Transcript Basic statistics - Iowa State University

Swine Statistics
Ken Stalder
Iowa State University
IOWA STATE UNIVERSITY
Department of Animal Science
Why are we teaching statistics in a swine
systems course?


Today’s pork operations are more sophisticated.
Operations involve a series of tasks

Breeding & Conception Rates
AI Technicians
Service Sires
Day effects

Farrowing rate
Fallout by day of gestation
Breeding issues
Health

Farrowing Production
Number born alive
Stillbirth rate
Number weaned
IOWA STATE UNIVERSITY
Department of Animal Science
Why are we teaching statistics in a swine
systems course?


Today’s pork operations are more sophisticated.
Operations involve a series of tasks

Variation about these number can be just as important when
evaluating production information
Conception rates day to day within a week
week to week within a month
month to month within a quarter
month to month within a season
month to month within a year
season within a year
year to year within a herd
herd to herd within a production system
production system within a country
Country within the world
IOWA STATE UNIVERSITY
Department of Animal Science
Example of this type of information
IOWA STATE UNIVERSITY
Department of Animal Science
Statistics meaning – dictionary.com
1.
( used with a singular verb ) the science that
deals with the collection, classification,
analysis, and interpretation of numerical facts
or data, and that, by use of mathematical
theories of probability, imposes order and
regularity on aggregates of more or less
disparate elements.
2.
( used with a plural verb ) the numerical facts
or data themselves.
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Mean
n
Y
Treatment
1
Treatment
2
Treatment
3
270
290
290
300
250
340
280
280
330
280
290
300
270
280
300
Total
1400
1390
1560
4350
5
5
5
15
280
278
312
290
IOWA STATE UNIVERSITY
Department of Animal Science
Example from Kaps and Lamberson,
Biostatistics for Animal Science, 2004
Basic statistical measures commonly used when
evaluating swine data
 What
does the mean really tell us?
 Depends



Evaluating titer levels for some disease
Determining whether a production value is within some
normal level
Temperature values


what the question is?
Most equipment does not hold values like temperature at a
mean but within some normal operating levels
Other examples
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 After
obtaining your data you can start to ask
yourself more complex questions
 May

seem simple but can be very descriptive
Range
Simply the high and low values
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Interpreting
 Depends



the meaning of the range
on data being examined
Within a production system
Within an operation
Within a group of pigs

Grow – Finish pigs
Using Paylean™



Nursery age pigs
Piglets in farrowing
Feeding animals
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Mean
n
Y
Range
Treatment
1
Treatment
2
Treatment
3
270
290
290
300
250
340
280
280
330
280
290
300
270
280
300
Total
1400
1390
1560
4350
5
5
5
15
280
278
312
290
270-300
250-290
290-340
270-340
IOWA STATE UNIVERSITY
Department of Animal Science
Example from Kaps and Lamberson,
Biostatistics for Animal Science, 2004
Basic statistical measures commonly used when
evaluating swine data
 After
obtaining your data you can start to ask
yourself more complex questions
 May


seem simple but can be very descriptive
Median
What is the median?


It is the middle number where ½ the values are above and ½
the values are below
Often good to understand along with the mean
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 How

do you calculate the median?
Line up all of the values

If you have an odd number of values it is the number with the
same number of observations on either side
5, 7, 13, 25, 99
Median = 13

If you have an even number of values it is the average of the
two middle numbers that have the same number of
observations on either side
5, 7, 13, 15, 25, 99
Median = (13 + 15) / 2 = 14
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 What

does the median tell us?
It is one measure of the variability?


Less precise than others
If the data are “close” normally distributed then the mean and
median values will be similar
Example
5, 7, 13, 25, 99
Median = 13

Mean = 29.8
Often used in when evaluating financial information
Net income from contract growers
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Mean
n
Y
Median
Treatment
1
Treatment
2
Treatment
3
270
290
290
300
250
340
280
280
330
280
290
300
270
280
300
Total
1400
1390
1560
4350
5
5
5
15
280
278
312
290
280
280
300
290
IOWA STATE UNIVERSITY
Department of Animal Science
Example from Kaps and Lamberson,
Biostatistics for Animal Science, 2004
Basic statistical measures commonly used when
evaluating swine data
 After
obtaining your data you can start to ask
yourself more complex questions
 May


seem simple but can be very descriptive
Mode
What is the mode?


It is the value that occurs most frequently
Similar to the median it is good to understand along with the
mean
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Mean
n
Y
Mode
Treatment
1
Treatment
2
Treatment
3
270
290
290
300
250
340
280
280
330
280
290
300
270
280
300
Total
1400
1390
1560
4350
5
5
5
15
280
278
312
290
270 & 280
280 & 290
300
280
IOWA STATE UNIVERSITY
Department of Animal Science
Example from Kaps and Lamberson,
Biostatistics for Animal Science, 2004
Basic statistical measures commonly used when
evaluating swine data
 After
obtaining your data you can start to ask
yourself more complex questions
 May
seem simple but can be very descriptive
 Often

Quartiles – divide up the data into 4 equal portions


see data presented in
Calculate means of the top 25% and bottom 25% of producers
Percentiles – divide up the data into 10 equal
percentages (10, 20, 30…….., 90, 100)

Calculate means from the top 10% of herds and bottom 10% of
herds
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 What

use do data have if broken up into
Quartiles or Percentiles

Compare to other similar operations within a production system
Benchmark against operations within a system
Establish production goals
Establish a bonus structure

Compare to other operations within a country
Really examining how an operation compare to others under similar
buying systems, financial circumstances, etc.

Compare to other operations across the world
Ultimately defines the most efficient production around the world
Generally the economic situation is most important in these
comparisons as it will ultimately determine where pork is produced
world-wide.
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 The
mean, mode, median, range are really
descriptive statistics.

Often used in a snap shot in time
 These
values really tell us nothing about the
variability with our data.
 Why
is understanding the variability or the
variation in a production system important?
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Standard
deviation = a measure of the
variability within some data
 Under
the normal distribution assumptions:
 68%
of the data will fall within ±1 standard
deviation of the mean
 95%
of the data will fall within ±2 standard
deviations of the mean
 Almost
all (99.7%) of the data will fall within ±3
standard deviations of the mean
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Standard
deviation = a measure of the
variability within some data
IOWA STATE UNIVERSITY
Department of Animal Science
Image from:
http://www.spcforexcel.com/normal-probability-plots
Basic statistical measures commonly used when
evaluating swine data
 Standard
deviation = a measure of the
variability within some data
68.3 % of the area represented ±1 SD
95.4 % of the area represented ±2 SD
99.7 % of the area represented ±1 SD
IOWA STATE UNIVERSITY
Department of Animal Science
Image from:
http://www.spcforexcel.com/normal-probability-plots
Basic statistical measures commonly used when
evaluating swine data
 Standard
deviation = a measure of the
variability within some data
 The
normal curve can take on a variety of
shapes depending on the variation that exists
within the data
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Standard
deviation = a measure of the
variability within some data
 What
does the SD really tell us when looking
data from swine operations?








Number born alive
Number weaned
Weaning weight
Weaning age
Return to estrus
Backfat
Loin muscle area
Percent lean
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures commonly used when
evaluating swine data
 Mean
Treatment
1
Treatment
2
Treatment
3
270
290
290
300
250
340
280
280
330
280
290
300
270
280
300
Total
1400
1390
1560
4350
5
5
5
15
Y
280
278
312
290
Standard
Deviation
11.0
14.7
19.4
22.5
n
IOWA STATE UNIVERSITY
Department of Animal Science
Example from Kaps and Lamberson,
Biostatistics for Animal Science, 2004
Basic statistical measures commonly used when
evaluating swine data
 Why
is understanding the variability or the
variation in a production system important?
 It
starts by understanding a term called
“throughput”

Units produced per unit of fixed costs




Lbs of pork per $ fixed costs a slaughter facility
Lbs of sausage per $ fixed costs in a processing facility
Pigs produced per $ fixed costs in a breed to wean operation
Variation usually takes away from throughput

Looking for variation causes to correct or reduce
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures to interpret scientific
publications
 Graduates
going into more technical fields
related to swine production
 Often
asked to interpret information for the
company you work for or the clientele you
serve
 Need
be able to understand scientific
publications
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures to interpret scientific
publications
 Most

No difference between treatment means
 What



papers have a null hypothesis or H0
does P < 0.05 mean
I reject the null hypothesis of no treatment mean
difference
I am saying there is a difference
5% chance of declaring a difference when it really was
not there.
IOWA STATE UNIVERSITY
Department of Animal Science
Basic statistical measures to interpret scientific
publications
 Most

No difference between treatment means
 Does




papers have a null hypothesis or H0
P > 0.05 mean that there is no difference?
Not necessarily
Did the research have enough animals or pens of animals
in each treatment to detect a difference
If not best to say no difference was observed
The more variation with the data measured the more
animals or pens of animals that have to be tested

Example
Reproductive traits – number born alive and weaned
Performance traits – growth rate and feed efficiency
Carcass traits – backfat, loin muscle area, percent lean
IOWA STATE UNIVERSITY
Department of Animal Science
Any questions or comments?
IOWA STATE UNIVERSITY
Department of Animal Science