Study Skills

?
  • Created by: aarafa11
  • Created on: 07-01-20 17:16
What is qualitative data
method of observation to gather non-numerical data.
1 of 50
Types of qualitative data
nominal, ordinal, (binary)
2 of 50
What is nominal
Variables with NO order or ranking. E.g: race or gender
3 of 50
What is ordinal
Can be quantitative or qualitative. Variables with an order and ranking. E.g: ranking 1-5 or ranking outstanding - disappointing
4 of 50
How to collect qualitative
interviews, written statement and documents
5 of 50
What is quantitative
measures of values or counts and are expressed as numbers
6 of 50
Types of quantitative data
continuous data or discrete data
7 of 50
How to collect quantitative data
surveys, observation, experiments and interviews
8 of 50
What is contentious data
When the values in the set can take on ANY value. finite or infinite.
9 of 50
What is discrete data
values are distinct and separate
10 of 50
Difference between discrete and contentious
contentious can be measured and that discrete can be counted
11 of 50
Example of continuous data
can be any number, even decimal, between an interval. [0,70] - can have 2.5. Height of a child
12 of 50
Examples of discrete data
the numbers you can get on a dice - cannot get 2.5. number of language spoken. ordinal data are discrete
13 of 50
Advantages of using old data for a research
useful, effortless, save time, ethical consideration have been done
14 of 50
disadvantages of using old data for a research
no control on how data was collected, might not align with research aims, data needs scrutiny be fore use
15 of 50
Disadvanage of continuous data
observer error can be reduced but CAN'T be ELIMINATED. Limits of instruments to measure
16 of 50
When to use categorical nominal
when the data CANNOT be put into a meaningful order
17 of 50
When to use categorical ordinal
when the data CAN be put into a meaningful order
18 of 50
The significance of p value
lower than the p value = significant = data does NOT come from the same population
19 of 50
what is p value
if sample being compared is coming from the same population, which you don't want
20 of 50
What is p hacking
1)when you continue to collect data even when p<0.05. 2) data manipulation. Excessive stats
21 of 50
How does collection too much data cause p- hacking
it becomes biased as the results are no longer significant
22 of 50
how does too much stat cause p hacking
more result could happen by chance alone
23 of 50
The types of error when comparing data
type 1 = false positive, type 2 = false negative
24 of 50
What is a type 1 error when collecting data
WRONGLY accepting relationship. "guilty until proven innocent"
25 of 50
What is a type 2 error when collecting data
FAILING to accepting relationship. "innocent until proven guilty"
26 of 50
when to use chi-squared
for categorical data
27 of 50
what are the conditions for chi-squared
must be an actual data (raw, proportional, ratio), simple random sample, catergorical, contigency table of <5
28 of 50
what "actual" data is not included as a condition for chi-squared
percentage
29 of 50
Types of test for a continuous data
a parametric test or a non-parametric test
30 of 50
what makes it suitable to use a parametric test for a continuous data
for a normally distributed data, sample independent from 2 population, 2 population having similar variance, no outlier
31 of 50
what makes it suitable to use a non-parametric test for a continuous data
any continuous data based on the ranks of the data values
32 of 50
type of tests to know if data is normally distributed for it to be PARAMETRIC
t-test, ANOVA, regression
33 of 50
what is a t-test
2 samples collected from the SAME continuous data
34 of 50
which hypothesis should you disprove
null, p>0.05 to have a significant difference
35 of 50
what does the t-test do
compares the mean and dispersion of 2 samples so you can establish if the data came from the same population
36 of 50
what are the criteria for a t-test
continuous, 2 groups, independent, random samples, no outliers, normal distribution, similar variance
37 of 50
what happens when the criteria for a t-test are NOT met
use a log-transformation, use a less sensitive test
38 of 50
what is a log transformation
it address skewed data by decreasing the variability. this causes the data to be close to the normal distribution
39 of 50
what are the less sensitive test when the criteria for the t-test ISN'T met
Welch's test (parametric), Mann-Witney U test or Kruskal Wallis test (both ordinal)
40 of 50
when to use Mann-Witney U test (ordinal)
when t-test won't work. when there is 2 samples
41 of 50
when to use Kruskal Wallis test (ordinal)
when t-test won't work. more than 2 samples
42 of 50
what test can you do for ordinal data
Mann-Witney U, Kruskal Wallis, Likert scales
43 of 50
When to use likert scales (ordinal)
when there is a 5-7 point scales, when the average is the standard interval data
44 of 50
What is ANOVA
Analysis of variance
45 of 50
Criteria for ANOVA
Observation independent. No outliers. Each sample is normally distributed. Variance if roughly equal
46 of 50
How is each of ANOVA's sample normally distributed
large sample size (≥2), tested using graphically, tested using Shapiro-Wilk test, test the skewness, test using kurtosis
47 of 50
the p value for each of the test required to know the normal distribution for ANOVA
Shapiro-Wilks :) p=0.150 :? 0.128 :/ 0.32 . Skewness :) 0.533 :? 0.662 :/ 1.105
48 of 50
How would you compare to sets of continuous data on the same sample
regression (Pearson product moment correlation )
49 of 50
Criteria for regression test
continuous
50 of 50

Other cards in this set

Card 2

Front

Types of qualitative data

Back

nominal, ordinal, (binary)

Card 3

Front

What is nominal

Back

Preview of the front of card 3

Card 4

Front

What is ordinal

Back

Preview of the front of card 4

Card 5

Front

How to collect qualitative

Back

Preview of the front of card 5
View more cards

Comments

JanaC

Report

this is sooo useful !! thanks sooo much

Similar Other resources:

OB
See all Other resources »See all FI resources »