BAS 301:
Research Methodology & Proposal
Writing
Taiwo C. OMOTORIOGUN, PhD
Contact
[Link]@[Link]
What is research?
The systematic investigation into the study of
materials, sources, phenomena, etc. in order to
establish facts and reach new conclusions.
What is a proposal
A document that outline a research problem; its associated
research questions, prior literature related to the area; and
the procedure to answering the research questions.
Centered around: what, why and how?
Components of research proposal
Research objectives
Specific things you want to achieve in your research
Clearly formulated objectives help to:
Research
Literature review 3-5 objectives
Guild your research design
methodology
The sections in proposals
must be connected
Examples of objectives
To identify bacteria present on the surface of cell phones (MCB)
To assess the levels of pesticides in the rice from Ladi farms (EMT)
To determine the geo-electric parameters of the subsurface layers (AGP)
To determine the prevalence of malaria infection in urban and rural dwellers in
ondo state (MCB)
To determine the ameliorative effect of lycopene-rich extracts on jaundiced-rats
(BCH)
To investigate the hepatoprotective roles of kola bark extract on high fat fed wistar
rats (BCH)
To determine differences in the biogas yield capacity of different biomass (food waste)
(BITECH)
To Isolate microorganisms that are contaminants of poultry feeds (MCB)
Examples of objectives
Determine abundance, density and diversity of large mammals on
wetlands;
Determine effect of wetland size on abundance, density and
diversity of large mammals in wetlands;
Determine effect of time of day on abundance, density and
diversity of large mammals in wetlands,
Determine effect of time of day on modes of wetland utilization
by large mammals
Examples of objectives
Determine the breeding time of birds in Elizade University Forest
Determine the breeding behaviour of birds in relation to sex
Determine the Sex composition of group size of breeding birds
using molecular methods
Data collection and analysis
400 level projects in view
Statistics:
The practice of data collection, organization, analysis,
interpretation and presentation
Types of Statistics
Descriptive statistics: frequency, measures of central
tendency and measures of dispersion
Inferential statistics: chi-square, t-test, ANOVA and so on.
Why Statistical Analysis?
Processed Data:
Raw Data
Central tendency
Raw data Statistics Variables
wing weight
species age sex wing weight 88.2
vilwe 4 F 83 32.7 Mean 4 36.71
vilwe 6 F 84 32.1 SD 5.19 4.98
vilwe 5 F 86 34 median 86 35.45
vilwe 6 F 81 28.2 Min 81 28
vilwe 4 F 83 35.6 Max 97 46
vilwe 4 M 94 42.6 Kurtosis
vilwe 4 M 93 40.7
vilwe 4 M 95 42.5 Skewness
vilwe 4 M 97 41.1 N = 38
vilwe 4 M 94 39.4
Processed Data:
Raw Data
Graphical representation
Raw data
species age sex wing weight
vilwe 4 F 83 32.7
vilwe 6 F 84 32.1
vilwe 5 F 86 34
vilwe 6 F 81 28.2
vilwe 4 F 83 35.6
vilwe 4 M 94 42.6
vilwe 4 M 93 40.7
vilwe 4 M 95 42.5
vilwe 4 M 97 41.1
vilwe 4 M 94 39.4
Processed Data:
Raw Data Graphical representation
Raw data
species age sex wing weight
vilwe 4 F 83 32.7
vilwe 6 F 84 32.1
vilwe 5 F 86 34
vilwe 6 F 81 28.2
vilwe 4 F 83 35.6
vilwe 4 M 94 42.6
vilwe 4 M 93 40.7
vilwe 4 M 95 42.5
vilwe 4 M 97 41.1
vilwe 4 M 94 39.4
Data distribution
Raw Data Processed Data:
Graphical representation
Raw data
species age sex wing weight Can wing length differ
between ale and female?
vilwe 4 F 83 32.7
vilwe 6 F 84 32.1
vilwe 5 F 86 34
vilwe 6 F 81 28.2
vilwe 4 F 83 35.6
vilwe 4 M 94 42.6
vilwe 4 M 93 40.7
vilwe 4 M 95 42.5 Would body mass be the
same in male and female?
vilwe 4 M 97 41.1
vilwe 4 M 94 39.4
Comparing two groups: same
variable
Processed Data
Raw Data Graphical representation
Raw data
species age sex wing weight
vilwe 4 F 83 32.7
vilwe 6 F 84 32.1
vilwe 5 F 86 34
vilwe 6 F 81 28.2
vilwe 4 F 83 35.6
vilwe 4 M 94 42.6
vilwe 4 M 93 40.7
vilwe 4 M 95 42.5
vilwe 4 M 97 41.1
Pre-insight into correlation
vilwe 4 M 94 39.4
or regression
Categorical variables take on values that
It is used to test for: are names or labels
Goodness of fit e.g. race, sex, age group
Homogeneity of variance
Degree of independence
When to use it
Discrete outcomes (categorical variables)
Ordinal and Dichotomous variables
−
=
Where: oi = observed count and
ei = expected count
H0: The distribution of bird species
across five habitat type is independent
of feeding guilds Output:
Number of cases in table:
Habitat types 651
Species agric sava fore wetl urban
Number of factors: 2
Fruigivore 10 30 16 7 14 Test for independence for
Granivore 35 39 40 30 11 all factors:
Carnivore 15 13 25 18 15
Chisq = 55.49, df
Insectivore 34 34 36 21 10
Omnivore 14 19 30 23 12
=20, p-value = 3.473e-05
Nectivore 34 13 16 18 9
H0: The distribution of bird species
across five habitat types is independent
of time of day Output:
Number of cases in table:
Habitat types 1200
Number of factors: 2
Time agric sava fore wetl urban
Test for independence for
ealymorng 60 78 80 45 37
all factors:
latemorn 50 67 69 50 25
afternoo 45 55 70 55 15
Chisq = 15.05, df
earlyeven 48 58 70 40 20
=16, p-value = 0.0213
lateeven 25 40 45 38 15
Appropriate
for One sample t-test:
comparing two ̅−
groups =
Two sample t-test:
Inappropriatefor
categorical/count
data ̅ − ̅
=
Sample size (n): n +
≥ 5 or ≤ 30
H0: The wing and weight measurement of
group A is equal to that of groupB at 5%
level
Raw data Output:
speci Mean of wingA = 88.34
es age sex wngA wngB wtA wtA
vilwe 4 F 83 76 32.7 33 Mean of wingB = 85.41
vilwe 6 F 84 60 32.1 32.5 t-value = 1.68
vilwe 5 F 86 89 34 36
p-value = 0.103
vilwe 6 F 81 85 28.2 27.8
vilwe 4 F 83 78 35.6 35.6
vilwe 4 M 87 80 33.6 32.8 Since the p-value >0.05, accept H
vilwe 4 M 95 90 41.5 42.5
vilwe 4 M 95 94 42.1 41.1 wing A = wing B
vilwe 4 M 94 93 39.7 39.4
vilwe 4 M 92 95 40.9 36.2
H0: The wing and weight measurement of
group A is equal to that of groupB at 5%
level
Raw data
Output:
speci
es age sex wngA wngB wtA wtA Mean of weightA = 36.71
vilwe 4 F 83 76 32.7 33 Mean of weighB = 36.45
vilwe 6 F 84 60 32.1 32.5
vilwe 5 F 86 89 34 36 t-value = 0.21
vilwe 6 F 81 85 28.2 27.8 p-value = 0.827
vilwe 4 F 83 78 35.6 35.6
vilwe 4 M 87 80 33.6 32.8
Since the p-value >0.05, accept H0
vilwe 4 M 95 90 41.5 42.5
vilwe 4 M 95 94 42.1 41.1 weight A = weight B
vilwe 4 M 94 93 39.7 39.4
vilwe 4 M 92 95 40.9 36.2
To investigate whether the mean of numeric
variables differs significantly between groups and to
quantify the difference (Kopp, 2016).
Appropriate for comparing three or more groups
Assumptions of ANOVA:
Additivity
Error: ~Ν , . Only valid when observation
follows the normal distribution
Assessing Normality
QQ Plots
Shapiro-Wilk test
Accounting for non-
normality
Transformation
Kruskal-Wallistest
Ignore the non-
normality
Assumption of Assumption of
Independence Constant variance
Repeated measurement Bartlett’s test
Similar materials for Levene test
Treatment A and B
Hierarchical data
Accounting for
Unequal variances
Accounting for Transformation
Independence Robust method: Bruner-
No formal test Dette-Munk
Mixed-effect analysis
Repeated measures
analysis
NOT NORMALLY DISTRIBUTED DATA
Metals in feathers
SPECIES Fe Pb Cu Zn Cd Cr Ni
OLISU 326.667 93.333 13.333 6.667 NA 39.13 20
OLISU 278.261 67.391 2.174 19.565 NA NA 8.163
OLISU 71.429 67.857 2.041 50 NA NA NA
OLISU 16.327 18.367 21.622 12.245 NA NA NA
OLISU 40.541 29.73 NA 48.649 NA NA NA
LITGR 61.739 14.783 3.478 20 NA 5.645 10.435
LITGR 33.597 8.3 2.372 9.486 NA NA 9.881
LITGR 102.419 13.71 0.806 4.839 NA NA NA
LITGR 7.983 1.681 0.84 5.882 NA NA NA
LITGR 0.42 2.101 0.21 2.311 NA NA NA
WESBL 1128.571 314.286 71.429 357.143 NA 28.571 342.857
WESBL 14.216 10.294 0.49 5.882 NA NA NA
WESBL NA NA NA NA NA NA NA
KRUSKAL-WALIS TEST
Not normally distributed data
Kruskal-Wallis
Fe Pb Cu Zn Cd Cr Ni
H
F
P
One way anova
Fe Pb Cu Zn Cd Cr Ni
df 9 9 9 9 3 3 5
SS 617405 48805 2047 52986 2981 2981 92274
MS 68601 5423 341.2 5887 993 993.5 18455
F 1.028 1.648 1.267 1.373 NA NA 525.7
P 0.456 0.175 0.346 0.270 NA NA 0.0019
Linearassociation
between two
variables
Range: -1 & +1
Types:
Pearsoncorrelation
Spearman Rank
Function
relation between variables:
dependent and independent variables.
For example,
Tree diameter and volume
Aboveground biomass and vegetation indices
Linear: = + + ⋯+ +
Nonlinear: = +
Assumption: ~Ν ,
Evaluation of Regression Analysis
Coefficientof determination (R2)
Root mean squared error (RMSE)
Overall significance of model and the
parameters
Limitation of standard regression
Standard regression cannot handle non-
normal response
Generalized linear models: Binary logistic,
Multinomial
R software is freely available and the most
flexible. A platform for graphics, statistical
computing and spatial analysis
Start now or never!!!!