12/02_MEETING Flashcards
TRUE OR FALSE: The typical recommendation for effect size is small-to-large.
TRUE.
TRUE OR FALSE: Power level would say “I detected light correctly” while Type I would say “I detected that this was incorrect”
TRUE.
TRUE OR FALSE: The higher value the better for your Type I error rate.
FALSE.
The lower.
TRUE OR FALSE: To detect sample size, you need the three variables of error rate, Type 1 error and power level.
TRUE.
TRUE OR FALSE: You cannot compute for prospective power analysis in R.
FALSE.
You totally can. There’s a syntax and you need the pwr library.
For linear regression, pwr.fw.test u=1 because simple linear regression sig. level = .01 (error rate) f2 = .02 (small to large) power = 0.80 (small to large)
Get the v (those are your participants)
For you to have this scenario, you need 585 (minimum) participants in your sample
TRUE OR FALSE: .02, .15, .35 is Cohen’s f2 for small-to-large, medium-to-large, and large, respectively
TRUE
Why does Sir Sagmit tell his students to plus minus 50 to 100 students?
Because you anticipate that you will lose participants (you might need to do outlier chuchu)
What does prospective power analysis serve?
To get sample size.
TRUE OR FALSE: When you conceive research that yields a linear regression, it is recommended that you use prospective power analysis to get sample size.
TRUE.
Bawal hula.
How do you conceive a simple linear regression?
Identify a predictor and an outcome from the variables in the file provided for.
How do you conceive a simple linear regression?
Identify a predictor and an outcome from the variables in the file provided for.
What is the generic question in simple linear regression?
Does the predictor predict the outcome? (Does dad indifference predict high standards for self?)
Which software does Sir Sagmit use to get linear regression?
JASP (not R, this time)
He doesn’t recommend bootlegging SPSS—best alternative is still R, but if you want the point and clic, JASP is one and JAMOVI is another one (fun fact: JASP and JAMOVI evolved from R)
What are the two types of regression JASP provides in its options?
Classical (frequentist)— Sir doesnt subscribe to this, but “unfortunately” we still need to learn this
Bayesian—“the future is Bayesian” but unfortunately no one is teaching Bayesian; Bayesian isn’t new, it’s from the 1960s but they didnt take it up; this has “complexity” and harder????
[SIR SHOWING RESULTS IN LINEAR REGRESSION THRU JASP?]
What does R^2 tell you
How much of the variable is being predicted by the equation of the line, or the statistical model?
R^2 = 0.004 (only 0.4% of your variable was being explained by your equation of the line)
What are the three parts of your linear regression results in JASP?
Model Summary - HIGHSTANDARDS
ANOVA
COEFFICIENTS
Under ANOVA, what does the p indicate?
“Imagine a world where anything that you say was simply due to chance. Nagkataon lang. So there’s nothing definite. So that means, it’s not statistically significant.
Now you observed 0.4% (this was predicted by the equation of the line), so how often will you find this in a world where predictions are random?
There’s a 46% probability that the predictions you actually observed that the predictions will be observed in a world where all predictions are random. In other words, your predictions might be simply random. (So it’s statistically non-significant)
Just think of it in the world where everything is chance.
When do we use y-intercepts?
When we try to look for predictions?
What does DADINDIFFERENCE of 0.053 (under understandardized in coefficients) mean or imply?
For every 1 unit increase in DADINDIFFERENCE, you predict a 0.053 increase in the other variable.
BUT YOU DONT WRITE IT THAT WAY IN YOUR REPORT. (refer to 9.09)
“When you see dat beta coefficient you know what’s going on”
Is the former statistically significant?
Check the p in the table.
It should be less than 0.05. (In this case, it’s not.)
[SIR SAID TO LOOK AT HIS FILE ABOUT EXPLAINING)
[Ysa, input what it looks like and apply the analysis is here.]
The beta coefficient is 0.3, for every one unit increase… there’s a 0.3 increase in the other one. (Review this pls)
[Johanna asked about covariance table and Sir is addressing it]
You know cor = cov/(SDxSD)
Where do you find the equation of the line?
Look at the coefficients table, and look at the variables
TRUE OR FALSE: It is possible for one variable to be statistically significant and another to be statistically insignificant.
TRUE.
*
How does COPEACCEPT predict?
After controlling for COPEACCEPTANCE, for every 1 unit increase in MOMINDIFFERENCE, you expect a 0.47 increase
“After controlling for” is important. There is a difference between statements that have this and do not have this
TRUE OR FALSE: The formula is always empirical divided by theoretical.
TRUE.
TRUE OR FALSE: The Blasians like dichotomous thinking.
FALSE. They hate it.
TRUE OR FALSE: Effect sizes became very popular due to estimation thinking.
TRUE.
TRUE OR FALSE: Estimation thinking is being debunked and now we have meta-analytic thinking.
TRUE.
“When you divide the world into two, you’re losing the fifty shades of gray.” (Estimation thinking)
TRUE.
TRUE OR FALSE: Our science (psychology) is probabilistic.
TRUE.
Dichotomous thinking is actually a violation to the nature of psychology, because nothing is absolute. (Shouldn’t we be dealing with estimates—what is the probability?)
What is the probability of an event under the given condition?
Conditional probability
Event is empirical, condition is theoretical, or what is ideal
TRUE OR FALSE: The p-value is a conditional probability.
TRUE.
The condition is the null hypothesis (nagkataon lang)
Probability of an event under a certain condition
TRUE OR FALSE: All our science is based on frequentist or statistical inference.
TRUE.
What is the dfiference between frequentists and bayesians when it comes to probability?
F - P (data/theory)
B - P(theory/data)
TRUE OR FALSE: We are putting the theory in the gold standard (or the hypothesis) when we look at the frequentist view.
TRUE.
Which view of probability leaves more for probability?
Bayesians view
r means relationship
yea
sana ol
Which view goes “IN THE LONG RRRRRRUN, you will observe in a world with no correlations, you’ll observe this 6 times”
Frequentist view
DID YOU KNOW THERE ARE 3 TYPES OF HYPOTHESIS TESTING
yessir yessir 3 bags full
What are the 3 types of hypothesis testing?
Test of statistical significance
Test of acceptance
Null hypothesis significance testing (NHST)
TRUE OR FALSE: Sir hopes you use NHST
False.
Hypothesis testing?
Estimate the strnegth of evidence against null hypothesis; which of the two competing hypothesis is more likely to reflect your data
Why does Sir not like NHST?
DECISIONS KASI and this is weird for psychology; it’s flawed! what would be the best alternative? some would say the baysians. but problem? no one’s teaching it. some say likelihood would be better (AIC-based procedures), however in the spirit of really understanding the nature of hypothesis testing (what i do in the undergrad) is to go back to the roots (aka test of statistical significance and test of acceptance)
What are the two types of hypotheses in the test of acceptance?
Main hypothesis and alternative hypothesis
