there a relationship between gender & self-reportphysical aggression?

Front 58

Statistical Significance

Back 58

Determine whether or not your correlation is statistically significant

-ie. if relationship is "real" or not

Want to prove change in DV= due to IV manipulation not chance

...significant tells you if it’s likely this came about by chance, or if it’s unlikely this resulted from chance

Front 59

Determining Statistical Significance from Table

Back 59

1) df = N – 2 (**if given N look 2 below)

**round down if not specifically given

2) Use the .05 column (usually)

3) If r value is greater than the one i the table= statistically significant

**the bigger it is= the more likely it is caused by IV manipulation and not chance

Front 60

Shared Variance

Back 60

Reflection of what percentage of change in one variable can be accounted for by a change in the other variable

-shared variance= "percent change accounted for"

Calculation= r2 (=%)

Front 61

Correlation vs Causation

Back 61

Correlation does not necessarily imply causation

BUT if we know that 2 variables are related, then we can begin to make predictions:

a) if hours of studying & GPA= correlated, could make predictions about 1, 5h, 10h etc...

Front 62

Prediction vs Regression

Back 62

Regression= correlation

Difference= language:

a) A associated w/ B= correlation

b) A predicting outcome of B= regression

Front 63

Variables in a Regression

Back 63

1) Predictor variable(IV/ X):– Variable that the researcher believes influences change in the other variable

2) Criterion variable (DV/ Y): – Variable that the researcher believes is beinginfluenced by changes in the other variable

Front 64

Other Uses for Correlation Tests

Back 64

1)Test re-test reliability –

-ie. expect to find a statistically significant, positive, andstrong correlation between scores from Time 1 andscores from Time 2

2) Concurrent/convergent Validity

-ie. expect to find a statistically significant, positive, and strongcorrelation between scores from new measure & other measure ofthat construct (established measure or other)

3) Divergent validity

ie. Expect to find a statistically significant negative, & strongcorrelation between scores from one measure & scores from ameasure that assess an opposing construct (e.g., anxiety vsd relaxation scores)

Front 65

Experimental Research

Back 65

Main goal of an experiment is to establish acausal relationship

-ie manipulate of one variable (IV) to observeits effects on another variable (DV) whileholding other extraneous or confoundingvariables constant

Front 66

2 Types of Experimental Research

Back 66

1) Between groups

2) Within groups

Front 67

Between Groups Experimental Research

Back 67

Randomly forming 2 or more groups from apool of subjects/participants

Each group receives a different experimentaltreatment/ condition (level of the IV) & the groups arecompared

ie. Effect of advisory message

– Condition A: With warning, Condition B: No warning

- DV: Desire to see the movie

Front 68

Within Groups Experimental Research

Back 68

There is only 1 treatment group & each subject/participant is given all levels of the IV

-make before/after comparison between scoresobtained at different levels of the IV for thesame participants

ie. Diagnosis of cancer on emotional well being

Measure before and after treatment

Front 69

4 Basic Elements of Experimental Research

Back 69

1) Manipulation- manipulation of the IV by the experimenter

2) Measurement- measure of the DV by the experimenter

3) Comparison- comparison of DV from different conditions

4) Control- Experimenter controls the confounding variables

Front 70

Critical components of Experimental Research

Back 70

1) Independent and dependent variables

2) Experimental and control groups

Front 71

Independent Variable

Back 71

Believed to be the cause ...manipulated by the experimenter

Will have at least 2 levels (treatments) of the IV.

Can be experimental (researcher assigns participants to conditions) or non-experimental (participants alreadybelong to their conditions)

Front 72

Factorial Design

Back 72

Research with more than one IV

-ie. Effects of diet (IV #1) & exercise (IV #2) on weight

Front 73

Dependent Variable

Back 73

The variable that you measure in response tomanipulating the IV

- measured for each level of the IV andthen compared across conditions

Front 74

A Multivariate Study

Back 74

When the study has more than 1 DV

Front 75

Groups in Experimental Research

Back 75

1) Experimental group= participants exposed to the manipulation.

2) Control group= participants that are not exposed to the manipulationand are used for comparison purposes.

Front 76

Extraneous vs Confounding Variable

Back 76

1) Extraneous variable = variablesintroduced in the experiment that have no effect on IV or DV

**ie. if noise occurs across all treatments

2) Confounding variable= an extraneousvariable that varies w/ the IV & therefore canaffect the DV

** threaten internal validity (unclear whether the IV or CV causedchange in the DV)

** ie. if noise is present during only one treatment

Front 77

Identifying Confounding Variables

Back 77

1) Common sense

2) Previous experience

3) Previous research

4) Known confounds

Front 78

Types of Known Confounds

Back 78

1) Within-groups

– History

– Maturation

– Repeated testing

– Instrumentation

– Regression to the mean

2) Between-groups

– Assignment bias

– Attrition

– Compensation

Front 79

Confounding Variable due to Within-group History

Back 79

External events occurring between the 1st & 2nd testing session

**The longer the time interval between Time 1 & Time2, the more opportunity for outside events to occur

(ie. School program to reduce racism among High Schoolstudents • Time 1 = August 30th 2011, Time 2 = September 30th 2011)

Front 80

Confounding Variable due to Within-group Maturation

Back 80

Systematic change in participant’s psychology or physiology that is not related to the treatment

i

**More likely to occur w/ elderly & children

ie. physical growth, cognitive development, wisdom, etc...

Front 81

Confounding Variable due to Within-group Testing Effects

Back 81

Possible effects of the pre-test on the post-test (thesame measure taken at 2 different times).

** Scores onthe test may be partially due to experience

Note: Just because you have a pre-test, does not necessarilymean that it will affect your post-test. Just means that itcould.

Front 82

Confounding Variable due to Within-group Instrumentation

Back 82

Changes in characteristics of a measurement over time

-ie. more precise

Front 83

Confounding Variable due to Within-group Regression to the Mean

Back 83

The tendency for an extreme score to be followedby a less extreme score

Front 84

Confounding Variable due to Between-group Assignment Bias

Back 84

Bias introduced when groups are noticeably different on certain characteristics.

Front 85

Confounding Variable due to Between-group Attrition

Back 85

Loss of participants from 1 or more groups of a study

OK if it is affecting all groups equally but problematic if effect is unequal

**often differential=Participants w/ certain characteristics are lost

... who, why and when did drop out occur is important information

Front 86

Confounding Variable due to Between-group Compensation

Back 86

Untreated individuals or groups learn abouttreatment received by others & want similartreatment

**battle this by stressing importance of a control condition and assuring that if successful, they’ll get the treatment

Front 87

Effect of Experimental Control

Back 87

Experimental control means ensuring thatextraneous variables don’t become confoundingvariables

Front 88

Methods of Protecting Against Confounding Factors

Back 88

1)Remove them

-ie. warning participant of possibility to remove from "surprise" factor

2) Hold them constant

-ie. keep constant across treatments

*can also be done by keeping participant variables constant BUT= limits external validity

3) Use a placebo control

-Remove confounding effect of experimental method itself

4) Match them

- Identify potential confound & ensure that it isequal across groups

(ie. male to female ratio represented in breakdown of groups)

5) Randomize them

Assign participants randonly toconditions in hopes that potential confounds are alsorandomly assigned, will equal out and no longer confound

**larger N= more sucessful

Front 89

Placebo Effect

Back 89

Psychosomatic effect that causes a condition/treatment to yield results without active component

Front 90

Types of Placebo

Back 90

1) Psychotherapies- Actual therapy vs. nonspecific therapy

2) Surgeries- Real surgery v. mock surgery

3) Medications- Meds v. sugar pills

4) Food- Energy bar v. non-energy bar

5) Drink- Alcoholic v. non-alcoholic drink

Front 91

Gaining Experimental Control

Back 91

Ensuring that confounds are not varying with the IV

**experience high degree of internal validity

Certain methods of gaining experimental control will decrease your external validity

ie. Matching & Holding confounds constant

Front 92

Between Group Design

Back 92

2 or more groups of participants are formed atrandom

- Each group receives a different level of the IV& the groups are compared

**because each participant= one treatment= one score

-furthermore scores are independent of each other

Front 93

Problem with Between Group Designs

Back 93

People are Different... going to respond differently to same treatment/ manipulation

1) Assignment bias- participants in condition A are older thancondition B

2) When individual differencescreate too much variability within groups= Mask real effects

Front 94

Solution to Individual Differences in Between Group Designs

Back 94

FREE YES :)

Front 95

Unsystematic Variance

Back 95

Variation within treatment group

**within-group variability be relatively equal across groups

... many statistical tests assume that youhave equal variability within groups

BUT amount of unsystematic variance within a group can vary between groups

**want to minimize within group variance as much as possible

Front 96

Systematic Variance (F)

Back 96

Variation between groups

WANT variability/ high systemic variance (F) to be as high as possible- reflects effect of manipulating A (the IV) on B (DV) (vs chance)

**want AVERAGE score to be different

Front 97

Implication of In group vs Between Group Variability

Back 97

Want to minimize within group variability and maximize between group variability

Why? The bigger the difference in betweengroup variability compared to withingroup variability, the greater chancethat your group difference isstatistically significant