Sie sind auf Seite 1von 30

Quantitative analysis

Quantitative data analysis uses


statistics
 Descriptive statistics
Summarises data. Makes a lot of data easily readable.

 Inferential statistics
Tells you something about the population often from a
sample

Remember the hierarchy of methods? The bottom, non


experimental, can usually just use descriptive statistics
Time (minutes) taken by a group of
students to answer a question

20 25 24 33 13

26 8 19 31 11

16 21 17 11 34

14 15 21 18 17
Frequency table

Minutes Frequency

5 ≤ t < 10 1

10 ≤ t < 15 4

15 ≤ t < 20 6

20 ≤ t < 25 4

25 ≤ t < 30 2

30 ≤ t < 35 3
Frequency

Below normal 5

Normal 10

Above normal 5
Favourite pizza toppings of 20
students
Coding
Pepperoni
Cheese
P Su P P P
Sausage
Supreme
Sa Sa Ch Ch Su

Results
Sa Sa P P P
Pepperoni 10
Cheese 4
P P Ch Ch P
Sausage 4
Supreme 2
Favourite Pizza Toppings

20%

40% Pepperoni
Cheese
50%
Sausage
Supreme

40%
Marking assessments
 12 pieces of students' work

 We may be interested in the overall performance of


those students (Descriptive)

 We would also be interested in the distribution or


spread of the marks… Why? (Inferential)
Measurements of Central Tendency
(Averages)

 Mean
 Median
 Mode
Work out the Mean, Median and
Mode for these student ages
 15
 14
 16
 17
 17
 16
 20
 11
 11
 11
 13
 15
Work out the Mean, Median and
Mode for this group of student ages
 15
 14
 16
 17
 17
 16
 20
 11
 11
 11
 13
 50
Dispersion example
Dr X & Dr Y
Estimation

 100 street lights (population)


 Randomly sample 10 (sample)
 Find 2 are faulty

 Generalise that;
Confidence Limits and Intervals

 If from one population different samples will produce


different means (averages) how do we know which one
best fits the population average?
 Can have a range of plausible values
 Upper and lower ends are confidence limits
 The range of plausible values is the confidence interval
Confidence level

 The plausible range is set at a certain level: a common


one being 95%

 95% confidence level for the confidence interval


obtained
Hypothesis

 With hypothesis testing rather than looking for a range of


values for example an unknown population mean
(parameter)

 If the result of the sample taken from the population does


fall into the confidence interval then not enough evidence
against the null hypothesis

 A statement is made as to whether some hypothesis is


plausible at some given chosen significance level
Significance testing

 Unlike the confidence level approach

 Significance testing provides a quantifiable measure


of evidence against the null hypothesis

 Uses a p-value (probability value)


p-value: (significance testing)

 Significance probability of 5% p-value 0.05


 If the observed result is
 p>0.10 little evidence against H0
 0.10 ≥ p>0.05 weak evidence against H0
 0.05 ≥ p>0.01 moderate evidence against H0 p≤ 0.01 strong
evidence against H0

 Roughly anything less then 0.05 is statistically significant.


Statistical significance

 Is where the observed results were unlikely to have


occurred by chance alone

 p-value –is the chance of seeing the observed result


given that the null hypothesis is TRUE

 Or how likely the observed results are given the


hypothesis of no real relation (null hypothesis)
Power

 Power of a test is the probability of rejecting the null


hypothesis when it is false

 So need to have sufficient power in a study to do this.

 If we know the sample size. Standard deviation and


the fixed level test results then can calculate the
power. Ideally we want a power of about 0.8. and
usually at least a 5% (0.05)significance level
Plotting the data
Bar Chart

80
70
60
50
Series 1
40
Series 2
30
Series 3
20
10
0
Category 1 Category 2 Category 3 Category 4
Pie chart

Sales

1st Qtr
2nd Qtr
3rd Qtr
4th Qtr
Box plot
Boxplot of C1

20

18

16

14
C1

12

10

6
Scatter chart

Y-Values
6
5
Axis Title

4
3
2 Y-Values
1
0
0 1 2 3 4
Axis Title
Activity – quantitative data analysis

 Using the data collected within your questionnaires


1. Make sense of your groups data
2. Create charts/graphs to illustrate your data
3. Present your findings to the group

Das könnte Ihnen auch gefallen