Unit 5 Research Designs and Sampling, Chapter 13 Sampling for Extensive Studies

Wildemuth chapter 13
sampling for extensive studies – conclusion to draw from your research will hold apply to a particular set of people or organizations
a sample was collected to study and plan to generalize your findings to all the people organizations representative sample
practice of sampling is intended to achieve greater efficiency/if done properly results in a more accurate findings because you can control your data collection procedures and other aspects of the procedure preferably

– Design your sampling procedures carefully/the selection of particular people organizations to participate will these does set of rules that reply to a carefully defined population


Went to selected a probability sample from the population of interest it needs to be reviewed subsequently
probability sampling: concepts and definitions: the population is the collection of elements but which we wish to make an inference – an element is an object in which a measurement is taken
the sampling frame is a list of all agile eligible elements in the population

Probability sample has two specific characteristics first every element of the fibrillation of interest has a known nonzero probability of being selected into the sample
the second important characteristic elements be selected randomly at some point in the process
does not mean haphazard selection Newmont line selection based on the theory of probability that will help you avoid bias in your sample

Procedures for selecting a probability sample
first define your population of interest then define more carefully those who have you something within the past year three times within the year different definition I fear population of interest will depend on your goal

Compromise the definition of the study population to make your sampling plan possible to carry out
define your target population and actual study population

Unit of analysis – could be individuals or could be groups or organizations that unit of analysis you are making clear which elements are eligible for inclusion

Identify or construct the sampling frame – list of elements in the study of population
matches the definition of your study population – missing some portion?
May be complete the contact information or inaccurate or duplications this list is flawed – you will be implicitly redefining your population

Finally select particular animal elements from the sampling frame – identifying specific people – probability sample you use

simple random sampling single sampling frame all the elements in it will have equal chance of being selected, generate a set of random numbers, work through the random numbers and to get the sample size you need

Systematic sampling
extremely tedious and error-prone instead use systematic sampling – random numbers chosen to select the first element from their every and element is selected and varies based on the size of the sample
should not be used if the frames order has some type of pattern it could introduce bias into the sample

Stratified sampling
population is divided into strata 1st/elements a random list sampled from each stratum – may be possible to use a smaller's sample size take your research more efficient

Define the strata – selective variable to divide your population
second decide in the number of strata
finally decide in the sample size

Cluster sampling
sampling units are groups begin the sampling procedure by randomly selecting a sample of clusters
in single stage cluster sampling you would then include in your sample all the elements within the selected clusters
minimize the cost of collecting data from a geographically widespread sample

Nonprobability sampling: concepts and definitions

Not possible to select a random sample from the population which interested? Not possible to identify create a valid sampling frame? Some type of nonprobability sample will need to be defined

Quota sampling
quota for each level of each characteristic/recruiting people until you have met your quota – resemble stratified sampling/no random selection of elements

Purposeful purposive sampling

Judgment sample relies on expert judgment of the person selecting the sample
based on their individual characteristics, a sample that is representative of the population in terms of both central tendency in range of characteristics – include both a typical staff member and more extreme cases/potential for bias

Snowball sampling
particularly sensitive or when eligible belt members may be particularly difficult to identify – ask each participant for suggestions for additional people
forward to five more people

Convenient sampling
try to recruit people because they are available
characteristics of the people who are eligible for study participation

Sample size
as efficient as possible – recruit and collect data from a small sample as possible
don't use a standard method parens (local tradition, at least enough, data availability, intuition, experience, negotiations) – confidence interval around the estimate – area can tolerate any parameter estimates – no fix criteria for setting acceptable confidence interval balance the importance of the accuracy

Special case usability testing special case five or six users can identify 80% of the usability problems homes there are – larger sample size significant portion of the usability problems

The effects of nonresponse in a sample
some of the participants you selected will not accept your invitation response rates only 63% – oversample to increase the number/will not necessary help avoided by a sample

Use appropriate data/use wording/well-designed questionnaires are study procedures = gift certificates cash other items

Unit 5 Research Designs and Sampling, Chapter 12 experimental studies

Wildemuth chapter 12
variables are manipulated and their effects upon other variables observed – basic definition provides a strong framework for understanding primary characteristics of a to experiment
1st some variables called independent variables are manipulated – you must exert control over their variation to be sure you can understand the effects of their understudy
independent variables are the input to the experiment
2nd of their those effects and other variables (call dependent variables) the dependent variable understood as the output from the experiment

characterized by control – all possibilities for variation are either controlled or they are varied systematically
three experimental designs
pretest protest control group design
posttest only control group design
factorial design

Randomization
the validity of your experiment – threats to internal validity
threats to external validity

Additional issues
experimental setting: lab or field
within versus between subjects designs
ethical issues in experiments

Principles of experimentation are simple: randomly's assigned the subjects to two or more groups and measure the outcomes resulting from the activation experienced age group
move beyond the basic characteristics of peer experiments you find their number of decisions to make about design
crucial that you understand your research question thoroughly and stated clearly
identify which independent variable should be manipulated and which dependency variables should be measured

Unit 5 Research Designs and Sampling, Chapter 11 Quasi-experimental Studies

Wildemuth chapter 11
definition used for accessing causal relationships by determining the impact of an intervention – teaching technique, electronic database or collection development Powell's policy or an outcome or effect of interest

Preexperimental designs
true experimental designs
quasiexperimental designs – used in natural setting/some control over the experimental conditions can be exerted you full control is either not possible or not desirable
distinguish first how these differ from those other types of designs – the pre-experiment and the true experiment

Amount of control exerted on next rainiest variables is the primary distinction between cause site experimental studies and true experimental studies
lack of control associated with the absence a random assignment
involve naturally occurring groups
only in true experiments are individual subjects randomly assigned to particular experimental conditions
quasiexperimental methods are designed for settings where utter designs are not feasible
when it's not possible to exert such control – random assignment or some other way/quasiexperimental designs offer an alternative for the researcher

Design lacks complete experimental control/weaker than true experiments
most appropriate approach for many types of studies of interest in information and library science
implemented in a more naturalistic setting – rather more control controlled laboratory setting
increasing ecological validity of the study
attainable in real-world settings
controlled experimentation in a laboratory setting may not translate to success and less controlled natural context

Use of quasiexperimental designs in information library science

Apply to 8 to 10% of research studies in the field of ILS
evidence suggests that two experiments designs are used considerably less frequently than quasiexperimental designs

Typically pragmatic rather than theoretical in nature take the form of applied research
within the framework of action research designs
to actively and directly impact procedures and services with the in a practice setting
research areas include instruction, evaluation, information seeking and professional development

Specific single and multi group designs demonstrate useful designs particularly appropriate for application ILS

Timeseries designs – based on intermittent measurements taken before an exposure to the treatment
within subjects designs
minimum of two data collections points in total encourages use of more than minimum for sufficient opportunities to assess effects of the treatment new paragraph non-equivalent control group design – most likely most frequently apply type also referred to his non-equivalent groups design/pretest posttest not equivalent couple control group design
control comparison study
counterbalance design – multiple treatments are interventions applied to each of the subjects
wrist is design and in interpretation any difference in results can be attributed to the treatment or intervention imposed by the researcher
such as bias and mortality affects – selection bias occurs when the groups being compared are different from each other in some systematic way
history affects – unrelated events associated with the study participants/that is their history during the course of the study
testing affects – pretest of some kind may influence ability ability of your study in two ways first administer close together in time/subject may recall the responses on the pretest to match those or change them for the posttest
second pretest questions may actually condition subjects responses to the intervention causing them to enact interact with it differently than if they had not completed the pretest paren reactive effective testing)
additional threats to be dealt validity

Very useful in situations which make it possible to implement a true experiment/cannot make comparisons/cannot randomly assign research subject to the two groups, plus I experimental design may allow you to address research questions with a balanced combination of rigor and naturalism

Unit 5 Research Designs and Sampling, Chapter 10 Delphi Studies

Wildemuth chapter 10

Oracle: something that is foretold by or as if by supernatural means: divination, prophecy, soothsaying, fitness a nation, vision from rosin budgets to the new thesauruses third edition


Introduction definition name for this Oracle because primarily used in forecasting future events based on the opinion of experts
technique for gleaning and refining the subject of input from a group of people, usually experts, in an attempt to achieve consistence consensus about some aspect of the present or the future

Method for structuring a group communication process so that the process is effective in allowing a group of individuals as a whole to deal with the complex problem
some assessment of the group judgment or view/some opportunity for individual to rape revise views and some degree of anonymity for the individual responses

Not used outside the defense community and no till 1964
forecasting future events has been the major application/versatile research tool that can be used in a bright variety of situations

Characteristics:
allows more people to participate in the study that can effectively interact face-to-face – controls the communication flow and helps the group to stay focused
allows participants from a diverse background and with diverse or opposing views to interact without being concerned about reactions
influences of dominant individuals are avoided – strength of personality eloquence and other personal influential factors much less physical visible because of anonymity
group pressure for conformity is avoided – participants exempted from the direct pressure to conform to majority opinions
effects of feelings and information communicated through body language – tone of voice, gestures, or local buyer minimized./Without seeing each other face-to-face not affected by nonverbal communications
timing costs are reduced – frequent face-to-face meetings are costly and time-consuming/less time and costs are needed

Criticism: net
lack of statistical test
lack of democratic demographic description of participants
selection of experts
lack of exploratory explanatory quality
Degree of Anna and Deputy

Things to avoid
imposing your own preconceptions
in a adequately summarizing and presenting group response
not ensuring common interpretation of the evaluation scales
ignoring rather than exploring disagreements – discouraged it dissenters dropout and artificial consensus is generated
underestimating demand nature of a multi-round Delphi study and not properly compensating the participants
ignoring missed understandings that may arise from differences in language and logic – participants come from diverse cultural backgrounds

Most appropriate situations with the truth of the matter cannot be known to direct observation
leverage the expertise of people who are thought about a particular issue or problem – seeking to find consensus
careful selection of panelists and iterative surveying of other opinions with feedback

Unit 5 Research Designs and Sampling, Chapter 9 Longitudinal Studies

Wildemuth chapter 9

Scope: process that occurs over time, observe it over time to understand more fully
behaviors change as the searchers learn more
cannot be answered with a static description and Association

Longitudinal research refers to a whole family of different research designs including repeated cross-sectional studies, not necessarily including the same study per Spence

Prospective panel designs
a) data are collected for each item or variable for two or more distinct time periods
be) the subjects or cases analyzed are the same or least comparable from once. To the next
C) the analysis involves some comparison of data between or among periods

Two or more participant panels may be included to improve the robustness of the design
data are collected at least three occasions so that trajectory of change can be observed

Advantages: comparing strengths and weaknesses to those of cross-sectional research designs
first: longitudinal research can examine changes or other processes that occur over time within individuals/examine the ways in each participant has changed from one time to the next/can examine the duration of particular episodes or phenomena

Second: basis for drawing conclusions about cause-and-effect/3 important criteria
the two of burials must cover very, this relationship between the must not be attributable to any other cause, and the variable to be believed to be the cause must precede or be simultaneous with the effect
researchers interested in making the strongest case that one variable causes another should consider a large and new to know study design.

Data collection and analysis in longitudinal studies
interviews, questionnaires or other types of measures to asked the participants
collect data from existing records
only rule of thumb is equitable data must be gathered over multiple occasions
make comparisons based on those data

Plan for three times you will collected and how they will be timed
many things, both theoretical and practical. Lifecycle of the process you are trying to observe and how quickly it occurs

Challenges: challenges associated with your sampling attrition from your sample, difficulty of measuring the same variable on each occasion and the effects of extraneous events (including your study procedures) on the phenomenon being absurd
plan for the study will need to conduct when your original plans don't work
anticipate the challenges you might face

Goal is to understand on the individual level phenomenon occurs over time
hold on to all your study participants over the course the study
methods for measuring the variables of interest – can be compared with each other
anticipate extraneous events that could have a biasing effect.

Unit 5 Research Designs and Sampling, Chapter 8 Naturalistic Research

Naturalistic research
if you are data that more closely reflects the real lived experiences of the population of interest – naturalistic research (studies that approximate natural, uncontrived conditions) eliminate new areas of behavior – difficult philosophical issues/practical problems.

Challenge the validity
laboratories studies fail to
gather a detailed unprejudiced record of people's behaviors, beliefs and preferences
explore people's behavior in the context of their own work and life new intensively observed particular elements of context such as setting and artifacts
discover the tactic meanings and understandings that are common in communication and social interaction.

Naturalistic research methods study people in their natural environment – going into the field/replicating elements of the natural environment elsewhere

Basic tension – naturalism and positivism (or rationalism) as modes of inquiry
fundamentally different philosophical approach – direct opposition to traditional scientific norms – causualty, reality, generalizability, and objectivity. Acknowledging that research findings are idiographic, reflecting only one view of one environment

Naturalistic inquiry axioms – practical warnings (complex work processes) alert to political and social issues within organizations/ multiple viewpoints are recognized

Doing naturalistic research
first step move from specific research question to design it helps you answer that question – listen in on a number of reference transactions, extensive notes on whether patrons and librarians said and did.
More naturalistic – enable structs ever be observation of people in their natural environments
they have chosen to come to the library make use of it services a part of ordinary life


Identified patterns and themes – relied on signs and other conventions/watch the students struggle/try to guess how

Worthwhile to document recurrent ways in which discrete activities are produced performed and accomplished by members time and time again
can be used to develop better theories and models of behavior, better programs and systems.

Approaches and trade-offs
organized in three main dimensions degree of naturalism, type of insight, and resources required.

Degree of naturalism
high degree of naturalism – situations and interactions he observed were uncontrived and occurred in the context of participants own work and experience
less naturalistic – directly in the community did not engage

Type of insight
specify the types of insight that are particularly relevant to your interest
architecture: people navigate and orient themselves, design your study, on how and when people get confused
fundraising and marketing: how people talk about the library, they relate to its part of professional, personal and social lives, design your study to focus on people's feelings about the library
choose a naturalistic study in situations where you want to gain insight into people's naturally occurring behavior in natural settings

Resources required
can demand vast amounts of time – rich data to be collected and interpreted
wide array of naturalistic techniques developed
lightweight approaches: (rapid assessment techniques) brief and focused forays
full-blown approaches – engage fully with the chosen domain, direct participation, emphasize open-ended exploration and understanding

Specific techniques
numerous decisions: research design, data collection methods, and data analysis methods.

Field observation: researcher is not attempting to become engaged simply an observer
continuous monitoring or sampling

ethnography
detailed in-depth observation of people's behavior, beliefs and preferences – ongoing basis/daily lives next sentence months or years of fieldwork – time to structure interpret and write about

Contextual inquiry
rapidly gathering – use with an information system design projects/support the design process
spend several hours over the course of one or many days with participant of the system being designed new apprentice to the participants master
participant teaches the researcher about the work processes of interest

Cognitive work analysis
like conceptual inquiry, descriptive approach – help us understand how people actually perform the work
the environment, perceptual, cognitive, and ergonomic attributes of people who typically do the task
provided details of the ways – applied cognitive work analysis in the research (naturalistic study carried out within corporate setting) – served as a framework for data collection and analysis
seven dimensions new the environment new to the work to main new organization new line the task and work to main terms
decision-making terms
strategies that can be used
actors resources and values

Quasi-experiments
fixed research design, similar to an experiment, participants are assigned to conditions – in some systematic way
assigning people randomly to particular conditions is basically anti-naturalistic, quasi-experiments are way to introduce naturalistic elements while maintaining some control

Conclusion reveals the fascinating complexity of human behaviors and enables collection of rich visual verbal and physical data
elusiveness of what you want to measure/seeming infinity of techniques
combine richness of naturalistic techniques, rigor of controlled studies and validated instruments.

Unit 5 Research Designs and Sampling, Chapter 7 - Case Studies

case studies

number of studies employing case studies dramatically increased 1980s 1990s

1.       the complexity of the unit is studied intensively

2.       definition: description of a particular situation or event

serves as a learning tool providing a formal framework for discussion

research: case study is defined as a reach search study focused on a single case or set of cases.

Research approach case study

11 characteristics

1 the phenomena is examined in a natural setting

2 data are collected by multiple means

3 one or a few entities (person, group, or organization) are examined.

4 the complexity of the unit is studied intensively

5 case studies are more suitable for the exploration, classification, and hypothesis development stages of the knowledge – building process; the investigator should have a receptive attitude toward exploration

6 no experimental controls or manipulation are involved

7 the investigator may not specify the set of independent and dependent variables in advance

8 the results derived depend heavily on the integrative powers of the investigator

9 changes in site selection and data collection methods could take place as the investigator develops new hypothesis

10 case research is useful in the study of why and how questions because these deal with operational links to be traced over time rather than with frequency or incidence new 11 the focus is on contemporary events

most often case studies are qualitative and conducted in the field

don’t ignore multimethod aspects of case study research – evidence collected in case studies may be qualitative or quantitative or both combination of both types could contribute to validity of method

research procedures should only be changed after careful consideration because changing could decrease the rigor of the study

when she case studies be used

1.       does the phenomena of interest and have to be studied in a natural setting

2.       does the phenomena of interest focus on contemporary events new line is a research question aim to answer how and why questions new I

3.       does the phenomena of interest include a variety of factors in relationships that can be directly observed

used in exploratory studies – defined phenomena worth studying

example: relationship between information search strategies and personal development theory

case study of a single person – interviewing and observing information seeking on the Internet.

Can be used as a pilot study – for trying out particular data collection methods specific context or become more familiar with the phenomena in a specific context

follow up on exploratory study conducted with another method

examine roles of librarians and remote access – impact on quality and effectiveness of the research process

open-ended interviews – Case study protocol was used to follow up – can be used on a preliminary study/weakness is its lack of generalizability.

Can also be used it is the descriptive research to depict comprehensively the phenomena of interest

over a seven-year period – case study method allowed an investigator to describe different aspects of the restructuring, including chronological, operational, and role-based

Facilitate evaluation research – specific organizational context – natural cases: is well suited to understanding the interactions between information technology related innovations and organizational contexts

can be directly applied to the improvement of information and library practice

Summary case studies useful in many different types of research: exploratory and confirmatory, descriptive and evaluative

ideal when a how or why question asked about a contemporary set of events over which researcher has no control

Designing a case study

first step clearly define your research question

theoretical and empirical literature may provide a sketchy foundation

literature review will be your first step – other critical steps

identifying your unit of analysis

selecting a case

or cases that will be the focus of your study/planning your data collection procedures

Identifying the unit of analysis

major entity that you are analyzing in your study – focused on individuals as a unit of analysis/aggregate entities like groups or organizations

primary defining characteristic – focuses on a single instance of the unit of analysis

multiple perspectives by gathering data – multiple units of analysis – aggregate to understand case that is focus of the study

Selecting a case

strategically select a case or several – theoretical purposes and the relevance of its case to these

theoretical sampling – replicate or extend an emergent theory

statistical sampling – focuses on selecting study participants representative of population

focus on single/compared to – in depth investigation/rich detail

five possible reasons for selecting a particular case

1.       representative

2.       critical case – essential for testing a well formulated theory

3.       three extreme or unique case

4.       revelatory case – eliminates previously inaccessible knowledge

5.       longitudinal study – repeatedly studied at different points in time

Multiple case studies called comparative case studies – combination of two or more single case studies

literal replication – similar

theoretical replication – cases that differ

collecting data

multiple methods of data collection – analysis of existing documentation, archival records, interviews, direct observation, participant observation and examination of physical artifacts, quantitative methods (questionnaires)

direct observation most frequently used – equate with field studies/not limited to direct observation

results combined through triangulation – process of using multiple perceptions to clarify types of triangulation

1. Data triangulation – combining data from different sources

2. Investigator triangulation – combining data multiple researchers

3. Methodological  triangulation – combining data collected via different methods

4. theory triangulation – combining data collected from multiple theoretical perspectives

data triangulation – data from several sources: cataloging staff, executive staff, policy and procedures manual, and Web server transaction log

Methodological l triangulation – direct observation of workplace, interviews, content analysis and transaction log analysis comparison of the findings – and share findings are valid

Strengths and weaknesses of case studies

lack of generalizability of study findings is the weakness – no basis for generalizing the findings beyond the setting in which he was conducted

particularization, not generalization – rich nests with which a particular setting or phenomena can be described

results from case studies are generalizable to theoretical propositions

theory can be tested through case study research

single case can be used to test a theory against a particular set of empirical circumstances

cannot be treated as a representative sample from population

four classes of criteria relevant to the quality of a case study research report

1. resonance criteria – degree to which the report fits, overlaps with, or reinforces selected theoretical framework

2. rhetorical criteria – deal with the form structure and presentational characteristics of the report/unity, overall organization, simplicity or clarity and craftsmanship

3. empowerment criteria – ability to evoke and facilitate action, empowerment related characteristics include fairness, educative newness and action ability

4. applicability criteria – feasibility of making inferences from the case study applying them in the readers context or situation. Transfer findings from one context to another, the relevant characteristics of both should be the same

Conclusion

how and why – examining contemporary events in a natural setting

single case – define a general conceptual category of property

multiple cases – confirm the definition

strength: flexibility of this research strategy, rich array of data collection techniques.

Triangulation of multiple data sources and data collection methods – support theory testing and development, describe the phenomena of interest. Go to sleep

Unit 5: Research Method, Design, and Sampling; Ethics; Theory, Connaway & Powell Chapter 4

Survey Research and Sampling

Survey: group of research methods used to determine the present status of a given phenomenon
- to make inferences about a large group of elements by studying  relatively small number selected from a larger group

define: to look over/see beyond

• Observations
• Population
• Data
• Bias

Major Differences Survey Types & Other Methods of Research

• Survey gathers contemporary data
• Historical Research concerned with past data

Contrast experimental -

• doesn't enable researcher to manipulate the independent variable
• less control of research environment
• not considered capable of definitely establishing causal relationship
• less rigorous than experimental
• better suited to study large number of cases
• geographically dispersed
• more appropriate personal factors/exploratory analysis of relationships

Types of Survey Studies

Exploratory Surveys

1. Literature surveys
2. Experience surveys
3. Analysis of "insight-stimulating" examples

Analytical and Descriptive Surveys

appropriate for data - quantitative in nature, require statistical assistance to extract meaning
descriptive survey - MOST COMMON TYPE OF SURVEY = (SURVEY RESEARCH METHOD)

• Other Types of Surveys

1. Cross sectional
2. Trend study
3. Cohort study
4. Panel study
5. Approximation of a longitudinal study
6. Parallel samples study
7. Contextual study
8. Socio-metric study
9. Critical incident study

Basic Purposes of Descriptive Surveys - describe characteristics of population of interest, estimate proportions, make specific predictions, test associational relationships

• can consider but not test causal relationships, 
• can test associations relationships
• do tend to 
• could conclude
• seems to be a correlation

other factors or variables

• could influence more
• cannot control so could not test
• relationship must make sense conceptionally regardless of methodology/technique

Basic Steps of Survey Research Overview

• Formulating Objectives
• Selecting Data Collection Techniques
• Selecting the Sample
• Collecting the Data
• Analyzing and Interpreting the Results
• Survey Research Designs
• Survey Research Costs
• shorten length of data collection
• reduce number of follow ups
• limit pilot or pretest to small no. participants
• shorten time spent developing data collection instruments - adapt to existing instruments
• make instrument as short as possible
• use non-monetary incentives to encourage respondents
• minimize staff costs
• shop around for least expensive supplies and equipment
• reduce number of survey activities
• minimize amount of time each activity takes

Sampling - MOST CRUCIAL STEP

Basic Terms and Concepts

• Universe 
• Population
• Population Stratum - subdivision of a population based on specification or characteristic
• Element - individual member
• Census - count of all and determine characteristics
• Sample - selection of units
• Case - individual members (lower case n)
• Sampling Frame - actual list of units from which the sample is selected

Types of Sampling Methods

• Non-probability Sampling
• Accidental Sample
• Quota Sample
• Snowball Sample
• Purposive Sample
• Self-Selected Sample
• Incomplete Sample
• Probability Sampling
• Simple Random Sample SRS
• Selecting the Simple Random Sample
• number sequentially elements in population
• determine how many are to be elected
• select 3 digit numbers from table to give every element a chance of selection
• choose starting point and pattern
• processed in pattern
• select those to skip
• Systematic Sample
• Stratified Random Sample
• Cluster Sample

Combination of Cluster, Stratified and simple random EXAMPLE

• entire 48 continguous state divided into small areas called PRIMARY SAMPLING UNITS (PSU) usually counties, metro areas, tele exchange, A stratified  random sample o 75 are selected from total

• PSU is stratified into large, small and smaller cities and/or rural areas. Each unit is a SAMPLE PLACE

• Sample place divided into CHUNKS/BLOCKS, a number are randomly selected

• Chunks broken down into SEGMENTS, containing 4-12 dwelling units, segments randomly drawn from each chunk

• DWELLING UNITS, constitute final sample. city directory can obtain telephone numbers for those chosen

Determining the Sample Size
general rule of thumb, larger the better, less than 100 not likely to represent population
General Criteria

1. degree of precision required between sample and population, less accuracy=smaller sample
2. variability of the population, greater variability=larger sample
3. method of sampling / stratified sampling requires fewer cases than simple or systematic random
4. way in which the results are to be analyzed / small=significant limitations on types of statistics

Use of Formulas

Sampling Error

Other Causes of Sampling Error

Non-Sampling Error

Sampling in Library Use

Non-Random Sampling

Random Sampling Over Time

Conclusions

Summary

Unit 5: Research Method, Design, and Sampling; Ethics; Theory, Connaway & Powell Chapter 3

Selecting the Research Method:

Applied Research

Action Research

Evidence Based Research

Evaluative Research

Cost Benefit Analysis

• net value
• reducing uncertainty
• buying service elsewhere
• librarian time
• improves performance or saves money

other kinds:

• cost-minimization
• cost-utility
• willingness to pay
• willingness to accept
• cost of time

Qualitative Research

Specific Research Methods:

Survey Research

Experimental Research

Historical Research

Operations Research

• formulate problem
• construct mathematical model to represent study
• derive solution
• test the model and solution
• establish controls over solution
• implementation

Modeling

Systems Analysis

Case Study

Delphi Study

Content Analysis

Bibliometrics

Task Based Research

Comparative Librarianship

Technology Based Research Methods

• incorporate transaction log analysis with other data collection methods
• with user demographic data
• allows for search behaviors to be analyzed in relation to the searchers experience with online systems, educational background, reason for the search, etc. 
• require researcher to infer less about nature of search and maintain validity of study

Ethics of Research

• mutually respectful
• win-win relationship
• weigh questionable practices against potential benefits

Guidelines for LIS Professionals

Ethics for Research on the Internet

Scientific and Research Misconduct

Unit 5: Research Method, Design, and Sampling; Ethics; Theory, Role of Theory Topic 13

Patten, Topic 13
The Role of Theory in Research

Theory: unified explanation for discrete observation

example: reinforcement theory : positive reinforcement increases frequency of response
but
intermittent reinforcement is more effective

Test hypothesis by deducing with theory

example: goals determine level of cognitive engagement
exceptions cause reformulating to account for discrepancy

research can induce theory

Qualitative Researchers call it grounded theory (based on observations)/regularly revised as new observations warrant.

Consider research topic to test some aspect of theory, easier to defend, has implications for validating and refining a theory.

no theory is universal, always exceptions - Trends across groups can be examined

Unit 5: Research Method, Design, and Sampling; Ethics; Theory, Ethical Concerns Topic 12

Patten Topic 12
Ethical Considerations in Research

Key to promoting ethical values is Informed Consent to Participants of:
1. general purpose of the research
2. what will be done to them during the research
3. what the potential benefit(s) to them and others might be
4. what the potential for harm to them might be
5. the fact that they may withdraw at any time with penalty

Participation in a Research Study could

• present hazards
• expose anxiety
• mental anguish
• sensitive topics - renew anxiety

Participants must be protected from physical and psychological harm

Participants have a right to privacy

• Confidentiality
• disguised or hidden identity

Knowledge of the purpose of the study can make it difficult, balance

Follow with debriefing,

• review purpose of study
• procedures
• results
• confidential 
• answer any questions about the study

Unit 5 Research Method, Design, and Sampling; Ethics; Theory - JML

Eldredge, J. D. (2004). Inventory of research methods for librarianship and informatics. Journal of the Medical Library Association, 92(1), 83-90. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC314107/

Traditionally: case study, program evaluation, and survey research methodologies

Now utilizing wider array of research methods :

Analysis -
detailed examination of anything complex - understand it's nature or essential features

Audit -
review multiple variables, identify strengths and weaknesses, strict criteria formulated in advance, what should be done, protocol or plan compares to actually do.

Autobiography -
written by self about self(whole life, segment or episode)

Bibliomining -
see Data Mining

Biography -
narrative account of notable's life (whole life, segment or episode)

Case Study -
describe / analyze self experiences with a process, group, innovation, technology, project, population program or organization
explicitly state, prior to beginning, questions, propositions, unit(s) of analysis, logic & criteria for interpretation

Citation Analysis -
see Descriptive Survey

Cohort Design - track over time, defined population sharing set of common characteristics - encountering possible intended or not exposure to phenomenon or subsequent observable change in population, consequence of exposure.

Comparative Study -
systematic effort to find similarities and differences between 2 or more observed phenomena

Content Analysis -
maps non-numerical artifacts (text) into matrix of statistically manipulated symbols, to reduce large body of qualitative info to manageable size

Data Mining -
discovery of meaningful patterns from low-level data with automated methods (statistical or AI tools) - in libraries: Bibliomining

Delphi method -
anonymous participants response to questions over iterations to reach quantitative group decisions.

Descriptive Survey -
seeks to ascertain respondents' perspectives or experiences on a subject in predetermined structured way.

Focus Group -
generates data with small group setting, analyzed to help in planning/decisions, evaluate programs, products/services - develop model /theory,

Gap Analysis -
survey to detect discrepancies/gaps customer expectations or org and ability to deliver

History -
seeks to recreate accurate past although their is dispute. reveal cause/effect

Longitudinal Study -
see Cohort Study

Meta-Analysis -
combine identical/comparable data sets, 2 or more studies to create larger pool of results to strengthen conclusion

Narrative Review - 
literature review on broadly defined subject, write intro overview, describe current research/controversies (offers concise intro to broad subjects)

Participant Observation -
researcher - community meet to study(understand situation from perspective of participants

Program Evaluation  -
systematic assessment of operation / outcome of program/policy, compare to explicit or implicit standards.

Randomized Controlled Trial -
carefully defined and assembled population. complies with predetermined inclusion and exclusion criteria. divided into random control group - std treatment or none

Summing Up -
cluster of methods (book written to describe) where meta-analysis cannot synthesize

Systematic Reviews -
minimize bias, integrating multiple studies / concise summary of best evidence to critically appraise and synthesize relevant studies. quantitative or qualitative, related to summing up

Unobtrusive Observation -
ethical concerns (participants do not know they are being studied)