Home
Subjects
Create
Search
Log in
Sign up
Upgrade to remove ads
Only $2.99/month
Home
Math
Statistics
Hypothesis Testing
Module 1 : Pitfalls for New Researchers - Research Methods
STUDY
Flashcards
Learn
Write
Spell
Test
PLAY
Match
Gravity
Terms in this set (21)
Statistical power
The likelihood that the study will detect an effect, when there is an effect to be detected.
○ Statistical power is a probability, and can vary between 0 and 1 (like all probabilities!).
○ Cohen argues we should aim for 0.8 in the behavioural sciences, meaning that you have an 80% chance of detecting a real population effect.
Type I Error
False positives - when you see things that are not actually there (Example - a pregnancy test reveals a male is pregnant)
Type II error
False negatives - when you fail to find things that aren't there (Example - a pregnancy test fails to reveal that an 8 month pregnant woman is expecting)
A level
a level refers to significance level (also called alpha) - a higher a level (or significance level) increases the probability that you reject the null hypothesis
Be cautious - you don't want to reject a null hypothesis that is actually true - or make a type 1 error
Increase the a-level of your statistical test, to improve statistical power
Not recommended - In research this is set to 0.5, the research community does not recommend you increase it - you will be frowned upon. This is because although it will increase your "hit rate" it will also increase your likelihood of finding a type 1 error. It is far worse to conclude to falsely conclude there is an effect in the population (a) then to detect a real population effect
Why is not recommended to increase the a-level of your statistical test (in order to increase statistical power)?
In research this is set to 0.5, the research community does not recommend you increase it - you will be frowned upon. This is because although it will increase your "hit rate" it will also increase your likelihood of finding a type 1 error. It is far worse to conclude to falsely conclude there is an effect in the population (a) then to detect a real population effect
When should you use a one-tailed test instead of a two-tailed test?
If you can confidently predict the direction of your hypothesized population effect, (from previous research).
Is a one tailed test or a two tailed test more statistically powerful?
One tailed test
By using reliable measurement instruments, you can reduce _______ ________, (unreliable measurement instruments are a major source of ____ __________ ). If possible only use instruments that have a reliability of over 0.7.
error variance
Use a ______ subjects design to reduce the ammount of error variance in your data, and increase statistical power
within
Increasing the size of your treatment effect (particularly relevant to intervention studies) will increase statistical power, but it will also
limit generalisability
Unless we conduct an ___________, there is no way of knowing whether a lack of statistical power prevented us from detecting a real population effect
a priori power analysis
The more ____ _____ we include in our study, the less powerful our univariate statistical tests will be (meaning our descriptive statistics, such as mean, mode and median)
dependent variables
What are five extraneous factors that can confound an intervention effect?
History, maturation, testing, instrumentation, regression to the mean
Regression to the mean
a statistical phenomenon that occurs whenever subjects in the intervention group have been selected on the basis of their extreme pre-test scores. Subjects with extremely high pre-test scores will almost certainly score lower on the post-test.
Without a control group, the one group pre-test post-test design is unable to stay with certainty that the intervention effect had no extraneous factors influencing it - thus the design has no ______ _____
internal validity
A design has ____ ______ when a change in the dependent variable (or outcome variable) can be attributed to the independent variable (or the intervention)
internal validity
If subjects are not _____ _______to groups, it is unlikely they will be matched in terms of extraneous factors (pre-test scores, age, level of education etc) which can confound the intervention effect
randomly allocated
_____ _____ _______ undermines the internal validity of the design
group non-equivalence
______ ______ when the findings can be generalised to the wider population
external validity
When using a ___-tailed test, you are testing for the possibility of the relationship in one direction and completely disregarding the possibility of a relationship in the other direction.
one
YOU MIGHT ALSO LIKE...
Research 2 final
59 terms
Research Methods 1 (Psych-2101) - Chapter 4 Vocabu…
36 terms
Ch6: T tests-tesing the differences between two me…
33 terms
Exs chapter 14
74 terms
OTHER SETS BY THIS CREATOR
Statistical Power and Single Subject Designs
41 terms
Screening tests, qual and SEM
52 terms
Qual, Screening Tests & SEM
113 terms
All Terms
36 terms
OTHER QUIZLET SETS
PT Sem quiz #2
19 terms
interwar and ww2
15 terms
CH. 12 Quiz
20 terms
Volume 21: Trust and Theodore Roosevelt
50 terms