CHAPTER 3

Autonomy issues

- Children should give the researcher assent while people over 18 should give the re-
searcher informed consent
- The adults should also be free and have the ability to make their own informed deci-
sion to voluntarily participate.

Coercion - the practice of persuading someone to do something by using force or threats.

Any procedure that limits a person’s individual freedom is coercive

e.g. professors who force the class to take a survey so as to award them a grade.

Withholding information and deception

- Providing too much information may potentially invalidate the results of the study
- Bc the participants may behave differently and unnaturally if they are told exactly
what is being observed
- However, the participants have to be thoroughly debriefed at the end.
Deception – the active misrepresentation of information about the nature of a study
e.g. Milgrams study whereby they were lied that the expt was about memory and learning

Providing informed consent in this case will bias the participant’s response.

Milgrams study also illustrates a type of deception whereby the participants become a part of
a series of events staged for the purposed of the study.

Problem of deception is not only in lab research

Humphreys (1970) studied sexual behavior in public restrooms – tearoom trade

- Served as a lookout for those engaging in such activity
- Employed considerable deception
- Also invaded privacy ( followed people home)

Is deception a major ethical issue?

3 main reason for a decrease in the type of elaborate deception as in milgrams expt
- Researchers are more interested in cognitive variables than emotions – thus use simi-
lar methods to researchers in memory and cog. Psychology
- General level of awareness of ethical issues has led researchers to conduct expts in a
diff way
- Ethics committees at unis and colleges are now reviewing proposed research, Elabo-
rate deception is a last resort for approval when there ar no alt procedures. Along with
Institutional review board
Importance of debriefing
- Explanarion of the purposed of the research that is given to participants following
their participaition
- Need to explain why they were deceived if they were
- Allows the participants to ask any qns, thus have an ethical and educational purpose
 - Participants should leave expt without any ill feelings
- In the milgrams study – he did sloppy debriefing.

Institutional Review Boards

IRB is responsible for the review of a research conducted within the institution.
- Local review agencies of at least 5 members
- Each college and uni in the US that receives funding has an IRB

IRB defines research as a systematic investigational research development testing and

evaluating designed to develop generalizable knowledge

Systematic – research - formal protocol that sets an objective and a seq of procedures to
obtain it
- Must be purposefully and intentionally designed
Generalizable knowledge
- A person must have the intent to create new knowledge with the results

Exempt research – research in which there is no risk is exempt from review – anonymous
surveys, questionnaires an educational tests are all exempt research.
- Naturalistic observation – exempt
- Archival research – data is publicly available – exempt

Exempt must be determined by IRB

Minimal risk research

- Risks of harm to participants are no greated than risks encounterd in daily life or rou-
tine physical/physhological tests
- E.g. recording weight,tests of sensory activity voice recordings , diagnostic echogra-
phy
- Moderate exercise by healthy individuals
- Research on individual or gp behavior or characteristics of individuals. ( no manipula-
tion of participants behavior i.e. no stress fo the particiants
 Research with non human animal subjects

- Institutional animal care and use committee – charged w reviewing animal research
procedures and ensuring all regulations are adhered to

Being an ethical researcher – the issue of misrep

Fraud – fabrication of data

- Detected when other scientists cannot replicate the results of a study
- Allegations of fraud cannot be done lightly bit must have possible explanations .
Plagiarism – misrepresentation of anothers work as your own
- Writers must give proper citaition, quotation marks

Word for word plagiarism – copied word for word , no citation

Paraphrasing plagiarism – paraphrased words, idea not cited
RISK-BENEFIT
CHAPTER 8 EXPERIMENTAL DESIGN

In the experimental design, the researcher attempts to control all extraneous variables ( also
known as confounding variables)

- Done by placing participants in a controlled environment (lab setting)

Confounding and Internal validity

- The only diff btw the experimental group and the control group should be the manipu-
lated variable.
- A good experimenter should also write out any possible pitfalls (confounding varia-
bles)

Confounding variable – A variable that varies along with the independent variable.
Confounding occurs when the effect of the independent variable and uncontrolled varia-
ble are intertwined, therefore you cannot determine which of the variables is responsible
for the observed effect.

e.g. independent variable – Watching violent tv shows

dependent variable – The level of aggression expressed by children
confounding variable – lack of parental control

Good experimental design requires eliminating all possible confounding variables that
could result in alternative explanations.

When the results of an experiment can be confidently attributed to the effect of the inde-
pendent variable, the experiment is said to have internal validity

Internal validity is the accuracy of conclusions drawn about cause and effect

EXPERIMENTAL DESIGNS THAT PROVIDE THE HIGHEST DEGREE OF IN-

TERNAL VALIDITY

Basic experiments

Simplest experiments have 2 variables – independent and dependent variables, the independ-
ent variable has a minimum of 2 levels (experimental and control)

Researchers must make sure that those are the only differences between the two groups

- Can be done through:

√ Experimental control – keeping the extraneous variables constant

 √ Randomization – making sure that any extraneous variables will affect both groups
equally

BASIC EXPERIMENTAL DESIGN – POSTTEST-ONLY DESIGN AND PRETEST-

POSTTEST DESIGN

POSTTEST ONLY DESIGN

1. Must obtain two equivalent groups of participants,

2. Manipulate the independent variable
3. Measure the effect of the independent variable on the dependant variable

1. Choosing the participants

Prodecures used must achieve equivalent groups to eliminate any potential selection
differences (the people selected to be in each group cannot have any systematic dif-
ferences such as income level/racial differences)

The participants must be randomly assigned

2. Manipulation of independent variable

i.e. decide on the 2 levels of the independent variable – e.g. one group receives treat-
ment and another does not / one group gets more reward than the other/two qualita-
tively different conditions e.g meditation vs exercise

3. Effect of independent variable is measured

Same measurement procedure is used, so that the comparison of the two groups are
valid and possible. Any difference in the groups on the dependent variable must be
attributed to the independent variable .

A statistical significance test may also be used to assess the difference between the
groups
PRETEST-POSTTEST DESIGN

Difference between posttest design and pretest posttest design is that for pretest posttest, both
a pretest and posttest are administered to the participants

Makes it possible to ascertain that the groups were in fact equivalent at the beginning of the
experiment.

However, the precaution is unnecessary if participants were randomly assigned to the two
groups.

The larger the sample, the less likelihood that there is that the groups will differ in any sys-
tematic way prior to the manipulation of the independent variable. As a rule of thumb, a gen-
eral minimum would be 20-30 participants per condition.

POSTTEST ONLY VS PRETEST POSTTEST DESIGNS Commented [db1]: *

Disadvantages of posttest only :

(-) Equivalence of groups
Posttest only design concerns the equivalence of the group whereby though randomization is
likely to produce equivalent groups, this might not be the case with small sample sizes.
As such, the pretest allows a researcher to tell if the groups are in fact equivalent to begin
with.
Advantages of pretest-posttest design
(+) Emphasis on change is incorporated into the analysis of the group differences, extent of
change in each individual can be examined. (e.g. in a smoking reduction programme)
(+) Pretest is also useful when participants drop out of the programme – especially when it
spans over a long period of time.

This dropout factor is called attrition (the loss of subjects who decide to leave an experiment)
or mortality – is a threat to internal validity bc the mortality rate is related to the nature of the
experimental manipulation.

With the pretest, it is possible to examine if attrition is a plausible alternative explanation -

e.g. if the programme was about smoking reduction, if only the light smokers remain, we can
see if the attrition rate has a plausible explanation for what actually happened.

Disadvantages of pretest-posttest
(-) may be time consuming and awkward to administer a pretest in the given experimental
condition.  may result in the sensitizing the participants to figure out what you are actually
studying ( issue of reactivity/responding in a socially desirable manner)  becomes very dif-
ficult to generalize the results to the people who have not received a pretest.

Pretest can be disguised by administering it in a completely different situation with a differ-

ent experimenter, or embed the pretest in a set of irrelevant measures such that it is not obvi-
ous that the researcher is interested in a particular topic

SOLOMON FOUR-GROUP DESIGN

 Whereby half the participants receive the posttest only, and the other half receive the
pretest-posttest design.

If there is no impact of the pretest, the posttest scores will be the same in the two control
groups.
ASSIGNING PARTICIPANTS TO EXPERIMENTAL CONDITIONS

Independent groups design – An experiment in which different subjects are assigned to

each group. Also called between subjects design (comparisons are made between different
groups of participants)

Repeated measures design - each participant is measured after receiving each level of the
independent variable. Also called a within subjects design (comparisons are made within ght
same group of participants)

Independent Groups design

In an independent groups design, different participants are assigned to each of the conditions
using a random assignment (usage of random “chance” procedure to determine which condi-
tion an individual will participate in)

Random assignment will prevent any systematic biases and the groups can be considered
equivalent.

Repeated measures design

Having the same individuals participate in all the same groups

Participants participate in both conditions  e.g. participants might read read a low meaning-
ful material first, take a recall test then do the same w a high meaningful test

Meaningfulness – ease with which new material can be related to something that is already
known.

Advantages and Disadvantages

(+) fewer research participants are needed – especially when participants are scarce and when
it is costly to run each experiment
(+) Repeated measures designs are extremely sensitive to finding statistically significant dif-
ferences between groups – bc you have data from the same people in both conditions
(+) much easier to see the independent variable (cont)
(-) different conditions must be presented in a particular sequence

- Order effect whereby the order of presenting the treatments affects the depend-
ent variable. E.g.
- Practice effect; the performance on the second task might improve merely because of
the practice gained on the first task. (also known as learning effect)
- Fatigue effect ; resulting in a deterioration in performance from the first to the second
condition as the research participant becomes tired,bored or distracted.
- Carry over effect; effect of the first treatment to carry over to influence the response
to the second treatment e.g. if the independent variable is the severity of a crime, the
more severe one might seem much worse to participants than it normally would how-
ever, as they are exposed to more and more severe crimes, the first one would not be a
severe. Or another e.g. when the first condition produces a change that is still influ-
encing the person when the second condition is introduced. E.g. the first condition in-
 volves experiencing failure at an important task, this may result in increased stress re-
sponses temporarily. When should the second variable be introduced? When is it too
soon?

Counterbalancing Commented [db2]:

Important for repeated measured design, due to the order effects. Counterbalancing is a
method of controlling for order effects in a repeated measures design by either including all
orders of treatment presentation or randomly determining the order for each subject.

By counterbalancing the order of conditions, it is possible to determine the extent to which

order is influencing the results. Extended to experiments with 3 or more groups (with 3
groups there are 3! = 6 possible orders )

Latin Squares

Technique to control order effects without having all possible orders is to construct aa latin
square : a limited set of orders constructed to ensure that
1- Each condition appears at each ordinal position and 2- each condition precedes and
follows each condition one time.

No. of orders in a latin square is equal to the number of conditions (i.e if there are 4 condi-
tions, there are 4 orders) , no. of participants tested in each order must be equal

Time interval between treatments

- Reseachers need to carefully determine the time interval between presentation of

treatments and possible activities between them.
- E.g. a rest period may counteract a fatigue effect
 - E.g. for drug tests, the time interval will be the period of time taken for the drug to
wear off
- However, this may result in a low attrition rate thus lowering the internal validity of
the expt
Choosing between independent groups and repeated measures designs

- You cannot use a repeated measures design when a procedure has irreversible effects
e.g. lobotomy,brain tissue removal

Matched pair design : instead of simply randomly assigning participants to groups,

you can match the people on a participat variable such as age or personality trait.

Matching variable will either be the dependent variable or a variable strongly related
to it.

Might even be matched on cognitive ability

Ensures that the groups are equivalent prior to introduction of the indep variable
Good esp when sample size is small
Result is a greater ability to detect a statistically significant effect of the independent variable
because it is possible to account for individual differences in responses to the independent
variable.

However, matching can be costly and time consuming because they require measuring the
participants before the expt takes place

Matched pairs design is most likely to be used when there are only a few participants availa-
ble

Points to take note of (chap 8)

CHAPTER 9 CONDUCTING EXPERIMENTS

Practical aspects of a research investigation

- Samples may be drawn from a probability or non probablility sampling technique

- Determination of sample size – generally, increasing the size of a sample increases
the likelihood that your results will be statistically significant

Manipulating the independent variable

- In order to manipulate an independent variable, you need to construct an operational

definition of the variable – turning a conceptual variable into a set of operations – spe-
cific instructions events and stimuli to be presented to the research participants
Setting the stage

- Supply the participants with the information necessary for them to provide their in-
formed consent to participate
- Includes the information about underlying rationale of the study
- You may also deceive the participants by saying that the study is intended for some-
thing else bc of the idea of reactivity – if your participants figure out what you are
studying, they may employ a response set that is socially desirable/acceptable  af-
fecting internal validity of your experiment
- If deception Is necessary, you have to debrief the participants at the conclusion of the
experiment
Types of manipulation

- Straightforward Manipulations
a. Researchers are able to manipulate an independent variable with relative sim-
plicity by presenting written,verbal or visual material to the participants – such
stimuli are known as straightforward manipulations
b. May be presented verbally, written form, videotape or with a computer.
- Staged manipulation
a. Reasearcher is trying to create some psychological state in the participants e.g.
frustration,anger or a temporary lowering of self esteem
b. Necessary to stimulate some situation that occurs in the real-world e.g mil-
gram obedience experiment
i. Employ a confederate- aka accomplice who is part of the manipula-
tion, this confederate is used to create a particular social situation.
c. Also called a staged manipulation or event manipulation
d. Example would be the Solomon asch line judgement task
- Strength of the manipulations
a. A strong manipulation maximizes the differences between the two groups and in-
creases the chances of that the independent variable will have a statistically signif-
icant effect on the dependent variable.
b. A confederate is usually used – in one group the confederate can be extremely
similar to the participants and in another the confederate will be dissimilar.
c. Particularly important in early stages of research whereby the researcher is trying
to show a correlation
 d. Principle of using the strongest manipulation possible should be tempered by two
considerations
i. External validity of a study- strongest manipulation may entail a situa-
tion that raraly, if ever occurs in the real world.
ii. Ethics – manipulation must be strong but within the bounds of ethics e.g.
fear and anxiety for example might not be possible because of the poten-
tial physical and psychological harm to participants
- Cost of manipulation
a. Researchers w limited monetary resources may not be able to afford expensive
equipment,salaries for confederates or payments to the participants in long-term
studies
MEASURING THE DEPENDENT VARIABLE

Types of Measures
- Self report measures
a. Can be used to measure attitudes, liking for someone, judgements about personal-
ity characteristics, intended-behaviours, emotional states, attributons about why
someone performed well or poorly on a task confidence in judgement, human
thought and behavior.
b. May be on a seven point scale from very unlikely to very likely

- Behavioural measures
a. Direct observations of behaviors
b. Examples include Whether an individual responds to a responds to request for
help etc
c. Need to quantify observed behaviours – i.e. rate of behavior,reaction time,dura-
tion of behavior
- Physiological measures
a. Galvanic skin response (GSR) – measure of general emotional arousal and anxi-
ety,measures the ekectrical conductance of the skin which changes when sweating
occurs
b. Electromyogram (EMG) – measures muscle tension and is frequently used as a
measure of tension or stress.
c. Electroencephalogram (EEG) – measure of electrical activity of brain cells, can be
used to record general brain arousal as a response to different situations, such as
activity in certain parts of the brain as learning occurs or brain activity during dif-
ferent stages of sleep.
d. Magnetic resonance imaging (MRI) – provides an image of an individual’s brains
structure, allows them to
i. Functional MRI allows the researchers to scan areas of the brain while a
research participant performs a physical/cognitive task.
ii. Data provides evidence fot what brain processes are involved in these
tasks
e. Other physiological measures include heart rate, analysis of blood,urine,tempera-
ture
Multiple measures

- A variable can be measured in a variety of concrete ways

- E.g. employee wellness on health ( more independent measurs)
a. Employee wellness can be measured by several physiological measures such a
blood pressure,cholesterol etc
b. More measures leads to an increase in confidence
- If there are more dependent measures, the question of order arises
a. Do the order of the measures matter?
b. Does responding to the first measure affect response on the later measures?

Sensitivity of the dependent variable

- The dependent variable must be sensitive enough to detect differences btw groups
a. e.g. do you like this person (yes/no) vs how much do you like this person ( 5/7 pt
scale)
- Sensitivity – the ability of a measure to detect differences between groups
- Ceiling effect - When a task is so simple that everyone does well regardless of the
conditions that are manipulated by the independent variable i.e. independent variable
has no effect on the dependent measure
- Floor effect – Task is so difficult that no body performs well
E.g : study examining the effect of crowding on various measures of cognitive task perfor-
mance.
- in the study, there was no effect of crowding on cognitive performances but it Is also possi-
ble that the measures were too easy or too difficult to detect an effect

Cost of Measures

- Some measures may be more difficult than others

- E.g. physiological measures vs paper and pencil self reports

ADDITIONAL CONTROLS

- Additional control procedures are necessary to address other types of alternative ex-
planations
- Usage of a control group allows for elimination of a variety of alternative explana-
tions for the results therefore improving internal validity
- Two general control issues concern participant expectations and experimenter expec-
tations

Controlling for participant expectations

- Demand Characteristics
a. Are features of the experiment that might inform participants of the purpose of the
study
 b. When they know of the expectations of the hypothesis of the study, they might
then behave to what you might want as compared to their true selves
c. SOLUTION –
i. deception, whereby you make them think that the experiment is
studying one thing when it actually is studying something else
ii. May be done by having filler items ( questions unrelated to the ac-
tual hypothesis ) on the questionnaire
iii. Indirect observation/field observation which are unobtrusive
measures to minimize the problem of demand characteristics

- Placebo groups
a. To measure if the effect is caused by the measure e.g. drug given or the expecta-
tions of the participants
b. One group maybe given the actual drug while the other gets nothing.
c. Reduces the placebo effect – improvement caused by the participants’ expecta-
tions about the effect of the drug
d. A placebo group in this situation is given a pill or injection containing an inert,
harmless substance. If the same improvement is resultant of both groups, all im-
provement could be caused by a placebo effect rather than an actual drug.
e. However, you have to be compliant to ethics principles – e.g. you cannot take
away medication from someone in poor health by placing them on a placebo ( be-
neficence is violated )

Controlling for experimenter expectations

- Expectancy effects or experimenter bias
a. Any intentional or unintentional influence that the experimenter exerts on subjects
to confirm the hypothesis under investigation
b. Sources of experimenter bias
i. Certain words may be emphasized when reading instructions to one group
but not the other
ii. There can be subtle differences in the way the experimenter reacts to the
participants
- Solutions to the expectancy problem
i. Experimenters should be well trained and should practive behaving consist-
ently with all participants
ii. Run all conditions simultaneously such that the experimenter behavior is the
same for everyone
iii. Automate procedures by using computers, voice recorders etc
iv. experimenters who are unaware of the hypothesis being investigated. i.e. only
person conducting the expt
v. Single-blind experiment : the participant is unaware of whether a placebo or
actual drug is being administered
vi. Double-blind experiment : neither the participant nor experimenter knows
whether a placebo or actual drug is being administered

FINAL PLANNING CONSIDERATIONS

Research Proposals
- Includes a lit review, with a background for the study as well as details of the proce-
dure with its full intent
 - Allow people with grants to have a quick overview
Pilot studies
- Whereby the researcher does a trial run with a small number of participants
- Pilot studies reveal any possible mistakes, if there are any confusing questions etc
Manipulation checks
- An attempt to directly measure if the independent variable manipulation has the in-
tended effect on the participants
- E.g. manipulating anxiety – manipulation check tells you if the high anxiety group
were really more anxious than the low anxiety one. – you may then change your pro-
cedure to increase the anxiety in the high anxiety groups if they are not significant
enough
- If there are no significant results, a manipulation check helps you identify it and sub-
sequently

Debriefing

Important for researcher to have a debrief after every research such that any deception or pro-
cedures used are fully explained

Participants may be asked not to discuss the study

Allows participants to ask questions

ANALYZING AND INTERPRETING RESULTS

- By means of statistical methods

COMMUNICATING RESEARCH TO OTHERS

- Write a final report to make your work tangible
- With participant info,procedures used etc

Professional meetings

- Gain funding through sponsored meetings

- APA ir APS (Association for psychological science)
Journal articles
- Work is peer reviewed and then eventually submitted to a journal
CHAPTER 10 COMPLEX EXPERIMENTAL DESIGNS

Increasing the number of levels of an independent variable

- A study with only 2 levels may not provide much information about the exact form
of the relationship between the independent and dependent variables
- Only 2 levels to a relationship may not result in data exact enough to present a posi-
tive monotonic relationship – which is not a strictly positive linear relationship

FACTORIAL DESIGNS

- Since a number of variables are operating to affect behavior, researchers often manip-
ulate more than one independent variable in a single experiment
- All levels of each independent variable are combined with all levels of the other inde-
pendent variables.
- Simplest design – 2 X 2 factorial design ( 2 independent variables, each having 2 lev-
els each)
a. E.g 2 independent variables might be – confederate sociability and confederate
food intake

Low sociability High sociability

Low food intake Low sociability – Low food High sociability – Low food
intake intake
High food intake Low sociability – High food High sociability – High food
intake intake

2 independent variables result in 4 different experimental conditions

 Format for describing factorial designs

IV – Independent Variable

2 X 3 factorial design – has 6 conditions

3 X 4 factorial design – has 12 conditions

Interpretation of factorial designs

- Yield information about the effect of each independent variable aka main effect
a. In a design with 2 independent variables, there are 2 main effects
- Yield information about an interaction – Situation in which the effect of one inde-
pendent variable on the dependent variable changes, depending on the level of
another independent variable.
a. An interaction indicates that the effect of one independent variable depends on
the particular level of the other variable
b. i.e. the effect the independent variable has on the dependent variable is dependent
on the level of the other independent variable.
c. E.g. sociability (sociable and unsociable) vs food intake ( high and low food in-
take)
Factorial designs with manipulated and nonmanipulated variables

- A factorial design with both experimental and non-experimental design is called an

(IV X PV) design – Independent variable X Participant variable
- Participant variable – personal attributes such as age, ethnicity, participant sex, per-
sonality characteristics e.g. males and females and high and low scorers on a person-
ality measure
- Intro verts and extraverts in distraction conditions (silence and television) with the de-
pendent variable being reading condition
a. Students generally did better in silence but the extraverts did considerably well in
both situations of silence and television

OUTCOMES OF A 2X2 FACTORIAL DESIGN

There are 3 possibilities

- There may or may not be a significant main effect for independent variable A
- There may or may not be a significant main effect for independent variable B
- There may or may not be a significant interaction between the independent varia-
bles.
Graphs of variables with no interaction are usually parallel to each other while graphs of vari-
ables with interaction intersect at one point at least.

Examples of factorial designs

1. Effect of age of defendant and type of substance abuse use during an offense on
months of sentence
a. Independent variable – type of substance abuse ( marijuana vs alcohol)
b. Age of defendant – 20 vs 50 y/o
c. Dependent variable – Months of sentence
2. Effect of sex and violence on recall of advertising
a. Sex of participant – male/female (IV)
b. Exposure to violence – nonviolent vs violent video (IV)
c. Number of ads recalled (DV)

Interactions and simple main effects

Simple main effect analysis examines the mean differences at each level of the independent
variable.

Simple main effect of the variable averages across the levels of the other independent varia-
ble with simple main effects.

Look at the results in a general manner and get a visible pattern – that will be the simple main
effect.
- E.g. sociable vs unsociable – look for the general increase in the sociability

Assignment procedures and factorial designs

Two basic ways of assigning participants to conditions

1. In an independent groups design ( between subjects design) ,different participants are
assigned to each of the conditions in the study
a. A different group of participants will be assigned to each of the four conditions.
i. E.g. if you have a 2X2 design and want to have 10 participants in each
condition, you will need 40 different participants ( 4 ( no. of conditions) X
10)
2. In a repeated measures design(within-subjects design) , the same individuals partici-
pate in all conditions in the study.
a. The same individual will participate in all of the conditions by completing all the
tasks
i. E.g. you have a 2X2 design and you want 10 participants in each condi-
tion, you will

3. Mixed factorial design using combined assignment

- Usage of both independent groups and repeated measures procedures
i. E.g. participant variable – extraversion is an independent group variable
ii. Distraction (TV or silence) is the repeated measures variable therefore all partici-
pants are studied with both distractions and silence.
 a. Therefore, 10 participants are assigned to each level of the independent group
iv anf then the same 10 are used A1 are used for both levels of B and same 10
for A2 are used for both levels of B.

INCREASING THE NUMBER OF LEVELS OF AN INDEPENDENT VARIABLE

- Anxiety level and task difficulty on performance

a. The higher difficulty tasks may result in a higher anxiety level  poor perfor-
mance ( interactions present )
- Too many independent variables cause the results to become more difficult to inter-
pret
- Interactions w more IVs are more difficult to describe or understand

CHAPTER 11 SINGLE-CASE,QUASI EXPERIMENTAL AND DEVELOPMENTAL

RESEARCH

Single case experimental design

Subjects behavior Is observed over time during a baseline control period before the manipula-
tion is introduced, followed by a manipulation.

Reversal designs (also called a withdrawal design) ABA

- Treatment is introduced after a baseline period and then withdrawn during a second
baseline period
- ABA design (A – baseline period , B- treatment period )
- May be improved with an ABAB design
a. Problems with ABA
i. Single reversal is not extremely powerful evidence for the effectiveness of the
treatment  might be due to a random fluctuation
ii. Not ethical to end with a withdrawal that Is beneficial to the participant

Multiple baseline designs

- Due to the unethical nature to reverse treatment that reduces dangerous or illegal be-
haviors
a. Multiple baseline across subjects – behavior of several subjects measured over
time, manipulation is introduced at a different point in time
b. Multiple baseline across behaviours – several different behaviours of a single
subject are measured over time ,at different times, the same manipulation is ap-
plied to each of the behaviours.( e.g. a manipulation could be a reward system)
c. Multiple baseline across situations – same behavior is measured in different set-
tings such as home and work . a manipulation is introduced at a different time in e
each setting with the expectation that a change in the behavior in each situation
will occur only after the manipulation
Replications in single case designs
- May greatly enhance generalizability of the results
- Powerful alternatives to more traditional research design
- Valuable for someone who is applying some change technique in a natural environ-
ment
- Complex statistical analysies are not required for single case designs
Quasi experimental design
- Focuses on the effect of an independent variable that must be inferred from an experi-
menr with only one participant – single case experimental
- Pre-experimental and quasi experimental design that may be considered if it is not
possible to use one of the true experimental designs described in the chapter.
- Consider research designs for changes that occur with age

Quasi experimental designs – a type of design that approximates the control features of
true experiments to infer that a given treatment did have its intended effect

Quasi experimental design allows us to examine the impact of an independent variable on

a dependent variable.

One-group posttest only design

e.g. investigate whether sitting close to a stranger will casue the stranger to move away

Flaws :

e.g. the avg amount of time before people leave is 9.6 seconds.
- you cant interpret the finding
- other reasons for not moving or moving
- one group posstest only design – one shot study
- Lacks crucial element of an experiment – control or comparison group
- there is a lack of internal validity as there are no comparisons to draw cause and effect.
- This design may be used for testing the effectiveness of a programme.

One group pretest-posttest design

Measure the participants before the manipulation and after the manipulation
Flaws :

- If you find a reduction in smoking, you cannot assume that the result was due to the
relaxation training program.
- Design has failed to take into account several alternative explanations which are
threats to internal validity

Alternative explanations to the threats to internal validity

History
- Any event that occurs between the first and second measurements but is not part of
the manipulation
- Any such event is confounded with the manipulation
- Threat to internal validity of the experiment

Maturation
- Any changes that occur systematically over time
- Participants may become bored,fatigued,wiser and hungrier over a long period of time
- Children become more coordinated and analytical
- May cause you to mistakenly attribute the change to the treatment rather than matura-
tion
Testing
- A threat to internal validity in which taking a pretest changes behavior without any
effect on the independent variable
- E.g. smoking treatment measure might cause them to take note of the number of cigar
ettes may just cause them to reduce the number of cigarettes they smoke
- Thus in this case this may not mean that the treatment worked, but rather a change in
mindset
Instrument decay
- Basic characteristics of the measuring instrument change over time
- E.g. human observers used to measure behavior : observers may gain skill, become
fatigued, or change the standards on which they assess the participant
- E.g. smoking example whereby they have to record down the no of cigarettes smoked,
they may forget to record down a cigarette that they smoked etc thus leading to an ap-
parent reduction in cigarette smoking

Regression towards the mean

- Statistical regression (regression towards the mean : the principle that extreme scores
ona variable tend to be closer to the mean when a second measurement is made)
- Likely to occur when participants are selected due to their extreme scores
 - High scores  become lower, Low scores  become higher ( towards the mean)
- If smokers are selected to their extreme scores, regression towards the mean may be
the reason for their reduction of smoking
- Can be solved by having a control group

Non equivalent control group design – selection differences or selection bias

- A quasi-experimental design in which nonequivalent groups of subjects participate in
the different experimental groups, and there is no pretest.
- Separate control group – whereby control and experimental group are not equivalent
and not by random assignment
- Differences are a confounding variable

Nonequivalent control group pretest-posttest design

- Most useful quasi-experimental design
- Pretest is present for the experimenters to know of the differences in the participant
- Still no presence of random assignment
- You can still observe the changes in scores from pretest to posttest
- If independent variable has an effect, the experimental control should show a greater
change than the control group

Propensity score matching of nonequivalent treatment and control groups

- Matching the conditions of the participants such as health conditions with each other
- Still possible that the two groups are different on other variables that are not measured
-
**Interrupted time series design and control series design

- Design in which the effectiveness of a treatment is determined by examining a series

of measurements made over an extended time period both before and after a treatment
is introduced, treatment is not introduced at a random point n time.

Developmental Research designs

 - Concerning changes in reasoning ability as children grow older, the age at which
self-awareness develops in children ets
- Major variable is age
- 3 types – cross sectional , longitudinal and sequential method
Cross-sectional
- Persons of different ages are only studied one point in time
- E.g. examining how the ability to learn a computer application changes as people
grow older
- 20,30,40,50 year old participant are studied once by observing their performance
Longitudinal Method
- Same group of people observed at different points intime as they grow older
- Most popular : Terman Life Cycle Study – 1528 California school children w intelli-
gence scores of at least 135 were studied. Participants who called themselves termites
were measured on numerous aspects of their cognitive and social development in
1921 and 1922  provided description about the lives of highly intelligent individu-
als and disconformed negative sterotypes  used to study health and longevity as a
result of personality
- May also get people to recall what happended in the past but this retrospective ap-
proach has its own problems whereby they cannot remember the exact event.
Cross sectional vs longitudinal
- Cross sectional is more common , less expensive and yields results immediately
- Longitudinal is less common, expensive and takes a long time to yield results
Disadvantages to cross sectional designs
- Differences may result from cohort effects – a group of people born at thabout the
same time and exposed to the same societal events;
Disadvantages to longitudinal design
- People die, move or lose interest in the study
Sequential method

- Combination of the cross sectional and longitudinal design

- First phase begins with the cross sectional methods
a. E.g. minimally compare 55 and 65 year olds
- Then each each individual is tested at least one more time
e.g. identify 6 different age cohorts and examine their self esteem scores, then saw their
change in self esteem scores periodivcally over the next few years – allows for a range of
data from a wide range of ages.
CHAPTER 14 GENERALIZATION

- Participants are rarely randomly selected from the general population

- Generally chosen bc they are most available

College students
- Smart 1966 – over 70% of the articles published between 1962 and 1964 are based on
college students
- Potential problem is that such studies use a highly restricted population
- Sense of self identity is still emerging, developing social and political attitudes in a
state of flux, high need for peer approval, unstable relationships
- Students as a group are more homogenous than nonstudent samples
- Students are more similar to each other than adults are to other adults
- Affects internal validity
- E.g. college students are less conservative and rated women and ethnic minorities
more favorably

Volunteers
- Are crucial to any research
- External validity may be compromised as the data from volunteers maybe different
from what would be obtained with a more general sample.
- Rosenthal and Rosnow (1975) reported that volunteers tend to be more highly edu-
cated, be of a higher socioeconomic status,more in need of approval and more social

Online research
- Researchers can find potential participants through online survey designs
- Has implications on external validity – though a online sample is more diverse than
college students, internet users still represent a unique demographic  e.g. living in a
urban/suburban area,higher income and being college educated are related to more
time online thus this demographic may not generalize well with the population
Sex, gender, race and identity
- Sex and gender
a. Sex – biological
b. Gender- social roles and identities associated with being male and female
c. E.g. achievement motivation only studied males and contraception only related to
females  results cannot be generalized
- Race and ethnicity
a. Construct of race is made more complex by the independent concept of ethnicity
b. Race and ethnicity is very important in the census and surveys in which the data
are going to be used for making policy decisions for things such as housing and
education
c. External validity may be compromised whereby the resulrs may not be general-
ized
Location
- Participants in one locale may differ from those in another.
- E.g. extraversion vs introversion may differ in people in different countries
- Studies lack external validity
Culture
- Some theories and cannot be generalized to other cultures
- E.g. Japanese engage in self criticism which can be seen as relationship maintaining
whereas Americans work to maintain and enhance self esteem

Nonhuman animals

- Able to generalize because underlying biological and behavioural patterns

- Applications include biological bases of memory,food preferences,sexual behav-
ior,choice behavior and drug addictions

Generalizing across methods

- Person who conducts the experiment is a source of external validity
a. Participants behave differently with experimenters of a different sex
b. Warmer personality also produce different

Pretests and generalization

- Researcher can. Be sure that groups are equivalent on the pretest
- Enables researcher to assess mortality (attrition) rate whereby it is likely that they will
withdraw from the experiment
- However, it limits ability to generalize ro populations that did not receive a pretest (
not taking the pretest may cause them to behave differently

Generalizing from lab settings

- Allows experimenter to conduct experiments in highly controlled conditions, how-
ever, internal validity of the experiment is strong
- Lab settings are not similar to that of the real world and hence external validity is
compromised
- Field experiment may be done to improve external validity (natural setting)
- Generalizability increases when multiple methods are used

Supporting good external validity

- Generalization as a statistical interaction
a. Interaction occurs when a relationship between variables exist under one condi-
tion but not the other
b. When there is an interaction, you might not be able to generalize the study
c. Researchers can address issues of external validity that stem from the use of dif-
ferent populations by including subject type as a variable in the study

- Replication  replication of a research is a way of overcoming any problems of gen-

eralization that occur in a single study
a. Exact replications  an attempt to precisely replicate the procedures of a study to
see whether the same results are obtained
i. Frequent replications confirms its reliability
ii. External validity of the original findings is increased
iii. Single failure to replicate does not mean much, though it is unrealistic to
assume that one single failure invalidates the previous research
 b. Conceptual replications  use of different procedures to replicate a research find-
ing
a. Attempt to understand the rs among abstract conceptual variables by using
new,or different operational definitions of those variables
b. Important in furthering our understanding of behavior
c. Same variable is operationalized in a different way
d. When conceptual replications produce similar results, our confidence in gener-
alizability increases

Evaluating generalizations via lit reviews and meta analyses

- Literature review  a written summary and evaluation of the existing literature on a

specific topic
a. Summarizes what has been found
b. Tells reader which findings are strongly supported and which are only weakly
supported
c. Points out inconsistencies and lacking areas
d. Discusses directions for future research
- Organizing, integrating and evaluating previously published material, authors of liter-
ature reviews consider the progress of research toward clarifying a problem
- Allows for finding trends and directions for future study

- Meta analyses  another technique for comparing a large number of studies in an

area  the researcher combines the actual results of a number of studies

a. Analysis consists of a number of a set of statistical procedures that employ effect

sizes to compare a given finding across many different studies
b. You can draw statistical conclusions from this material
c. A method for determining the reliability of a finding by examining the results
from many different studies

Using research to improve lives

- APA president George miller spoke of psychology as a means of promoting human
welfare
- Allows research and improvement in health related behaviours related to stress,heart
disease and STI, Law and criminal justice ( showing how law enforcement officers
can improve accuracy of eyewitness identification ) education ( methods for encour-
aging academic performance) , work environments ( providing workers with more
control and improving how people interact with computers and machines wtd)
- Psychologists provide info about mental health,parenting etc via the internet
CHAPTER 12 UNDERSTANDING RESEARCH RESULTS : DESCRIPTION AND
CORRELATION

- DESCRIBING RESULTS
a. Comparing group percentages
i. Describing your results in terms of percentages e.g. 60% of males and
50% of females
ii. Continue by running statistical analyses
b. Correlating individual scores
i. Needed when there are no distinct groups of subjects
ii. Indiviuals are measured on 2 variables e.g. do people who sit near the front
have higher grades?
iii. Consider the rs between where they sit vs the grades they obtain
c. Comparing group means
i. Calculate the means and compare them in the control and non-control
groups – run statistical analyses on them

- Graphing Frequency distributions

a. Pie charts – good for data converted into percentages of a whole population
b. Bar graphs- showing 2 or more possible responses
c. Frequency polygons – useful to represent data in interval or ratio scales
d. Histograms- scale values are continuous and are quantitative

- Descriptive statistics

a. Central tendency  mean,median,mode

b. Variability  standard deviation measures the spread of the data

- Correlation coefficients (describing the strength of relationships)

a. Pearson product moment correlelation coefficient (used with interval or ratio scale
data )
i. Provides data about the strength of relationship between two variables and
indicates the direction (+/- 1.00 is the strongest of correlations, the sign in-
dicates the direction)
b. Correlations are visualized in the form of a scatterplot
c. Important considerations
i. Restriction of range  when the scores on a variable are limited to a small
subset of their possible values, thus internal validity and external validity
is flawed
ii. Curvilinear relationship  product moment correlation is designed to de-
tect only linear relationships
a. If relationship is curvilinear the pearson product moment correlation is
unable t detect the relationship

Effect size : extent to which two variables are associated  ranges from
0.00 to 1.00
 r^2 value is also sometimes referred to as the percentage of shared vari-
ance between two variables

Regression equations  calculations used to predict a persons score on one variable

based on another variable that is already known
When behavior is a variable, It is called a criterion variable
Predictor variable – variable that is used to make a prediction of an individuals score
on another variable

R^2 – multiple correlation coefficient

A correlation between one variable and a set of predictor variables

e.g. predicted GPA is a combination of college grades,GRE test scores,Gre subject

test scores, favourability of recommendation letters

Partial correlation – the correlation between two variables with the influence of a third
variable statistically controlled fr
e..g researcher is interested in a measure of number of bedrooms her person as an in-
dex of household crowding
then he administers a cognitive test to the children from the households  children in
more spacious houses score higher suspects social class to be a third variable 

Null hypothesis : population means are equal,observed difference is due to random error
Research/alternative hypothesis : population means are not equal,difference is not due to
chance

Statistical significance – rejection of null hypothesis when an outcome has a low probablility
of occurance

T and F tests

T test – commonly used to examine whether two groups are significantly different from each
other

a. Is obtained t has a low probability of occurance ( lies within the critical region) then
the alt hypothesis is rejected
b. T= group diff / within group variability

Most commonly used alpha level – 0.05

F test
- used to ask if there is a difference between three or more groups
- evaluate factorial designs for interactions or analysis of variance
- Used when there are more than 2 levels of an independent variable

Refer to book for the anova formula

Type 1 and 2 errors

Type 1 error – reject null hypothesis when null hypothesis is true

Type 2 error – Accept null hypothesis when null hypothesis is false

Choosing a significance level

Interpreting nonsignificant results

Power analysis
Power = 1-p ( type two error)

Major statistical programmes : spss, sas , systat

- Alllowf for accurate information on statistical significance, and also facilitate graphic
displays of data
Research with multiple independent variables