Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
73 Cards in this Set
- Front
- Back
Fabrication
|
don't run participants and make up data
|
|
Falsification
|
don't accurately represent the data
|
|
Plagarism
|
ideas of others that are reported, proposed, or reviewed as your own
|
|
Accepted use of scientifiec misconduct
|
need to do it unkowingly, unintentionally, or recklessly
must have evidence to back up misconduct or evidence it was not done on purpose |
|
Example of Misconduct
Wakefield |
1998 Lancet
link between autism and MMR vaccine (n=12) NOT REPLICATED NIH replicated it and found no correllational effect 2010, Lancet formally retracted this study. Evidence was fabricated |
|
Example of Misconduct
Marc Hauser |
Harvard U
"trialblazer of animal psychology" students turned him in on summer 2010 for 8 instances of misconduct article retracted from science journal all other research is now being questioned |
|
Example of Misconduct
Hwang Woo-suk |
South Korea
research on cloning stem cells cloned cow 2004- published in Science- stem cell made from other than embryo 2005- published Science- cloning and stem cells 2005- one scientists retracted his name so all of work is questioned 2006- elections and social issued were argued because of this guys research |
|
Example of Misconduct
Climategate |
(Al Gore) 2009
scientists found to manipulated data to support crisis 6 commitees did not find any evidence of fraud/miscaunduct Only showed pattern of failure to display a proper degree of openness |
|
Misrepresentation in Publication
|
take things out of context
make an inappropriate claim in correlation based on data selectivity of discussion of data is NOT ok |
|
Obligation to publish
|
even if you do not agree with the findings, you MUST publish
|
|
Citation Practices
|
must cite not only the lastest finding of journal, but also cite who found it first
also, it is good to give both sides of findings |
|
Methods Section
|
you must give every detail so anyone can re-do it
|
|
Results
|
discuss all data
show results in graphical form that represents data properly also with tables and figures |
|
Contributing Factors to Misconduct
|
There is an overwhelming obligation to get/ find significance
"publish or parish" |
|
Faster Findings
|
Don't have to put in as much work
|
|
Recognition for finding certain data
|
get attention that helps you further your career
the issue is lack of training which inevitably causes lack of knowledge |
|
Poor Supervision
|
Can lead to:
falsification of data fabrication of data misrepresentation of some data |
|
Reporting Scientific Misconduct
|
What's at stake?
the department is no longer looked upon positively power differential-- turn in teacher, they can possibly d things to affect your further career grant funding goes other studies may have been based off of the unethical one |
|
Whistleblowing
|
"ratting someone out"
|
|
Obligation to report misconduct
|
Any and all misconduct must be reported
|
|
Presumption of Innocents
|
anyone you are reporting, you must presume innocence b/c so much is at stake
|
|
APA Ethical Guidelines
|
Authorship
Follow Guidelines Errors Reviewer Ethics |
|
Authorship
|
Those who made the substantial scientific contribution to the study
who was most influential and ordered that way most work goes first, then less and less OR head author goes last to look better if masters or dissertation, YOU are first |
|
Original data must not be published elsewhere in multiple journals
|
this is due to copyrights
one journal at a time due to publishing copyrights |
|
Following APA ethical standards
|
Must keep all raw data, materials for 5 years
-- at regis it's only 3 years no plagarism! |
|
Error in APA
|
Can be corrected by retraction of entire journal OR formal correction
|
|
Reviewer Ethics APA
|
Peer-reviewed document by other scientist before it is published, so it is up to the reviewer to be ethical and never steal the ideas of the journal they are reviewing
|
|
Stats- Frequency distribution
|
Shows the overall pattern of data
can be displayed in table or graph OR can do relative frequency which is a percentage |
|
Histogram
|
bar graph
horizontal- class of data vertical-frequency can connect with dots if wanted |
|
Dot Plot
|
histogram with dots instead of bars
|
|
Descriptive Statistics
|
Used to descibe/summarize data
2 main types: central tendency dispersion (how spread out is the data?)- important in determining the importance of info |
|
Central Tendency
|
"Center" of distribution of scores
Mean, Median, Mode |
|
Mean
|
Average of data
X with bar over it |
|
Median
|
midpoint of distribution
remember to put data in numerical order |
|
Mode
|
bimodal or trimodal (or more) can exist in data
|
|
Dispersion
|
Range
Standard Deviation Variance |
|
Range
|
Difference between highest and lowest data
|
|
Standard Deviation
|
The measure of a sample of data that is used to represent the entire population the sample is taken from
|
|
Variance
|
The sum of the squared deviations from the mean, divided by the number of scores
|
|
Normal Distribution
|
Standard Bell Curve
In a normal bell curve, the normal distribution of mean, median, mode will be the same with different dispersion |
|
Population
|
complete correlation of those who could be measured
|
|
Sample
|
partial collection you DO measure
can take an infinite number of samples |
|
Distribution of sampling means
|
The means of the different means of sample groups all compared to one another
|
|
Standard error of mean
|
Standard Deviation of the distribution of the sample means
how much variability is there in the sample means |
|
Small Standard Error
|
Very close to the true population mean
usually, the bigger the n (sample size) the smaller the SE Law of Large Numbers -as the sample size increases, the mean of the sample becomes a better approximation of the entire population |
|
Test Hypothesis
|
Null Hypothesis
Experimental (alternative) Hypothesis If there is a significant difference between groups, we can reject the null hypothesis If there isn't a significant difference, we fail to reject the null hypothesis |
|
Null Hypothesis
|
the 2 conditions do not differ
no effect of IV, things have not changed |
|
Experimental (Alternative) Hypothesis
|
The IV has an effect on the DV
Experimental condition is different from control experiment |
|
P-Value Significance
|
p<.05
reject the null hypothesis and don't accept the the experimental hypothesis there is a significant difference "statistically significant" p>_.05 fail to reject the null not statistically significant "not statistically significant" |
|
Type 1 Error
|
false decision that an effect does exist
|
|
Type 2 Error
|
Failure to detect that the treatment had an effect
-due to sample size |
|
t-test
|
What we usually use when we are comparing 2 samples
use 1 IV to compare 2 levels/groups conditions |
|
Independent Sample t-test
|
Between-subject t-test
|
|
Dependent Sample t-test
|
Within-subject t-test
|
|
ANOVA
|
analysis of variance
-F typically used for 3 or more levels |
|
One-way ANOVA
|
1 IV is varied systematically to test difference between 2 or more groups/ level
|
|
Multifactor ANOVA
|
Between/ Within versions
More than 1 IV |
|
Omnibus F-test
|
If you have more than 2 groups and there is significance, you need to perform a pairwise comparison to determine where significance is (between what 2 groups)
|
|
Ceiling Effect
|
everyone gets everything correct
|
|
Floor Effect
|
No one gets anything correct
|
|
How to avoid ceiling and floor effects
|
questions too easy/hard
possible significance or not in IV but report none when should or report some when shouldn't |
|
Regression to the Mean
|
Regression Artifact
in any score you have true measurement and chance/error take 2 independent measures from same distribution. samples from mean on the first test will tend to be closer to the mean on the second test EG SI Jinxs EG lowest scoring school one year moves closer to mean next year and highest scoring mean the second year moves closer to mean next year EX head-start |
|
Reliability of Results
|
law of large #s
|
|
Experimental Reliability
|
likely hood significance is due to chance
you test by replication |
|
Direct Replication
|
replicate as closely as possible with as few changes as possible
|
|
Systematic Replication
|
certain factors changed that are thought to be irrelevant to see if overall outcome changes
EG passage length: long vs short EG type of writing style |
|
Conceptual Replication
|
Repeat same concept/phenomenon in a completely different way
same construct but change way of measurement or manipulation EG measure happiness on scale of 1-10 then measure happiness via facial muscles |
|
Replication and Sampling
|
be sure that the sample population can apply to overall population
EG Yale students don't apply to all students EG psych students dont apply to all students at regis u |
|
Test-Retest Reliability
|
exact same test given twice
|
|
Parallel Forms Reliability
|
2 math tests that cover the same material but with different questions.
|
|
Power and Effect Size
|
probability of rejecting the null hypothesis when it's actually false
Impacted by 2 main things sample size difference between the means EG huge sample size, small difference between the means but get significance.. is it really? |
|
Power and Magnitude of Effect Size
|
power and significance are greatly impacted by n
large n- with some diff between means, more likely to get significant difference small n- with same difference between group means, less likely to get significant difference |
|
Effect Size
|
calculate the strength of relationship
AKA "Magnitude of effect" Tells you if you have substantially important difference Must report both significance and effect size |