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;
109 Cards in this Set
- Front
- Back
|
What are the two different branches of stats? |
1. Despcriptive stats: summarise and describe numbers from a research study 2. inferential stats: sample > population draw conclusions and inferences go beyond the research study |
|
|
which 4 components determine which statistical method should be used? |
1. no. of variables (two vs. more than two) 2. level of measurement (i.e. Nominal, ordinal, interval and ratio) 3. type of comparisons (i.e.difference between conditions vs. relationships among variables) 4. research design (i.e. correlational design, experimental design,or quasi-experimental design) |
|
|
what is data? |
- variables organised for analysis |
|
|
what is a variable and what are the different types? |
- something that varies and that is measured
1. categorial -Arbitrary values represent categories 1= robbery, 2 = assault) 2. Discrete– can only have certain values within a range (217) 3.Continuous-can take on any value (i.e. 32.89038520) |
|
|
what is the difference between independent and dependant variables? |
- independant: controlled or manipulated by the researcher - the cause and effect (gender, age) - dependant: not controlled or manipulated by the researcher - the outcome (speeding fines) |
|
|
what are the different levels of measurement? |
- nominal data - ordinal data - interval data - ratio data |
|
|
what is nominal data? |
- categories associated with the variable are ‘different’ (gender, cities, make/model of a car) yes/no |
|
|
what is ordinal data? |
- Categories associate with the variable are ‘different’ AND the categories are ‘rankable’ (likert scales (agree, disagree,strongly disagree), school/military rank, letter grades. |
|
|
what is interval data? |
Categories associated with the variable are -‘different’ AND the - categories are ‘rankable’ - AND the intervals are ‘equidistant’ – equal (i.e. temperature) |
|
|
what is ratio data? |
- Categories associated with the variable are ‘different’ AND the categories are ‘rankable’ AND the intervals are ‘equidistant’ AND there Is a ‘true zero’ (i.e. victimisation rate) (homicides in united states) |
|
|
what are the different types of research designs? |
- experimental: -quasi experimental - correlational |
|
|
describe an experimental design |
- infer causation if researcher has manipulatedthe IV. True experiment involves random allocation to minimise confoundingvariables |
|
|
what is a quasi- experiment? |
- those that compare outcomes for an IV, but the IV has not been manipulated or randomly selected |
|
|
what is a correlational design? |
- they test relationships, cannot infer causation |
|
|
what is a popualtion? |
- an entire set of events or group of people- otherwise known as parameters |
|
|
what is a sample? |
- a selection of people from a populations - these are known as statistics |
|
|
what is a sampling error? |
- the difference between the population parameter and the sample statistic |
|
|
what are the ads and dis of a mean score? |
ads: - can be manipulated algebraically - stable across samples ( a better estimate) dis: - influenced by extreme scores - value may not exist - assumes interval properties |
|
|
ads and dis of median scores? |
ads: - unaffected by extreme scores and skewed distributions - doesn't require assumptions about interval properties of the scale dis: - does not enter readily into equations -not stable from the sample to sample |
|
|
ads and dis of mode scores? |
ads: - must occur - represents largest number of scores - highest probability of being chosen - applicable to categorial data dis: - changes with different data grouping - may not be representable |
|
|
when would you use mean to determine the typical case? |
- if variable is numeric and doesn't have extreme scores |
|
|
when would you use the median to determine the typical case? |
- if variable is numeric and has extreme scores |
|
|
when would you use the mode to determine a typical case? |
- if variable is categorical and modal category is observed |
|
|
what is a normal (curve) distribution? |
- SYMETRICAL - tails never touch x-axis : ends go into infinity - mean, median and mode are all the same: located at centre - bell shape |
|
|
what is a leptokurtic, platykurtic and mesokurtic shape? |
lepto: highly peaked platy: flattest meso: in between flat and highly peaked |
|
|
define dispersion? |
- numerical information of all the relative casesto the typical case/ how much variation in all the scores |
|
|
what is a standard deviation? |
- how scores differ from the average (mean) score in a sample -most common - is always a positive - square root of the variance |
|
|
what is variance? |
- it indicates how different scores are within a sample - the division between the average and the total of scores |
|
|
how do SD and variance differ and are similar? |
similar in that they compare every case to the average case different in that variance is unstandardised and SD is standardised |
|
|
how to calculate the range: |
- lowest minus highest score - |
|
|
what does SD indicate: - how to calculate |
- how scores differ from the average score in a sample - - square root of variance (always positive) - smaller value = closer to mean, larger value, further away from mean |
|
|
what is split file useful for: |
- to compare groups |
|
|
how is probability calculated: |
- dividing number of occurrences of a particular outcome by number of possible outcomes |
|
|
what is conditional probability? |
the likelihood a specific event will occur, given the occurrence of another |
|
|
what is joint probability: |
likelihood two specific events or more will occur |
|
|
importance of SND: |
- compare scores from different samples |
|
|
what are Z-scores: calculation? |
- tell us how many SD above or below the mean - identify location = observed scores - mean divided by SD = X-M/ SD |
|
|
what is sampling error? |
- sample doesn't accurately represent population |
|
|
what is central limit theorem? |
- greater number of samples the closer our estimate to population estimate |
|
|
what is sampling distribution? |
- spread of statistics (means) from each sample drawn from population |
|
|
what are confidence intervals? |
- quantifies how likely our sample statistic approximates the population parameter |
|
|
what is recode? what is compute? |
- a change in the variables or the way they have been measured - makes calculations from variables |
|
|
how to calculate the standard error? how to calculate upper limit and lower limit? |
SE = SD/square root of N upper = M + (1.96 x SE) lowwer = M - (1.96 x SE) |
|
|
what is SE? |
The Standard deviation of the sampling distribution |
|
|
what is a hypothesis? |
an expectation of the outcome |
|
|
what is a null hypothesis? when would you reject and not reject the null? |
- no relationship or difference - can't prove something true but can prove it to be false reject= p value < .05 fail to reject = p value > .05 |
|
|
what is type 1 and type 2 error |
1. rejecting the hypothesis when it is actually true 2. not rejecting the hypothesis when it is actually false |
|
|
What is the null hypothesis for the following researchhypothsis: On average, sex offenders will recidivate if, upon their release,they live within 300m of a place where children gather. |
Answer: There is no relationship between places where children gather and sex offender recidivism
|
|
|
what is a one-tailed test? |
- directional - null hypothesis = no relationship between two variables - research hypothesis = positive relationship between two groups (group 1 score higher than 2) - looking at one end of distribution (highest 5%)` |
|
|
what is a two tailed test? |
- non-directional or bi-directional - research hypothesis = there is a relationship between two variables or a difference - looks at both ends of distribution (lowest and highest 2.5%) |
|
|
what does power refer to? |
- ability to correctly reject null hypothesis when it is false (1 - Beta) |
|
|
what is a Brivariate correlation? |
relationship between 2 variables |
|
|
what are the steps in examining relationships? |
1. hypothesis 2. scatterplot 3. correlation coeficient 4. interpret results |
|
|
What is the relatonship between age of first arrest and total number ofarrests? hypothesis/ null and research? |
- there is a relationship - there is no relationship - there is a negative/ positive relationship between age of first rest and total number of arrests |
|
|
what is a correlation coefficient? examples? |
measure of relationship between variables 1. pearsons r - parametric / 2. spearman's rho- non parametric |
|
|
how to calculate Pearsons r effect size/ variance = |
squaring r r x r .4 x .4 = .16 |
|
|
how to calculate pearson's r |
r = X and Y vary together / X and Y vary separately |
|
|
what is a partial correlation? |
- tell correlation between 2 variables while disregarding another - first order correlation |
|
|
what are the assumptions of pearsons r |
- must be continous - related pairs - normally distributed - using histograms - linea and homo (same) scedastic (variances) - if violated use spearmint rho |
|
|
what is spearmint rho? |
- non-parametric - measure of linearity in ranks |
|
|
what is Cronbach's alpha? |
- indicates internal consistency - correlation between the item and the contract - if above.7 considered reliable - bigger the better ranges from .77 to .88 |
|
|
what is - point biserial correlation? - phi coefficient? |
- one dichotomous variable - both variables dichotomous |
|
|
numbers of weak/moderate and strong relationships = perfect = |
weak = .10 - .39 moderate - .40 - .69 strong = .70 - .99 perfect = -1 or 1 |
|
|
what is the CAVED scale? |
the stolen item was: Concealable Removable Available Valuable Enjoyable Disposable |
|
|
what do correlation and regression measure? |
relationships |
|
|
mean testing examines what among groups? |
- differences - |
|
|
t tests examine differences between how many groups? ANOVas examine differences among how many groups? |
- 2 means - more than 2 means |
|
|
Is the mean of one group different from the mean of another group? what test to use? |
- independent t test / testing null hypothesis |
|
|
Is the mean of a group taken at time 1 different from the mean of thesame group collected at time 2? |
- related samples / paired samples t-test |
|
|
Is the mean of a group different from a known mean? |
- one sample t-test |
|
|
how to calculate the t statistic and what is it = |
between group variance / within group variance - a ratio |
|
|
how to calculate degrees of freedom = |
DF = (N group 1 - 1) + (N group 2 - 1) |
|
|
how to report Cohens D = |
d = M1 - M2 / Mean SD Mean SD = SD1 + SD2/ 2 |
|
|
Cohen reporting sizes -small -medium -large |
.2 .5 .8 |
|
|
when would you use the top row of levlens t test? |
when it is not significant |
|
|
assumptions of t tests |
- continuous DV and categorical IV - normal distribution - independence of groups - homogeneity of variance |
|
|
what is the calculation of a linear equation? |
outcome = slope *(x = independent variable value) + intercept (constant) |
|
|
what is residual variance? |
square of the standard error of the estimate |
|
|
What would be the predicHon of the 2013 result, if they receiveda score of 12 on the 2010 test? slope = .94 intercept/constant = 7.90 |
answer = 19.18 |
|
|
power = |
1- Beta beta = probability of type 2 error |
|
|
when is Chi-square used? assumptions? |
- when you have categorical or frequency data - no more than 25% of cells and has expected frequency less than 5 |
|
|
chi-square equation? (X2) |
X2= E (O-E) 2 / E O=observed E = Expected |
|
|
calculation of degrees of freedom = |
k - 1 k= number of categories |
|
|
can Chi square have negative values? |
NO |
|
|
what is goodness of fit test? |
- interested in frequencies of one variance - only categorical data |
|
|
when is chi square test of independence used? |
- test if there is a relationship between 2 variables |
|
|
when is a one way ANOVA used? |
- categorical and continuous data - tests differences between more than 2 means |
|
|
what is an alternate or experimental hypothesis? |
- at least one population means is different from the others |
|
|
what is the f ratio? |
between / within group variance = calculate meaner each 3 groups - calculate grand mean (3 means divided by 3) F = treatment effect + difference due to chance / difference due to chance |
|
|
how to calculate the between group variance? |
group/error |
|
|
what is a post hoc analysis |
- makes comparisons between two or more group means after an ANOVA therefore use LSD, HSD or Bonferroni procedure |
|
|
LSD? when to use and what does it do |
- use only if the ANOVA f is significant - reduces type 1 error - works with groups smaller than 5 |
|
|
what is bonferroni |
- family wise error rate / comparisons - reject null when p< .017 |
|
|
One-way between (independent measures) groups ANOVA: |
Compare three or more independent groups (IV) on acontinuous DV. (extension of the independent-measures t-test). |
|
|
One-way within (repeated measures) groups ANOVA: |
Compare the same sample on a continuous DV at three or more points in time (IV) (extension of the repeated measures/paired samples t-test). |
|
|
Two-way (factorial) between groups ANOVA: |
Two IVs (one with 3 or more levels). Examine the effect of eachof the IVs on a continuous DV (main effects). Examineinteraction between the two IVs (interaction effect). |
|
|
how to calculate effect size? |
partial Eta squared = n2 |
|
|
ADS and DIS about non-parametric tests: |
- Fewer assumptions about the distributions (e.g., normality) • Simple • Unaffected by outliers (use ranks) Disadvantages • Lower power (need larger sample size for same power ofparametric test) • Rank randomised tests – deal with ranked data |
|
|
parametric and non-parametric test for two independent samples? |
- para: independent samples t-test - non: mann-whitney U test |
|
|
parametric and non-parametric test for two related samples: |
para: related samples t-test non: wilcoxin's t-test |
|
|
parametric and non-parametric test for three or more independent groups: |
para: one-way ANOVA non: kruskal-wallis one-way ANOVA |
|
|
parametric and non-parametric test for three or more repeated samples: |
para: repeated measures ANOVA non: friedman's rank test for k correlated samples |
|
|
what is a mann whitney Test ? |
- scores for groups are ranked - examines number of times one condition is ranked higher than the other - sig difference means distributions had difference central tendencies |
|
|
what is wilcoxon's matches pairs test? |
- changes scores to ranks - Tests null hypothesis that distribution of difference scores issymmetric about zero - test presented as T |
|
|
what is a kruskall wallis ANOVA? |
- tests hypothesis that all sample were drawn from identical populations - ranks all scores then sums the ranks for each group |
|
|
what is freedman rank test? |
- applied to ranks instead of raw scores - tests the null hypothesis that the scores for each treatment were drawn from identical populations |
|
|
assumptions of a non-parametric test? |
- can handle outliers - no normal distribution |
|
|
central tendency value used for: categorical categorical order or ranked continuous (skewed) continuous (normal distributuion) |
- mode -median -median -mean |
|
|
when looking at correlations use what test for - continuous data - ranked data |
- pearson - spearman |
