within subjects design some times called this

research study repeats measures of the same individuals under different conditions

compares two or more different treatment conditions (or compares treatment & Control)

same group of individuals in all treatments

looks for differences between treatment conditions within same group of participants

Confounding from environmental variables - ex. one group evaluated in the morning and the other in the afternoon

Confounding from time related variables - time between the first measurement and the final measurement.

time related confounding variables

history

 environmental events other than the treatment that change over time ;

only becomes confounding if it influences at least one treatment condition different and enough of the participants to have effect on group

longer trial, outside event could effect one day treatment and not next

time related confounding

maturation

particularly concern with elderly/children ;

treatment may be effected by new knowledge/skills/strength etc.

research situations that occur over long period of time

time related confounding

Instrumentation

more likely to occur with behavioral observation measures

particular concern when treatment is over period of time

same behavior might be judged differently at different times.

inconsistent measuring means

regression toward the mean

statistical regression

extreme scores on any measurement to move toward mean when measurement is repeated

threat because changes that occur in participants scores from one treatment to the next can be cause by regression instead of the treatment

Order effects

when ppl tested in series of treatment conditions, one treatment may influence following treatments

practice - participant gains experience through repeated tests

fatigue - decline in performance as treatment continues

change in participant behavior/performance related to general experience in research study

not related to specific treatment

ex.. practice effects/fatigue

ex of carry over effect

 in which the subjective perception of a treatment condition is influenced by its contrast with previous treatment

using:

randomization

holding them constant

matching across treatment conditions

• Nonexperimental research strategy

• Compares scores from different groups/conditions, do attempt to limit threats to internal validity

• Cohort effects/generation effects

Nonequivalent group design / Differential research

• Participant cannot use random assignment since there are two groups already formed (i.e. high school classes)

• Threat to IV: individual differences

• Differential research group design

• Compares pre-existing groups to establish difference

• Usually studies participant characteristics like gener, race, etc.

• Nonequivalent control group designs

• Uses pre-existing groups, one of which serves in the treatment condition and the other in the control condition. No random assignment.

• Posttest-only nonequivalent control group design

• Compares two nonequivalent groups. One group is measured after receiving a treatment and the other group is measured at the same time, but does not receive treatment.

• Pretest-posttest nonequivalent control group design

• Compares two nonequivalent groups. One group is measured twice, once before a treatment and once after. The other group is measured at the same two times, but does not receive treatment either time.
  

• Cross-sectional developmental design

• This research design uses different groups of individuals, each group representing a different age. The different groups are measured at one point in time and then compared. 

• Ex: comparing a group of 20 year olds to a group of 50 year olds

• Participants observed before and after treatment

• Does not have a control group

• Time-series design
  

• Design that has a series of observations for each participant before a treatment and a series of observations after. The treatment is a manipulation by the researcher.

• Longitudinal design

• Examines development by observing group of cohorts over time

• Ex: Group of 40 year old participants measured 20 years later at 60 years old.

• Pretest, posttest design

• Nonexperimental design where each participant in a single group is measured before and after the treatment to test the effect of the treatment

• summary value that describes a population (i.e. average score for population)

• summary of results

• mean, median, mode

• variance, standard deviation

• incorrectly accepting the hypothesis by stating that there is a significant result when there is actually no relationship. (can occur when the sample is unusual/extreme)

• stating that there is no significant effect when there actually is. (can occur when the effect is small)

·      Two treatment within subjects – A/B

o   Advantages;

§  Easy to conduct, results are easily interpreted, differences can be maximized, easy to completely counterbalance

o   Disadvantages;

§  Does not provide complete picture of the relationship between variables

·      ONLY Two points of data

o   Analysis; repeated measures t-test, are treatments significantly different

§  Correlational research strategy

·      Observe two variables as they exist naturally for a set of individuals

o   Scatter plot – strong positive, weak positive, moderate neg etc

§  truth of research/accuracy of results

·      Does study accurately answer the question intended

·      Verses validty of measurements

·      Single unambiguous explanation for the relationship between 2 variables

·      Extent research can be generalized to people, settings, times, measures, characteristics than those used in study

·      Multiple treatment design – A/B/C/D

o   Advantages; more likely to reveal functional relationship, more convincing cause/effect claim

o   Disadvantages; difference between groups may be too small to find significance, increased likelihood of attrition, difficult to completely counterbalance

§  Stop answering phone calls, die,

o   Repeated measures ANOVA, are any of the treatment means significantly different from each other, is so where. 

o   Each individual in one group is matched with a participant in each of the other groups, with respect to a variable considered to be relevant to the study

·      Within vs Between

o   Individual differences;

§  Problem: may become confounding variables or increase variance

§  Solution: within-subjects designs reduce this variance

o   Time-related factors and order effects

§  Problem: may distort results

§  Solution: between-subjects design only measure each individual once

o   Number of participants

§  Problem: some populations are hard to recruit or costly

§  Solution: within-subjects designs require fewer participants

·      Limitation to counterbracing

o   May distort treatment means

§  Not necessarily a problem, need to compare the means together

o   Asymmetrical order effects

§  Order effects may not take place evenly

o   Number of treatments                      

o   Switch to between subjects

§  When strong order effects are expected

simplest

least scientific 

superficial/face value of measurement procedure

subjective judgement and difficult to quantify

superficially appears to measure what it claims to measure 

when scores obtained from a new measure are directly related to scores obtained from an established measure of the same variable

consistency between two different measuring procedures of the same variable 

if the first, known, procedure is valid, we can infer the second also is. 

Just because two sets of measurements are related doesn't mean they are identical

demonstrated when scores obtained from a measure accurately predict behavior according to a theory

Construct validity 

an ideal/goal that develops gradually

based on research studies that use the same measurement procedures

Convergent validity 

demonstrate different measurement procedures "converge/join" on the same construct

Divergent validity 

measure two constructs to show there is little to no relationship

quality of a measurement procedure

based on the assumption that the variable being measured is stable or constant 

·      Between groups – disadvantages

o   Requires a lot of people

o   Individual differences between groups can become confounding variables

§  Alternate explaination

o   Individual differences in each treatment can create high variance

§  Obscures difference

·      Within groups- advantages

o   Removes potential for counfound due to individual or group differenes

o   Variability due to individual differences can be measured and removed

·      Within subjects – disadvantages

o   Confounding from environmental variables

o   Confounding due to time related factors

§  History, maturation, instrumentation, testing effects, regression

·      Kids grow up and learn things they already would have been learning, might night be attributed to the study at all

o   Switch to between subjects

§  When strong order effects are expected

·      Design does NOT eliminate individual differences, they are measured and removed with statistical analysis but they do exist. **

individual differences may become confounding variables or increase variance 

MAJOR disadvantage of between subjects design 

use within subjects if high difference detected

time related factors and order effects. 

within subjects design, 

between needs a lot of people

matched subjects design 

ex, two boys get the same results on first test. treatment applied to both and then new scores are compared between the two boys. 

typically involve comparison of scores from different groups or conditions

not manipulated

usually a participant variable (college v no college) or time variable (before v after) 

No attempt to control threats to validity 

no manipulation

between subjects design

non equivalent group design

Compares preexisting groups of indiviudals (Not randomly assigned)

ex...

Differential research

Posttest - only non equivalent control group design

Pretest - posttest non equivalent control group design

cross sectional developmental design

within in subjects

Pre-Post designs

compares two or more scores for one group of participants 

pretest - posttest design

time-series design

longitudinal developmental design

classified as quasi experimental

in pretest posttest design you can see if any trends happen before treatment even starts. compared to after 

can observe how behavior changes as people age without waiting for a group to grow older

data can be collected in short period of time 

no long term cooperation between and participant 

individuals not followed over years

subjects grew up in very different times, 

·      individuals that have lived in similar environments

differences between cohorts caused by unique characteristics or experiences other than age    

experimental research design.. uses results from SINGLE participant or subject to establish existence of cause/effect relationship

most developed by behaviorists examining operant conditions

beginning variable, represented by letter A

ex. resting heart rate before treatment 

multiple measures obtained during baseline and each treatment condition

ensure no outside factors influencing behavior until researcher intervenes and changes to next treatment condition

results as presented do not represent true experiment 

no control over extraneous variables

hard to prove internal validity 

before treatment/after treatment results may simply be chance 

series of observations of the same individual under the same conditions

could be baseline or treatment 

observations made when treatment is administered

identified by letter B

occurs when a series of measurements are all approximately the same magnitude

same data clusters around horizontal line

Change in average level

immediate change in level

change in trend

latency of change

** most convincing evidence for difference between phases occurs when the data show a large immediate change 

series of phases including baseline phase followed by treatment phase and at least one replication of a baseline, followed by treatment 

ABAB

could be unethical depending on the testing

may not change behavior

alternative technique that eliminates need for a return to baseline. 

well suited for evaluating treatments with long lasting/permanent effects 

est cause and effect relationship between treatments and behaviors using only single participant

flexability 

single case design - more control, experimental

case study - observational 

only describes ONE participant, hard to generalize

multiple observations, possibility of sensitization

no statistical controls, 

help organize, summarize, and simplify results obtained from studies    

o   methods that use the results obtained from samples to make generalizations about population

·      Corresponding summary value that describes a population

o   Ex. Average score for a population, GPA, age of students, favorite color of children

·      value that describes a sample

o   Averages, students at YOUR college, GPA of students in YOUR caf, etc

·      Statistic (____) has a corresponding parameter (_____).

computed for a sample

entire population

set of scores grouped into organized display that shows entire set    

shows a scale of measurement (categories) along horizonal axis, and frequencies on the vertical. 

o   Four types of measurments

§  Nominal, ordinal, interval, and ratio

o   bar graphs. Height is the frequency, bars touch

o   shows point above each score that the height indicates the frequency

Bar Graph

o   When categories on scale are not numerical, they are represented as a ....

ways to summarize all data using single number to define center of distribution    

·      used whenever mean is used.

o   Approx. average distance of scores from mean

·      score’s distance from the mean

o   Deviation score – X – mu

o   Standard deviation – squared variance

o   Variance  = standard deviation^2

·      measure of how spread out the numbers are  - Greek letter sigma

o   Square root of the variance

·      average of the squared differences from the mean.

·      Correlation Coefficient (r)

o   Direction – Positive (+) or negative (-)

o   Degree – Number (0-1)

o   Form – Type of Correlation (Pearson [r]/Spearman[r s]