# geography f764 - skills

?
• Created by: charlie
• Created on: 18-02-15 12:35

## Methods of Sampling

SAMPLING FRAME

SPATIAL - where that location will be (river study)

NON SPATIAL - who to ask (questionnaire)

METHODS

• Random
• Stratified
• Systematic
• Pragmatic
1 of 39

## 1. Random

DESCRIPTION

• random number tables to select sample point with every idividual variable having equal chance of being picked

• stat. tests
• no bias

• may not be representative of total statistical population (miss points)
• takes a long time
2 of 39

## 2. Stratified

DESCRIPTION

• underlying subdivisions in total stat. population are taken into account
• propotinally sampled

• no points missed or overepresented

• possible bias - invalidate inferences made from statistical tests
• possibly no stat test
• need to get info of underlying patterns
• groups under / over represented in proportional sampling
3 of 39

## 3. Systematic

DESCRIPTION

• points are selected at regular intervals
• spatial (where) + non-spatial (who)

• easy and quick
• allows even coverage to test hypothesis

• bias
• interval may coincide with data / location
• possibly no stat
• missed vatiations + regularities not typical
4 of 39

## 4. Pragmatic

DESCRIPTION

• and where changes can be observed

• realistic
• risk assessment based

• bias
• possibly no stat test
5 of 39

## Units of Sampling

Point

• number of individual points

Grid

• Quadrat used at indivial points

Line

• transect - measurements along a line

Belt transect

• wider transect
6 of 39

## Accuracy

DESCRIPTION

• level at which data is exact and free from error

ERRORS

measurement error -

• repeating increases room for error

operator error -

• idividual / equipment / climatic
7 of 39

## Reliability

DESCRIPTION

• extent to which sample data reflect the greater whole statistical population

ERRORS

sampling-

• inappropriate sampling frame
• small sample size
8 of 39

## Risk assessment

IDENTIFY RISK

EVALUATE SEVERITY (1-5)

CONTROL MEASURES

RE-EVALUATE SEVERITY (is it acceptable with control?)

9 of 39

## Pilot study

DESCRIPTION

• a small scale preliminary study used to evaluate successes / weaknesses and improvements to study design prior to performance of a full scale research project

• identify improvements
• if given time is acheivable
• if location is appropriate
• effectivenesss of sample method - representative, accurate, reliable
• risk assessment
• equipment effectiveness

• takes time
10 of 39

## Techniques of Presenting

TRENDS to be represented

SPATIAL - patterns between areas

TEMPORAL - patterns overtime

TECHNIQUES

• maps (QUAL/QUANT)
• graphs (QUANT)
• diagrams (QUAL/QUANT)
• annotated field sketches / photos (QUAL)
11 of 39

## Maps (spatial)

1. Chloropleth

2. Dot

3. Proportional Symbols

4. Isoline / Isopleth

5. Flow line

12 of 39

## 1. Chloropleth Maps

DESCRIPTION

• ratio values expressed as densities or percentages
• use of area

METHOD

• standardized values (%)
• divide into classes
• class sizes (fixed / mathematical calc / dispersion in data)
• title, key, scale, direction

• easy + useful for area census data / good visual patterns

• conceal variations in areas / abrupt changes at bounday / large units dominate / broad boundaries
13 of 39

## 2. Dot Maps

DESCRIPTION

• ratio data
• areas

METHODS

• find base map with boundaries
• decide on dot value (represent low values) + size (just touch in high density areas)
• place on maps
• key, title, scale, direction

• only one which provides info within wards
• visual + spatial patterns + not interupption at area boundaries
• statistical ananlysis

• high densities dots merge
14 of 39

## 3. Proportional Symbols

DESCRIPTION

• ratio data in specific location / or absolute

METHODS

• suitable scale base map
• choose symbol + scale
• calculate size / area of symbol
• plot in central of area
• add key, title, scale and direction

• good visual + spatial
• data can be recovered and can show whole range of values

• hard scale to decide upon / larger symbols misread / difficult to place on small maps
15 of 39

## 4. Isoline / Isopleth

DESCRIPTION

• ratio data
• uses points

METHOD

• suitable scale base map
• decide on number and value of isolines
• draw by interpolation
• number isolines
• add key, title, scale, direction

• clear spatial patterns / shows gradual changes / guess of values between data

• need a lot of points / assuming constant change / only used if variable changes with space
16 of 39

## 5. flow line

DESCRIPTION

• ratio data
• uses lines to show flow paths (routed/ actual pathways or non-routed/ straightlines)

METHOD

• suitable scale where flow lines can be recorded
• number of classes + scale of line (width)
• key, title, scale, direction

• visually shows patterns of movements + quick/easy
• can show different variables of direction

• very generalised + hard to stat. test
• carefully chose width scale
17 of 39

## Graphs (non-spatial)

Tables

Charts

• Bar Charts
• Divided/Stacked Bar Charts
• Pie Charts
• Rose/Star Charts

Graphs

• Line Graphs
• Scattergraph
• Triangular Graphs
• Lorenz Curves
• Kite Diagrams
18 of 39

## Tables

DESCRIPTION

• nominal data (presence or abscence) in form of parish names
• interval/numerical data in form of % of arable
• freq. distribution - values grouped into numerical values
• used to construct histogram

• backbone for then contructing futher presentation techniques
• clearly group data into individual classes
• statistical tests

• no spatial pattern / visual represenatation
19 of 39

## Bar Charts

DESCRIPTION

• interval and discrete data used
• quantities to be compared
• width constant and legth varies

• visual with temporal/spatial
• stat. tested + comparable

• non-continuous data cant be used
20 of 39

## Stacked/Divided Bar Chart

DESCRIPTION

• quatity that can be divided into component parts
• absolute values or porportions

• easy to draw / interpret
• visual pattern identified
• stat. test

• too many divisions will confuse
21 of 39

## Pie Charts

DESCRIPTION

• uses % data
• circular chart divided into segments of a whole sample

• visual and easily comparable
• precise % allowing stat. tests

• not always exact %
• too many divisions confuse + no temporal change can be shown
• when comparing, order of division has to be standardised
22 of 39

## Rose/Star Charts

DESCRIPTION

• used to show direction
• length/width of bars show frequency

• clearly show direction and spatial changes
• visual pattern represented
• can recover stat. if accurrate + to scale

• width/length scale hard to produce
• may not be able to state exact direction
23 of 39

## Line Graph

DESCRIPTION

• variations in absolute or %
• X axis independent and Y dependent
• X influences Y

• trends, patterns and anomalies highlighted

• only continuous
24 of 39

## Scatter Graph

DESCRIPTION

• show an overall trend, not just a line of points
• simplest way of looking at relationship betweek 2 variables
• first stage in stat analysis
• closer points in straight line = stronger relationship

• correlation suggested (line of best fit)
• spearman's rank
• anomalies

• few points can skew data
• too many confuse
• only between 2 variables
25 of 39

## Triangular Graph

DESCRIPTION

• visually represents 3 variables
• must add up to 100% on each side of equilateral triangle
• spatial and temporal

• 3 variables can be visually compared

• only %
• can be difficult to read
• can be difficult to construct
26 of 39

## Lorenz Curve

DESCRIPTION

• line graphs that show inequalities in distribution of a phenomena
• cumulative frequency curves
• striaght line shows even distribution
• Lorenz curve shows actual distribution and how much it deviated from even
• more concave = greater inequality
• 'gini' coeffiecient calculated to show degree of inequality

• viual patterns identified
• plot more than 1 distribution
• gini coefficient calculated

• time conuming to calculate cumulative freq.
• no exact quantative measure of degree of inequality
27 of 39

## Kite Diagram

DESCRIPTION

• shows how a phenomena's occurence changes over distance

• visually effective
• recove stat. tests
• relative phenomena compared

• only show % change, not absolute values (Bradshaw - no values)
28 of 39

## Sketch Map

DESCRIPTION

• box to keep in area sketching
• add only essential infromation (locational reference points)

• simple and easy to interpret

• to be effective needs scale, key, labels, direction
29 of 39

## comparing maps and photos

MAPS

• plan view indicating scale + direction
• shows what exists in area (characteristics / economic activity ...)
• flexible scales
• GIS - shows layering
• sometimes dated - doesnt show present
• shows scales + spatial data

PHOTOS

• aerial, oblique, ground level, satellite views
• capture at instant in time
• shows what is there at that time
• shows variety of data - colours, veg...
30 of 39

## GIS (geographical information system)

AEGIS 3 software used in connect with pilot study

• data stored, organised, combined with other data and then displayed

• cope with large amount of data
• cope with large variety
• easily change scale
• dynamic - cope with change
• allows spatial data to be investigated to show patterns / trends
• information tied to places

USES

1. collecting / recording data - (primary / weather  + secondary / census data)

2. presenting data - (area / line / point / symbols / colours / overlayed / compared)

3. analysing / interpreting - (investigate + answer questions)

31 of 39

## Types of data

PRIMARY

• unprocessed
• not analysed or interpreted
• geogrpaher has direct control

SECONDARY

• dervived from published documents that has been anlysed/ interpreted
• includes processed census data, research papers, published maps, textbooks

QUALITY

Nominallowest quality (0 or 1) / only stat tests are Mode + Chi Squared

Ordinal - rank ordering / only stat tests are Spearman's rank, median, percentiles

Interval - discrete scale + allows magnitude / stat tests include mean, SD, Spearmans, regression

32 of 39

## Stat. tests 1.

Mean - includes anomalous data

Median

Mode

Standard deviation - measures central tendancy around mean

Kurtosis - the taller/steeper the curve the lower the standard deviation

Distribution - normal (bell shaped curve) +Ve skew (mean greater than mode) -Ve skew (mode greater than mean)

33 of 39

## Stat. tests 2 - Spearman's

Spearman's Rank Correlation Coefficient

DESCRIPTION

• doesnt have to be normally distributed data

POSITIVES

• easy and simple
• doesnt have to be normally distributed (test other variables)

NEGATIVES

• a poor statistical test
• no magnitude
• no explanation for causes of strength of relationship

34 of 39

## Spearman's significance testing

DESCRIPTION

• degree of freedom (N-2)
• standard significance level 0.05 (95%)

• can say whether results are respresentative of total stat. population
• whether they are significant
• reduces problem of BIAS / due to CHANCE
• whether we can be 95% sure of this
35 of 39

## hypothesis

NULL H1

• relationship found in sample is not representative of total stat. population
• testing for no relationship

ALTERNATIVE

• relationship in sample does accurately reflect total statistical population
36 of 39

## Stat. tests 3 - Pearson's

Pearson's product moment correlation coefficient

DESCRIPTION

• have to have each individual point normally distributed

• statistically better
• more specific
• gives magnitude

• can only use normally distributed interval data
• harder to construct
• doesnt give reason for relationship
37 of 39

## Stat. tests 4 - Chi squared

DESCRIPTION (e.g. corrie orientation)

• observe differences between comparable sets of data (manufacture set from averages)
• null hypothesis put forward - random distibution
• data collected so can be grouped into classes
• compare to significance test to confirm whether any deviation from random in observed data is by chance
• if > critical value then we can reject the null hypothesis

38 of 39

## Stat. tests 5 - nearest neighbour analysis

DESCRIPTION

• allows us to statistically describe the distribution of settlements
• use 'central place theory' to test this
• whether they are clustered / random / regular

0 = CLUSTERED (H1)

1 = RANDOM (H0)

2.15 = REGULAR (H1)

• can be due to many different factors
• e.g settlement distributions
• whether you measure as crow flies or along roads
• SERVICES affect location
39 of 39