Information research methods in Library and Information Management are critical for extracting meaningful insights from the vast amounts of data generated by information users and resources. In Kenya, information professionals working in libraries, archives, and information centres must apply rigorous data analysis techniques to support decision-making, improve service delivery, and enhance resource management. This chapter focuses on the analysis of data, emphasizing the identification of appropriate techniques to handle different types of data encountered in the library and information environment. Mastery of these techniques enables professionals to interpret data accurately and provide evidence-based recommendations.

3.1 Data analysis techniques

Data analysis techniques are the structured methods a researcher uses to turn raw figures and responses into information that supports decisions. In a library or information centre, the data may come from circulation records, door counts, database usage logs, catalogue records or user satisfaction surveys. Choosing a technique depends on the type of data, the research question and the tools available. The techniques in this topic are descriptive statistics, correlation analysis, regression and cluster analysis, and each is introduced after the identification step that precedes all analysis.

3.1.1 Identification

Identification in data analysis refers to the process of recognising and categorising the type of data collected, the appropriate analytical approach, and the tools required for processing. This step is crucial for library and information management professionals who must ensure that data from sources such as user feedback, bibliographic records, or digital databases is analysed accurately to inform service improvements. The four techniques below are the main quantitative techniques identified for library research, and each is matched to a question the researcher wants to answer.

Descriptive statistics

Descriptive statistics summarise a dataset so that its main features can be seen at a glance, without drawing conclusions beyond the data in hand. A librarian at a county library, for example, may summarise daily visitor counts for one week to report the average number of users and how much the counts vary. Descriptive statistics are usually the first technique applied, because they reveal errors, unusual values and the general shape of the data before any deeper analysis.

The main groups of descriptive measures are:

• Measures of central tendency: The mean (sum of values ÷ number of values), the median (the middle value when data are ordered) and the mode (the most frequent value) show the typical value. The median is preferred when a few extreme values distort the mean.
• Measures of dispersion: The range, variance and standard deviation show how widely values are spread around the centre. A small standard deviation means daily visitor numbers are fairly steady.
• Frequency distributions: Tables that count how often each category or value occurs, for example the number of users in each category (students, researchers, staff).
• Percentages and proportions: Express a part as a share of the whole, such as the percentage of respondents rating the service as good.
• Graphical summaries: Bar charts, pie charts and histograms present frequencies and distributions visually. The interpretation of such visuals belongs to the later topic on data analysis results.

Worked example: daily visitors to a college library from Monday to Sunday were 120, 135, 150, 110, 160, 145 and 130.

Mean = sum ÷ n

Mean = 950 ÷ 7

Mean = 135.7 visitors per day

Median: the ordered values are 110, 120, 130, 135, 145, 150, 160, so the middle (4th) value is 135 visitors.

Range = highest − lowest

Range = 160 − 110

Range = 50 visitors

Sample standard deviation = √[Σ(x − mean)² ÷ (n − 1)]

Sample standard deviation = √(1821.43 ÷ 6)

Sample standard deviation = √303.57

Sample standard deviation = 17.4 visitors

Correlation analysis

Correlation analysis measures the strength and direction of the relationship between two variables. A librarian may want to know whether users who attend more information literacy sessions also borrow more books. The result is a correlation coefficient (r) between −1 and +1: values near +1 show a strong positive relationship, values near −1 a strong negative relationship, and values near 0 little or no linear relationship.

Key points to identify before using correlation:

• Pearson correlation coefficient: Used for two numerical variables (interval or ratio data) that are roughly linearly related, such as hours spent in the library and books borrowed.
• Spearman rank correlation: Used when data are ordinal, such as satisfaction ratings (poor, fair, good), or when numerical data are not normally distributed.
• Direction: A positive r means both variables rise together. A negative r means one rises as the other falls, for example more overdue fines and fewer visits.
• Strength: Values from about 0.7 to 1 (or −0.7 to −1) are usually described as strong, around 0.3 to 0.7 as moderate, and below about 0.3 as weak. These bands are guidelines, not fixed rules.
• Correlation is not causation: A strong relationship does not prove that one variable causes the other. A third factor, such as the examination period, may drive both.
• Sample size: A coefficient from a very small sample is unreliable and should be treated cautiously.

Formula: r = Σ(x − x̄)(y − ȳ) ÷ √[Σ(x − x̄)² × Σ(y − ȳ)²]

Worked example: five students attended 1, 2, 3, 4 and 5 information literacy sessions and borrowed 2, 4, 5, 4 and 5 books in the semester.

x̄ = 3 sessions, ȳ = 4 books

Σ(x − x̄)(y − ȳ) = 4 + 0 + 0 + 0 + 2 = 6

Σ(x − x̄)² = 10

Σ(y − ȳ)² = 6

r = 6 ÷ √(10 × 6)

r = 6 ÷ 7.746

r = 0.77

This is a strong positive relationship, although five students is too few to generalise from.

Regression

Regression goes beyond correlation by building an equation that describes how one variable (the dependent variable) changes when another (the independent variable) changes. It allows a library to predict an outcome, such as the number of books a student is likely to borrow given the number of training sessions attended. Simple linear regression uses one independent variable, while multiple regression uses several.

Points to identify when choosing regression:

• Dependent variable: The outcome being explained or predicted, such as books borrowed per semester.
• Independent variable: The factor believed to influence the outcome, such as sessions attended, enrolment size or opening hours.
• Simple linear regression: Fits a straight line, y = a + b × x, where a is the intercept and b is the slope (the change in y for each one-unit rise in x).
• Multiple regression: Includes several independent variables at once, for example sessions attended, year of study and distance from the library, to see the effect of each while holding the others constant.
• Logistic regression: Used when the outcome is a yes or no category, such as whether a registered user becomes an active borrower.
• Coefficient of determination (R²): Shows the share of variation in the dependent variable explained by the model. In simple regression, R² = r².
• Assumptions and limits: The relationship should be roughly linear, and predictions should stay within the range of the data used to build the model.

Worked example: using the five students above.

Slope b = Σ(x − x̄)(y − ȳ) ÷ Σ(x − x̄)²

b = 6 ÷ 10

b = 0.6 books per session

Intercept a = ȳ − b × x̄

a = 4 − 0.6 × 3

a = 2.2 books

Regression equation: books borrowed = 2.2 + 0.6 × sessions

Prediction for a student attending 4 sessions:

Books borrowed = 2.2 + 0.6 × 4

Books borrowed = 4.6 books

R² = 0.77² = 0.6, so about 60% of the variation in borrowing is explained by sessions attended.

Cluster analysis

Cluster analysis groups cases, such as library users, branches or titles, so that members of the same group are more alike than members of different groups. Unlike regression, there is no outcome variable and no groups are defined in advance; the groups emerge from the data. A university library might cluster users by visits, loans and e-resource downloads to discover groups such as heavy print borrowers, mainly online users and occasional visitors, then design services for each group.

Points to identify about cluster analysis:

• Purpose: Segmenting users, grouping similar collection items or comparing branches with similar usage patterns.
• K-means clustering: The researcher chooses the number of groups (k) and the method assigns each case to the nearest group centre, repeating until the groups are stable.
• Hierarchical clustering: Cases are joined step by step into larger groups, and the result is shown in a tree diagram called a dendrogram, which helps decide how many clusters to keep.
• Distance measure: Similarity is judged by distance. For two variables, Euclidean distance = √[(x₁ − x₂)² + (y₁ − y₂)²].
• Standardising variables: Variables measured on different scales (for example visits and minutes online) should be converted to a common scale so that one does not dominate.
• Naming clusters: The researcher must interpret each group from its average values and give it a meaningful label.

Worked example: User A made 10 visits and 2 downloads in a month, and User B made 4 visits and 8 downloads.

Distance = √[(10 − 4)² + (2 − 8)²]

Distance = √(36 + 36)

Distance = √72

Distance = 8.49

A large distance suggests the two users belong to different clusters.

Steps in a basic cluster analysis:

1. Define the purpose: State what the groups will be used for, such as planning separate services for different user groups.
2. Select the variables: Choose measures that describe user behaviour, for example visits, loans and downloads per month.
3. Clean the data: Remove or correct missing and clearly wrong values, since a single extreme value can distort a group.
4. Standardise the variables: Put all variables on a comparable scale.
5. Choose the method and number of clusters: Use k-means when the number of groups is known or can be tested, or hierarchical clustering to see how the groups form.
6. Run the analysis: Use statistical software such as SPSS or R to assign each case to a cluster.
7. Describe and name each cluster: Compare the average values of each group and label them.
8. Check usefulness: Confirm that the groups are distinct, reasonably sized and useful for decisions.

Identifying data types

Data types determine the nature of the analysis and the statistical methods applicable. Library surveys may include quantitative data such as the number of books borrowed, or qualitative data such as user comments on service quality. Understanding data types ensures that professionals apply the right analysis techniques.

• Nominal Data: Represents categories without any order, such as types of library users (students, researchers, staff). Analysis methods include frequency counts and mode determination.
• Ordinal Data: Reflects categories with a meaningful order but no fixed interval, like satisfaction ratings (poor, fair, good). Techniques include median calculation and non-parametric tests.
• Interval Data: Numerical data with equal intervals but no true zero, such as calendar years of publication. Analysis may involve mean and standard deviation.
• Ratio Data: Numerical data with a meaningful zero point, for example, the number of hours a user spends in the library. Enables the full range of statistical analyses including ratios and coefficients.
• Qualitative Data: Non-numeric data such as interview transcripts or open-ended survey responses, which require thematic or content analysis.

Identifying analytical approaches

Once data types are established, the next step is to determine the analytical approach suitable for the research objectives within the library setting. Summarising data to describe a dataset is covered under descriptive statistics above.

• Inferential Analysis: Draws conclusions about a population based on sample data, for instance, estimating the satisfaction of all registered users from a sample of respondents.
• Exploratory Analysis: Investigates data to discover patterns without prior hypotheses, useful in analysing new digital resource usage trends at university libraries.
• Predictive Analysis: Uses historical data to forecast future trends, such as estimating future book acquisitions based on past borrowing patterns.
• Diagnostic Analysis: Examines why something happened, for example why loans of a particular subject fell during a term.
• Qualitative Analysis: Involves coding and categorising textual data to identify themes, often used in evaluating user feedback at health information centres.

Identifying appropriate tools and software

Information professionals must select tools that fit the data type and analytical approach. Accessibility, cost, and user expertise influence tool choice.

• Statistical Packages: Software like SPSS and R is suitable for quantitative data analysis, offering a range of statistical tests, including correlation, regression and cluster analysis, and visualisations.
• Spreadsheet Software: Microsoft Excel is useful for simple descriptive statistics and data visualisation in public and academic libraries.
• Qualitative Data Analysis Software: NVivo and Atlas.ti facilitate coding and thematic analysis of textual data, supporting in-depth user satisfaction studies.
• Library Management Systems: Integrated systems such as Koha often include built-in reporting tools for circulation and cataloguing data analysis.
• Open Source Tools: Free tools such as PSPP and LibreOffice Calc provide cost-effective options for resource-constrained institutions such as county libraries and SACCO information centres.

Recognising data quality issues

Identification also involves assessing the quality of data before analysis. Library data may suffer from incompleteness or inaccuracies due to manual entry errors or system limitations.

• Completeness: Ensuring all necessary data fields are filled, such as borrower demographics in circulation records.
• Accuracy: Verifying that data entries reflect true values, for example, cross-checking acquisition dates with supplier invoices.
• Consistency: Maintaining uniform data formats across datasets, critical in multi-branch library networks.
• Timeliness: Using up-to-date information to reflect current trends, vital for decision-making in dynamic environments like university libraries.
• Validity: Confirming that data measures what it is intended to measure, such as ensuring survey questions align with the intended user experience metrics.

Identifying ethical considerations in data analysis

Ethical identification ensures that data handling respects user privacy and complies with legal frameworks such as Kenya's Data Protection Act.

• Confidentiality: Protecting user identities during analysis, especially for sensitive information in medical libraries.
• Informed Consent: Ensuring users understand how their data will be used, as in user experience studies.
• Data Security: Implementing measures to prevent unauthorised access to data, crucial in digital library repositories.
• Transparency: Being clear about analysis methods and limitations when reporting findings to stakeholders.
• Bias Avoidance: Recognising and mitigating any biases in data collection or analysis to maintain integrity in research outcomes.

Chapter Summary

This chapter explored various data analysis techniques essential for interpreting research data, including descriptive statistics, correlation analysis, regression, and cluster analysis, each serving a specific purpose in identifying patterns and relationships within data sets. It then examined how to interpret data analysis results through comparative, visual, theoretical, and contextual lenses, supported by a case study to illustrate practical application. The discussion on findings emphasized the importance of validation methods and procedures to ensure accuracy and reliability of research outcomes. The chapter also detailed the results and analysis process, highlighting critical steps such as data preparation through formatting and transforming, data exploration, statistical analysis, and the subsequent interpretation and conclusion of findings. Effective presentation and communication of results were underscored as vital for conveying research insights clearly. Finally, the chapter covered the structure and guidelines for preparing a research proposal, providing a foundation for systematic investigation and reporting.