Karteikarten
Ch7-ch11

Karten in diesem Stapel (36)
- 1mode & how to determine unimodel/ multimodalmost frequent score
qualitative / nominal
determined by steam or leaf/histogram - 2median & how to determinemiddle iteam 50% above/under when in rank order
ordinal or ranked measure/when data is scewed
in boxplot - 3mean / deviation scoreaverage / when mean is subtracted from a score
- above score: pos. deviation - 4dispersionextent to which scores differn from another/ spread/scatter
- 5variance formulaincorporates every score and makes positive
score(X) - mean(M)² / N - 6standard deviation formula

- 7what does SD tell me?the smaller, the less variability from mean
the larger the greater the dispersal of scores - 8quartiles interquartile rangeq1 25% q2 50% q3 25%
interquartile: 50% --> less influenced by extermes, more stable measure of variablity - 9frequency polygonmost common way of representing frequency distriputions, e.g. easiest: line graph
- 10negativly skewed positivly skewedneg: left tail longer, right the majority of scores
pos: right tail longer - 11normal distributionperfectly symmetrical + unimodal
mode, media, mean the same
bell shape
symmetrical
tails dont reach x axis
6 SD
degree of peakesness: kurtosis
kurtosis + skeweness = 0 - 12Z-score formulaZ= ( score(X) - mean(M) ) / SD
- 13Z-ScoreA Z-score is number drawn from a (standard) normal distribution with mean zero (0) and standard deviation one (1).
numerical measurement of a value's relationship to the mean in a group of values/
If a Z-score is 0, it represents the score is identical to the mean score. - below mean
reveal the nr. of SD from the mean
- enables us to compare across different distributions - 14testing for normalityH0= normal distribution
- histogram
- skewness + kurtosis = 0
- Kolmogorov-Smirnov-Shapiro Wilk : need to be non significant, if the p-value is more than the chosen alpha level, then normal (H0 supported)
-Q-Q plot: how obtained scores differ - 15level of significane/ alpha level definition (0,05 / 0,01)the probability value that defines very unlikely outcomes
- the probability of a value between +/- 1,96Z from the mean is 95% (5% are above/below)
- the probability of a value between +/-2,58z from the mean is 99% (1% above/below) - 16skeweness / skeweness ratio- skewness=0=symmetrical distribution
- pos+high = asymmetrical, long tail to the right
• neg = asymmetrical, long tail to the left
skewness/SDerror of skewness
normal 0 bis +/- 2,58 (1%significance) - 17kurtosis/ kurtosis ratio- kurtosis of 0 = normal
- neg = flatter distribution
- pos: more peaked than normal
ratio like skewness ratio
divided by SDerror - 18sampling errorvariantion between parameter of pop and the sample statistic
- 19sampling biastendeny to favor the selection of certain sample elements over others
- 20Standard error of the mean (SEM) - deifinition - formulastandard deviation of the distribution of sample means around the population mean
SEm= SDs / Wurzel von Ns - 21parameterdescriptive measure of a characteristic of the entire pop of all oberservations of that characteristic (e.g. mean)
- 22Central limit theoremAs N increases, the sampling distribution of same sized sample means approaches a normal distribution with the sampling distribution mean equal to the population mean
(e.g. 9%% of means of all same-sized samples lie betwen +/- 2,58 SEm - 23normal: 68% lie between 95% lie between 99% lie between+/- 1
+/- 1,96
+/- 2,58 - 24sampliing error of population formulaWurzel aus: (pq/N)
- 25sampling framlist of target population in which we are interested
- 26probability sample vs non-probility sampleeach element has a chance to get seelected, SE+reliability can be calculated, generalizible
while non - probability sample is not rnado, cannot be calculated and not generalized - 27probability sampling methods1. random sampling (random process)
2. systematic sampling (select every Nth iteam)
3. stratisfied sampling (subgroups according to characteristics - take proporations from each strata/subgroup)
4. Cluster sampling (natural groups of pop) - 28non-probabiltiy sampling methods(shd only be used if propability sampling not possible)
1. opportunity sampling (conveniencly available)
2. judgement sampling (expert's identification)
3. quota sampling (certain # of each category/characteristic)
4. snowball sampling (contacts) - 29Hypothesis testing: 1) if p<0,05: 2) if p>0,051) less than 5%, z-score less than 1,96: not significant, reject H0=H1
2) more than 5%: significant, retain H0 - 30type 1 errorwhen you reject H0 though it is correct / false positive/ when you say things are significant differently but they are NOT
- 31type 2 errorH0 approved but H0 is false
you miss a significant relationship exisiting - 32what happens to type 1/2 error when lowering alpha from 0,5 to 0,1 levelType 1 error is lowered, as it is less likely that you reject H0
though type 2 error is higher, as you might miss a significant relationship
--> set alpha before testing - 33confidence level / confidence interval definitionprobability that the population parameter falls within confidence interval/
range within which we believe the true poplation estimate to lie
- more confident than a point estimate but less precise - 34effect sizemeasure of how large the effect of IV on DV
extent to which the proportions do not overlap
probability, significane or t-test dont give the full picture - 35measures of effect size + formula- standardized mean difference (d)
d=(Mexp-Mcontrol) / SD
- correlation coeffications: r and phi
- eta² (correlation ratio, strength for curvilinear relationship) - 36when is effect size d small/large.2 85% do overlap (small)
.5 67% do overlap (medi)
.8 53% do overlap (large)
1 no difference of groups
Informationen
- Beschreibung
- research Karteikarten am Ch7-ch11, erstellt von Nina H am 16/10/2016.
- Erstellt von
- Nina H
- Aufrufe
- 3
- Bewertung