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;
51 Cards in this Set
- Front
- Back
- 3rd side (hint)
|
What is Science |
Body of knowledge, field, or approach to studying variables producing verifiable results |
Knowledge, field, study = verifiable results |
|
|
Verifiable |
Good hypotheses, objective, replicable methods |
|
|
|
Objective/Replicable |
O: Not influenced by individual, not subjective R: repeatable |
|
|
|
Hypothesis |
Tentative explanation of a phenomenon Evidence that hypo is true = theory |
|
|
|
Good hypotheses |
Logical: - reasonable explanation Testable: - explanation for relationship btw variables that can be defined or measured Refutable: - can be proven false Positive: - explanations about prescence (not abscence) of relationship btw variables |
LTRP |
|
|
Pseudoscience |
Ideas based on nonscientific theory, faith, and belief |
All the wrong things used |
|
|
Scientific Steps |
1. Observe phenomenon 2. Develop hypothesis 3. Make prediction 4. Evaluate prediction 5. Address hypothesis |
ODMEA |
|
|
H1 |
Research hypothesis or alternative hypothesis - predicts relationships btw variables |
Something there |
|
|
H0 |
Null hypothesis: - Predicts no relationship btw variables |
Nothing there |
|
|
Type 1 Error and Alpha |
Rejection of Null when actually true Alpha: probability of type 1 error - significance level - p-val |
No relation = true |
|
|
Type 2 Error & Beta |
Acceptance of Null when actually FALSE Beta: probability of Type 2 error |
Acceptance |
|
|
Power |
Correct choosing H0 |
|
|
|
Simple random sampling |
Equal chance at being selected |
|
|
|
Systematic sampling |
Pick starting point go to nth number |
A - nth sampling |
|
|
Cluster sampling |
Select group measure all of them, hence cluster |
Starbucks |
|
|
Stratified random sampling |
Purposefully select particular demographic then randomly select individuals w/in each category |
Selection of race or other demographics |
|
|
Proportionate sampling |
Each group represented proportional to population. |
If 10% of pop has freckles, 10% of sample will have freckles |
|
|
Probability sampling |
1. Everyone has non-zero chance of being selected 2. |
|
|
|
Convenience Sampling |
Easily accessible people |
|
|
|
Quota Sampling |
- Represent certain groups - Decide amount of people from each group - Not Random |
|
|
|
Snowball sampling |
Participants recruit others like them |
Throwing a snowball |
|
|
Sampling Distribution & Standard error of mean |
Sample distribution of mean: Mu * X bar
Standard error mean: Sigma * x bar Distribution of statistic. Draw random sample, calculate sample mean, repeat until having many sample means |
|
|
|
Skewness |
Larger positive val = more positively skwed Lump left = positive
Larger neg val = more negatively skewed Lump right = negative |
|
|
|
Kurtosis |
- Middle peak of curve, if not normally distributed it is kurtosis Negative = flat and wide, platykurtic Positive = tall and narrow, leptokurtic |
|
|
|
Central Limit Theorem |
Mean of sampling dist of mean (mu*xbar) = mu (pop mean) Variance = sigma^2/N (sample size) Standard error of mean (mu*xbar) > pop SD (sigma), gets smaller as n increases Approaches normal dist as sample size (n) increases, and then leptokurtic |
|
|
|
Z-scores |
2%, 14%, 34% Way to estimate probability oucomes Z = x - mu/sigma -1 if z score negative |
|
|
|
Categorical (frequency) data |
Anything like male, female, any number that has 1, 2, 3 |
|
|
|
Face validity |
Does it look valid, unscientific |
|
|
|
Predictive validity |
Accurately predict behavior according to theory |
|
|
|
Concurrent validity |
When scores obtained form new measure correlate with scores from more well established measured |
|
|
|
Construct validity |
Grows overtime with studies; scores obtained from measurement behave exactly the same as variable itself Ex: temperature in predicting aggression |
54 |
|
|
Convergent validity |
Measurements converge on same construct. Correlates with other measures of similar constructs |
|
|
|
Divergent |
Measurement doesn't correlate with measures of dissimilar constructs |
|
|
|
Reliability |
Valid: Circle within reliability, E.g. if I measure height and intelligence, it's reliable bc height but not valid Reliability: Consistency of measurements - if I take survey and get one score, then take it in 6 months, if I get same score it's reliablr |
|
|
|
Correlation |
Measure of relationship btw two variables Pearson's r (+1, -1) Closeness to regression libe = correlation strength |
|
|
|
Strong correlation |
+- .75 to +1 |
|
|
|
Weak correlation |
0 to +-.25 |
|
|
|
Medium correlation |
+-.25 to +-.75 |
|
|
|
Guidelines Correlation |
Can have stat sig with large enough sample size, even w/ weak correlation
Can have strong correlation, but not stat sig bc small sample size
APA: r(N-2) = #, p...#
N = numbers of pairs of group scores
Spearmans r = correlation btw ranked or order variables |
|
|
|
Point Biserial Correlation |
rpb instead of r Correlation btw one continous variable and one dichotomous (variable that can take only two different vals) Phi(ø) Correlation btw two dichotomous (only two vals) variables |
|
|
|
Regression |
Prediction of one variable from knowledge of one or more variables
Equation: y = bx + a
Error of prediction (residuals) = difference btw y and y hat
Standard error of estimate: avg squared deviations: SD of points above and below regression line
Strong correlation = less error Weak = more Perfect = 0 |
|
|
|
T test |
Determines difference btw two means of two different populations |
|
|
|
One sample t test |
Sample vs population mean dif btw X bar and mu |
Ex: Difference btw SAT score in one state vs another |
|
|
Dependent samples t test |
Dif btw two related samples - Everyone gives 2 scores N * 2 = Number of scores Df = N - 1, N = #of pairs of scores |
Ex: dif in depression before therapy and after |
|
|
Independent samples t test |
Dif btw 2 unrelated samples N - 2, N = #of participants |
Ex: dif in depression btw those who did vs did not receive therapy |
|
|
Pearson's r |
Correlation btw two variables |
|
|
|
Spearmans r |
Correlation btw ordered or ranked variables |
Ex: behavior rank and symptom rank |
|
|
R^2 |
Variability in y related to x not caused |
|
|
|
Chi-Square |
Non-parametric determines if there is diff btw groups of Categorical data Also proportions, used single classified variable
Analyzes frequency/Categorical data/count
Df: G-1 groups - 1 |
# of something Ex: If we want to know who has more depression btw young and old participants |
|
|
Chi-square test of Independence |
Pearson's chi square or chi-square test of association determines of there is relationship btw two categorical variables
Participants classified on basis of two variables simultaneously
Critical Val = (R-C)(C-1) Any number of groups are possible 2x2 2x3 etc. For 2 variables If contingency table too small, chi-square not valid For <_ 9 cells, guidelines for min sample size is 5x # of cells Ex: 2x3 (6 cells) contingency table needs 30 participants |
|
|
|
Chi-square assumes |
independence of observations I.e. one person's scores don't affect another's scores One and only one score Larger N = more likely for stat sig |
|
