Quantitative and Qualitative methods in Library Research
Amy Catalano, Ed.D., MLS, MALS
Associate Professor of Library Services, Hofstra University
Library research tends to be rather poor.
Generally, studies are related to immediate practices and situations and are not generalizable.
Many academic librarians have not been trained to do empirical research
Which methods to choose?
Many new researchers make the mistake of choosing a method before they have a research question.
The research question will define the method
But, can you do the study well with the resources that you have?
Choosing a topic
âHow I done it goodâ
Solving a problem you or your colleagues have had
Explore a theory (e.g., of information-seeking behavior)
Test the efficacy of an intervention
Exploring use of a library service
Explore use of a purchased resource (e.g., a database)
Look at reports published by ACRL which identifies areas of needed research. For example: The Value of Academic Libraries by Megan Oakleaf
Read through TOC/abstracts of recent issues of journals.
Designing a research question
A research question should be focused enough to be answerable in an article, for example.
âWhat are the information seeking behaviors of college students?â May be too broad.
âWhat are in the information seeking behaviors of History graduate students writing a capstone paper?â is more focused.
Identify a problem
Develop a hypothesis
Empirically test that hypothesis by analyzing data
Implications for your Study
Does your study have implications for others in your field? If not, you should do some rethinking.
You study should:
-universally be able to be researched by others
-control: have parameters and identify factors that will affect your results (e.g., SES)
Research design basics: Participants And Sampling
Population: The larger group from which you will draw your sample (e.g., 250 undergraduate students registered with the Students with Disabilities Office)
Sample: Should be random, and therefore generalizable to the chosen population
However, convenience sampling is often the norm.
When doing a teaching intervention, for example, entire classes can be randomly assigned to conditions. This is cluster sampling.
Qualitative: between 1 and 70 (or even more). Although more than 20 is rare. More is not always better or useful in qualitative research.
It is better to ask is the sample representative of what you are investigating?
Quantitative: Varies by method: 30 for each group in experimental research. For surveys 10-20% of the population, although this depdents of the size of the population.
The more the merrier!
The larger the population, the smaller the sample size you need. Beyond 5000 Ns is irrelevant and 400 is adequate. For a samll population (100), no point in sampling. Populaiton of 500, 50% should be sampled.
Research Design Basic: Instruments
If you are going to use a survey, interview protocol or test, you should ensure its reliability and validity. (pp. 61-66, Seligpani)
Reliability: How well an instrument consistently measures whatever it is measuring.
Validity: How well the instrument measures what it is supposed to measure. Several types of validity (e.g., content)
How doesâ¦, Why doesâ¦, (Process questions: the answer is not a number, more exploratory).
Interviews, observations, discourse analysis
Some results can be quantified
What are the perceptions of history students on the availability of primary sources electronically?
This question can be answered through different methods:
A case study
Not meant to be generalizable, so a small sample is acceptable
Diverse views are helpful
Structure or unstructured (or semi-structured)
Listen more, talk less. Ask open ended questions.
It is best to record an interview, which will allow you to take notes.
Watching participants to examine a phenomenon.
E.g., Observing students navigating a database without instruction to determine how they instinctively search.
Participant or non-participant
Taking notes and having a rubric
It is better to have more than one observer and to calibrate training and check on consistency of observations
Like interviews, but includes several individuals to allow for a collected understanding
All participants must get their say
Recording and transcribing a good practice as the researcher will need to mediate the focus group
A follow-up interview is a good practice.
Data-mining: e.g., catalog use, use of services, via Millenium
Can also be quantified
Analyzing qualitative data
Coding: with transcripts of interviews, you may want to code particular words or phrases (with a number, for example) to determine whether a pattern emerges
Grounding results in the current literature
Triangulation and cross-checking: using multiple methods , data collection strategies, and sources to get a clear picture of what is being studied.
To what extent did the distance education group perform better on the post-test than the face-to-face instruction group?
The question calls for a number as an answer.
Manipulation of an independent variable
Removal of the influence of any other variable (can do this with some stats tests)
All factors, except for the independent variable, should be the same
Use of pre-post test (often, but not always)
Ex. Do students who receive instruction via social media perform better on a digital literacy test than students who do not?
Quantitative methods: Surveys
When selecting an instrument search the literature first for one you can use or adapt
Check the reliability and validity
Administration in paper/person gives you a higher return. Online distribution is easier, but there is a lower rate of return (usually 10%) and responses tend to be biased.
Some survey types: tests of information literacy, service satisfaction and use, user-perceptions
Question phrasing is important to validity!
Sampling: convenience, random, snowball
Quantitative methods: Bibliometrics
Citation analysis: An examination of patterns or frequency of citations, authors, topics, methods etc.
May be used to link scholarly works to other authors
May be used to indicate the impact of a journal
Not many opportunities in librarianship
Statistical compilation of the results of many studies on one topic. The results are generally the effect of an intervention.
This is a systematic literature review
Often some systematic evaluation of existing studies is a part of the review
Appealing to librarians because it draws on their data mining
A common method in the health sciences an among health sciences librarians
Research Design basics: Data Analysis
Nominal: number stands in for a word, e.g., 1=female, 2=male
Ordinal: order 1st, 2nd, 3rd
Continuous: numbers from 0-?
Allows the researcher to generalize to a population
SPSS, SAS, or Excel allows a researcher to perform inferential statistics.
Inferential Statistic Tests
Chi-Square: Nominal Data, tests a hypothesis
ANOVA/T-test: compares groups on an independent variable
Regression: Determines the weight of a predictor variable. Also determines which of several variables predicts an outcome
You can combine quant with qual to get a better picture of your inquiry.
For example, you can interview a selection of participants from those you have surveyed to determine why they answered in a particular manner.
Once you select a method, be sure to read further on best practices.
Institutional Review Board
If you are going to be interacting with human beings in some way, you generally need to send a proposal to the IRB.
These humans include: students/faculty/staff at your institution, people at another institution (and you will need to work with the IRB at the institution as well), or anyone you plan to interact with via the telephone, survey, or test, for example.
You should be familiar with the Belmont report and basic Human Research principles.
A tutorial, quiz and certificate are available here:http://phrp.nihtraining.com/users/login.php
Connaway, L. S. (2010). Basic research methods for librarians (5th ed.). Santa Barbara, Calif: Libraries Unlimited.
Academic library research: perspectives and current trends. (2008). Chicago: Association of College and Research Libraries.
SPSS for Psychologists