Lies, damned lies & statistics Communication Research week 10.
-
Upload
quentin-cooper -
Category
Documents
-
view
223 -
download
1
Transcript of Lies, damned lies & statistics Communication Research week 10.
Communication Research Spring 2005 2
Basics of descriptive statistics
Statisticians use mathematical methods to analyse, summarise and interpret data that have been collected
Descriptive statistics describe the basic features of the study and allows the researcher to get a feel for the data
The choice of statistical method of analysis depends on the data that have to be analysed
Communication Research Spring 2005 3
Descriptive vs inferential statistics
Descriptive statistics refer to methods used to obtain, from raw data, information that characterises or summarises the whole set of data
Inferential statistics allow us to generalise from the data collected to the general population they were taken from
Communication Research Spring 2005 4
Descriptive Statistics
Qualitative Quantitative
FrequencyRelative frequency
Percentage
Measures of Central TendencyMeasures of spreadFive number system
TablesPie ChartsBar Graphs
Tables HistogramsBox plotsBar chartsLine charts
Communication Research Spring 2005 5
Different statistical measures
Raw data is unorganised but can be tabulated to make it easier to understand and to interpret
It is usually presented as a frequency table or graph
A frequency chart will allow a researcher to see trends or groupings of data and how they are distributed
Communication Research Spring 2005 6
Data: The raw material of statistics. Numbers that result from measurements or counting.
Statistics: The field of study concerned with the collection, organization, summarization and analysis of data and the drawing of inferences about a body of data when only a part of the data is observed.
Sources of Data
Routinely kept records
Surveys
Experiments
External Sources
Some Basic Concepts Related to Statistics
Communication Research Spring 2005 7
Random Variable: A variable whose values arise as a result of chance factors and cannot be exactly predicted in advance.
Population: A population of entities is defined as the largest collection of entities for which we have an interest at a particular time.
Sample: A part of a population.
Some Basic Concepts Related to Statistics
Communication Research Spring 2005 8
Statistical Inference
The procedure by which we reach a conclusion about a population on the basis of the information contained in a sample that has been drawn from that population.
Simple random sample
If a sample of size n is drawn from a population of size N in such a way that every possible sample of size n has the same chance of being selected, the sample is called a simple random sample. 2 out of 4=2C4=6
The Simple Random Sample
Communication Research Spring 2005 9
Characteristics of each distribution
Location – where on the axis is the distribution positioned?
Dispersion – how broad is the distribution? Shape – what is the form (appearance, pattern) of
the distribution? The type of data you have to analyse will determine
the statistical measure chosen Statistics describing the location of the distribution
are called measures of central tendency
Communication Research Spring 2005 10
Measures of central tendency – the mean The mean is the sum of all observed data
values divided by the sample size (the arithmetic average)
Describing data that are interval or ratio in nature (eg speed of response, age in years) calls for the mean
One of the main disadvantages is that it is most profoundly affected by extreme scores
Communication Research Spring 2005 11
Calculating a Mean ScoreScores:79818286868891939597
total = 878Divide by n = 10 scoresMean = 87.8
Communication Research Spring 2005 12
Measures of central tendency – the median
The median is the score or the point of distribution above which one half of the scores lie eg in a simple set of scores such as 1, 3, & 5 the median is 3
The median is best suited to data that are ordinal or ranked ( eg birth order, rank in class)
To compute the median Order the scores from lowest to highest Count the number of scores Select the middle score
When the number of scores is even, find the mean of the two middle scores
eg 31 33 35 38 40 41 42 43 44 46 47 48 49 50 N = 14 (no of scores); Median = (42 + 43) ÷ 2 = 42.5
Communication Research Spring 2005 13
Two distributions of scoresDistribution 1 Distribution 2
24 24 25 25 26 26
Mean = 25 Range = 3
16 19 22 25 28 30 35
Mean = 25 Range = 20
Communication Research Spring 2005 14
Measures of central tendency – the mode
The mode is the most frequently observed value in the frequency distribution ie it is the score that occurs most frequently
The mode is best used for nominal data and for data that are qualitative in nature such as gender, eye colour, ethnicity, school or group membership
In the following list of numbers: 58 27 24 41 27 26 41 53 24 29 41 53 47 28 56 The mode is 41 because it occurs 3 times
A common mistake is to identify the mode as how frequently the value occurs (3) not the value itself (41)
Communication Research Spring 2005 15
Which measure when?
Which measure of central tendency?
Measure Level of measurement Examples
ModeNominal or categorical – ie
qualitativeGender, hair or eye colour, group membership, ethnicity, school etc
Median Ordinal or ranked Rank in class, birth order
Mean Interval and ratio Speed of response, age in years
Communication Research Spring 2005 16
Three Measures of Variability
Range: the difference between the highest and lowest scores in a distribution of scores.
Variance: a measure of dispersion indicating the degree to which scores cluster around the mean score.
Standard deviation: index of the amount of variation in a distribution of scores.
Communication Research Spring 2005 17
Standard deviation
SD is a measure of the variability indicating the degree to which all observed values deviate from the mean
SD can only be used for interval and ratio data It is the most frequently used statistic as a
measure of dispersion or variability The larger the SD, the more variable the set of
scores is
Communication Research Spring 2005 18
COMPUTING DEVIATION SCORES
Raw Mean DEV. SQUAREDscore score deviation score 4 - 10 = -6 36 8 - 10 = -2 4 9 - 10 = -1 110 - 10 = 0 010 - 10 = 0 010 - 10 = 0 012 - 10 = 2 413 - 10 = 3 914 - 10 = 4 1690/9 = 10.00 = MEAN
70/9 = 7.77 = VarianceSTANDARD DEVIATION: (Square Root of Variance) = 2.79
Communication Research Spring 2005 19
Types of Variables
Variable Element that is identified in the hypothesis
or research question Property or characteristic of people or
things that varies in quality or magnitude Must have two or more levels Must be identified as independent or
dependent
Communication Research Spring 2005 20
Independent Variables
Manipulation or variation of this variable is the cause of change in other variables
Technically, independent variable is the term reserved for experimental studies Also called antecedent variable,
experimental variable, treatment variable, causal variable, predictor variable
Communication Research Spring 2005 21
Dependent Variables
The variable of primary interest Research question/hypothesis describes,
explains, or predicts changes in it The variable that is influenced or changed by the
independent variable In non-experimental research, also called criterion
variable, outcome variable
Communication Research Spring 2005 22
Relationship Between Independent and Dependent Variables
Cannot specify independent variables without specifying dependent variables
Number of independent and dependent variables depends on the nature and complexity of the study
The number and type of variables dictates which statistical test will be used
Communication Research Spring 2005 23
Issues of Reliability and Validity
Reliability = consistency in procedures and in reactions of participants
Validity = truth - Does it measure what it intended to measure?
When reliability and validity are achieved, data are free from systematic errors
Communication Research Spring 2005 24
Threats to Reliability and Validity If measuring device cannot make fine
distinctions If measuring device cannot capture
people/things that differ When attempting to measure something
irrelevant or unknown to respondent Can measuring device really capture the
phenomenon?
Communication Research Spring 2005 25
Other Sources of Variation Variation must represent true differences Other sources of variation
Factors not measured Personal factors Differences in situational factors Differences in research administration Number of items measured Unclear measuring device Mechanical or procedural issues Statistical processing of data
Communication Research Spring 2005 26
Types of variables
Data
Variables
Quantitative(numeric)
Qualitative(categorical)
Discrete Continuous Nominal Ordinal
Communication Research Spring 2005 27
Definitions
Variable: a characteristic that changes or varies over time and/or different subjects under consideration.
Changing over time Blood pressure, height, weight
Changing across a population gender, race/ethnicity
Communication Research Spring 2005 28
Definitions (con’t) Quantitative variables (numeric): measure a
numerical quantity of amount on each experimental unit
Qualitative variables (categorical): measure a non numeric quality or characteristic on each experimental unity by classifying each subject into a category
Communication Research Spring 2005 29
Categorical variables
Nominal: unordered categories Race/ethnicity Gender
Ordinal: ordered categories likert scales( disagree, neutral, agree ) Income categories
Communication Research Spring 2005 30
Univariate statistics (numerical variables)
Summary measures Measures of location Measures of spread
Overall pattern (distribution) Unimodal (one major peak) vs. bimodal) (2 peaks) Symmetric vs. skewed Outliers-an individual value that falls outside the
overall pattern
Communication Research Spring 2005 31
Skewness
The skewness of a distribution is measured by comparing the relative positions of the mean, median and mode. Distribution is symmetrical
Mean = Median = Mode
Distribution skewed right Median lies between mode and mean,
and mode is less than mean
Distribution skewed left Median lies between mode and mean,
and mode is greater than mean
Communication Research Spring 2005 32
Relative positions of the mean and median for (a) right-skewed, (b) symmetric, and(c) left-skewed distributions
Note: The mean assumes that the data is normally distributed. If this is not the case it is better to report the median as the measure of location.
Communication Research Spring 2005 33
Summary statisticsMeasures of spread (scale)
Variance: The average of the squared deviations of each sample value from the sample mean, except that instead of dividing the sum of the squared deviations by the sample size N, the sum is divided by N-1.
Standard deviation: The square root of the sample variance
Range: the difference between the maximum and minimum values in the sample.
( )∑=
−−
=n
ii xx
ns
1
2
1
1
( )∑=
−−
=n
ii xx
ns
1
22
1
1
Communication Research Spring 2005 35
Graphical display of numerical variables (histogram)
Class IntervalFrequency
20-under 30 6
30-under 40 18
40-under 50 11
50-under 60 11
60-under 70 3
70-under 80 10
10
20
0 10 20 30 40 50 60 70 80
Years
Frequency
Communication Research Spring 2005 36
86
76
23
77
81
79
68
77
92
59
68
75
83
49
91
47
72
82
74
70
56
60
88
75
97
39
78
94
55
67
83
89
67
91
81
Raw Data Stem
2
3
4
5
6
7
8
9
Leaf
3
9
7 9
5 6 9
0 7 7 8 8
0 2 4 5 5 6 7 7 8 9
1 1 2 3 3 6 8 9
1 1 2 4 7
Graphical display of numerical variables (stem and leaf plot)
Communication Research Spring 2005 37
NegativelySkewed
PositivelySkewed
Symmetric(Not Skewed)
S < 0 S = 0 S > 0
Graphical display of numerical variables (box plot)
Communication Research Spring 2005 38
Summary measures Count=frequency Percent=frequency/total sample
The distribution of a categorical variable lists the categories and gives either a count or a percent of individuals who fall in each category
Univariate statistics(categorical variables)
Communication Research Spring 2005 39
Displaying categorical variables
Rank Cause of Death
Frequency (%)
1 Heart Disease
710,760 (43%)
2 Cancer 553,091 (33%)
3 Stroke 167,661 (11%)
4 CLRD 122,009
( 7%)
5 Accidents 97,900
( 6%)
Total All five causes
1,651,4210
20
40
60
heart cancerstrokeCLRDaccident
heart cancer stroke CLRD accident
Communication Research Spring 2005 40
Common Applications T-Tests – the independent t-test is used to test for a
difference between two independent groups (like males and females) on the means of a continuous variablecontinuous variable.
one sample – compare a group to a known value For example, comparing the IQ of convicted felons to the known average of
100)
paired samples – compare one group at two points in time For example, comparing pretest and posttest scores
independent samples – compare two groups to each other
Communication Research Spring 2005 41
Common Applications
The Pearson's correlation is used to find a correlation between at least two continuous variables. The value for a Pearson's can fall between 0.00 (no correlation) and 1.00 (perfect correlation).
Other factors such as group size will determine if the correlation is significant. Generally, correlations above 0.80 are considered pretty high
Communication Research Spring 2005 42
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Days AbsentNon-significant t-test
60
50
40
30
20
10
0
Common Applications
Num
ber
of
people
Male Female